cyberbullying_detection.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "G10_ALDA_project.ipynb",
      "provenance": [],
      "collapsed_sections": [
        "BeKpTOD-UdB_",
        "JE4VJHfwdxg8"
      ],
      "toc_visible": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "Mjeo8OioPx7K"
      },
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "import glob\n",
        "import matplotlib.pyplot as plt"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_SX8f4DOP2g6",
        "outputId": "ba200002-bde9-46ec-8fad-078845709693"
      },
      "source": [
        "import os\n",
        "from google.colab import drive\n",
        "drive.mount('/content/gdrive/')\n",
        "os.chdir(\"./gdrive/My Drive/ALDA project/\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mounted at /content/gdrive/\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WkjbUuMJRBcp",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "bfc667c6-d258-4f75-e433-64f95343311c"
      },
      "source": [
        "!pwd"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/content/gdrive/My Drive/ALDA project\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SBGttTkFRnrP"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6FImbNIgNNfP"
      },
      "source": [
        "#Data Pre processing"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "r_EeYijxUj-Y"
      },
      "source": [
        "## Kaggle Dataset Cleaning"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Lm3KK5pbSp92",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "outputId": "9bbb273d-26de-4076-92ed-7d5dee5d7395"
      },
      "source": [
        "#kaggle_files = glob.glob(\"/kaggle_parsed_dataset.csv\")\n",
        "kaggle_df = pd.read_csv(\"./kaggle_parsed_dataset.csv\")\n",
        "kaggle_df['Source'] = kaggle_df.shape[0]*['kaggle_df']\n",
        "kaggle_df.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>index</th>\n",
              "      <th>oh_label</th>\n",
              "      <th>Date</th>\n",
              "      <th>Text</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>20120618192155Z</td>\n",
              "      <td>\"You fuck your dad.\"</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>20120528192215Z</td>\n",
              "      <td>\"i really don't understand your point.\\xa0 It ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>2</td>\n",
              "      <td>0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>\"A\\\\xc2\\\\xa0majority of Canadians can and has ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>3</td>\n",
              "      <td>0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>\"listen if you dont wanna get married to a man...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>4</td>\n",
              "      <td>0</td>\n",
              "      <td>20120619094753Z</td>\n",
              "      <td>\"C\\xe1c b\\u1ea1n xu\\u1ed1ng \\u0111\\u01b0\\u1edd...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   index  ...                                               Text\n",
              "0      0  ...                               \"You fuck your dad.\"\n",
              "1      1  ...  \"i really don't understand your point.\\xa0 It ...\n",
              "2      2  ...  \"A\\\\xc2\\\\xa0majority of Canadians can and has ...\n",
              "3      3  ...  \"listen if you dont wanna get married to a man...\n",
              "4      4  ...  \"C\\xe1c b\\u1ea1n xu\\u1ed1ng \\u0111\\u01b0\\u1edd...\n",
              "\n",
              "[5 rows x 4 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3VeU5uZwUViV",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "b0234464-792e-432d-bdf4-2b47d0c854d9"
      },
      "source": [
        "kaggle_df.isna().sum()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "index          0\n",
              "oh_label       0\n",
              "Date        1242\n",
              "Text           0\n",
              "dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PmjPwHpSfvkV"
      },
      "source": [
        "kaggle_df = kaggle_df.drop(['index', 'Date'], axis=1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JK63ZWSdg67C",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        },
        "outputId": "d1560512-8c9e-4bbb-9dfb-11dda71b5236"
      },
      "source": [
        "kaggle_df"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>oh_label</th>\n",
              "      <th>Text</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1</td>\n",
              "      <td>\"You fuck your dad.\"</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0</td>\n",
              "      <td>\"i really don't understand your point.\\xa0 It ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0</td>\n",
              "      <td>\"A\\\\xc2\\\\xa0majority of Canadians can and has ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0</td>\n",
              "      <td>\"listen if you dont wanna get married to a man...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0</td>\n",
              "      <td>\"C\\xe1c b\\u1ea1n xu\\u1ed1ng \\u0111\\u01b0\\u1edd...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8794</th>\n",
              "      <td>0</td>\n",
              "      <td>\"Never really gave it much thought. I just fig...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8795</th>\n",
              "      <td>0</td>\n",
              "      <td>\"Nadie se salva de la regla 34 xd\"</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8796</th>\n",
              "      <td>0</td>\n",
              "      <td>\"Question: Are you a boy or a girl?\"</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8797</th>\n",
              "      <td>1</td>\n",
              "      <td>\"Leave your email or phone number and maybe yo...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8798</th>\n",
              "      <td>0</td>\n",
              "      <td>\"From the scenarios you present, I see you bel...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>8799 rows × 2 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "      oh_label                                               Text\n",
              "0            1                               \"You fuck your dad.\"\n",
              "1            0  \"i really don't understand your point.\\xa0 It ...\n",
              "2            0  \"A\\\\xc2\\\\xa0majority of Canadians can and has ...\n",
              "3            0  \"listen if you dont wanna get married to a man...\n",
              "4            0  \"C\\xe1c b\\u1ea1n xu\\u1ed1ng \\u0111\\u01b0\\u1edd...\n",
              "...        ...                                                ...\n",
              "8794         0  \"Never really gave it much thought. I just fig...\n",
              "8795         0                 \"Nadie se salva de la regla 34 xd\"\n",
              "8796         0               \"Question: Are you a boy or a girl?\"\n",
              "8797         1  \"Leave your email or phone number and maybe yo...\n",
              "8798         0  \"From the scenarios you present, I see you bel...\n",
              "\n",
              "[8799 rows x 2 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "l7M8dNrgVqOE"
      },
      "source": [
        "## Twitter Dataset Cleaning"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "v50duvU0Vx90",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "outputId": "716c8f2d-117c-47af-e1cb-3a000efa3a75"
      },
      "source": [
        "twitter_df = pd.read_csv(\"./twitter_parsed_dataset.csv\")\n",
        "twitter_df['Source'] = twitter_df.shape[0]*['twitter_df']\n",
        "twitter_df.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>index</th>\n",
              "      <th>id</th>\n",
              "      <th>Text</th>\n",
              "      <th>Annotation</th>\n",
              "      <th>oh_label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>5.74948705591165E+017</td>\n",
              "      <td>5.74948705591165E+017</td>\n",
              "      <td>@halalflaws @biebervalue @greenlinerzjm I read...</td>\n",
              "      <td>none</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>5.71917888690393E+017</td>\n",
              "      <td>5.71917888690393E+017</td>\n",
              "      <td>@ShreyaBafna3 Now you idiots claim that people...</td>\n",
              "      <td>none</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>3.90255841338601E+017</td>\n",
              "      <td>3.90255841338601E+017</td>\n",
              "      <td>RT @Mooseoftorment Call me sexist, but when I ...</td>\n",
              "      <td>sexism</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>5.68208850655916E+017</td>\n",
              "      <td>5.68208850655916E+017</td>\n",
              "      <td>@g0ssipsquirrelx Wrong, ISIS follows the examp...</td>\n",
              "      <td>racism</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>5.75596338802373E+017</td>\n",
              "      <td>5.75596338802373E+017</td>\n",
              "      <td>#mkr No No No No No No</td>\n",
              "      <td>none</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                   index                     id  ... Annotation oh_label\n",
              "0  5.74948705591165E+017  5.74948705591165E+017  ...       none      0.0\n",
              "1  5.71917888690393E+017  5.71917888690393E+017  ...       none      0.0\n",
              "2  3.90255841338601E+017  3.90255841338601E+017  ...     sexism      1.0\n",
              "3  5.68208850655916E+017  5.68208850655916E+017  ...     racism      1.0\n",
              "4  5.75596338802373E+017  5.75596338802373E+017  ...       none      0.0\n",
              "\n",
              "[5 rows x 5 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 5
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xQDEs6OTWyXY",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        },
        "outputId": "ec2c534a-c21a-499d-d922-7eaa2266e828"
      },
      "source": [
        "twitter_df"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>index</th>\n",
              "      <th>id</th>\n",
              "      <th>Text</th>\n",
              "      <th>Annotation</th>\n",
              "      <th>oh_label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>5.74948705591165E+017</td>\n",
              "      <td>5.74948705591165E+017</td>\n",
              "      <td>@halalflaws @biebervalue @greenlinerzjm I read...</td>\n",
              "      <td>none</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>5.71917888690393E+017</td>\n",
              "      <td>5.71917888690393E+017</td>\n",
              "      <td>@ShreyaBafna3 Now you idiots claim that people...</td>\n",
              "      <td>none</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>3.90255841338601E+017</td>\n",
              "      <td>3.90255841338601E+017</td>\n",
              "      <td>RT @Mooseoftorment Call me sexist, but when I ...</td>\n",
              "      <td>sexism</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>5.68208850655916E+017</td>\n",
              "      <td>5.68208850655916E+017</td>\n",
              "      <td>@g0ssipsquirrelx Wrong, ISIS follows the examp...</td>\n",
              "      <td>racism</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>5.75596338802373E+017</td>\n",
              "      <td>5.75596338802373E+017</td>\n",
              "      <td>#mkr No No No No No No</td>\n",
              "      <td>none</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16846</th>\n",
              "      <td>5.75606766236475E+017</td>\n",
              "      <td>5.75606766236475E+017</td>\n",
              "      <td>Feeling so sorry for the girls, they should be...</td>\n",
              "      <td>none</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16847</th>\n",
              "      <td>5.72333822886326E+017</td>\n",
              "      <td>5.72333822886326E+017</td>\n",
              "      <td>#MKR 'pretty good dishes we're happy with' - O...</td>\n",
              "      <td>none</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16848</th>\n",
              "      <td>5.72326950057845E+017</td>\n",
              "      <td>5.72326950057845E+017</td>\n",
              "      <td>RT @colonelkickhead: Deconstructed lemon tart!...</td>\n",
              "      <td>none</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16849</th>\n",
              "      <td>5.74799612642357E+017</td>\n",
              "      <td>5.74799612642357E+017</td>\n",
              "      <td>@versacezaynx @nyazpolitics @greenlinerzjm You...</td>\n",
              "      <td>none</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16850</th>\n",
              "      <td>5.68826121153684E+017</td>\n",
              "      <td>5.68826121153684E+017</td>\n",
              "      <td>And before you protest that you're *not* mad, ...</td>\n",
              "      <td>none</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>16851 rows × 5 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                       index                     id  ... Annotation oh_label\n",
              "0      5.74948705591165E+017  5.74948705591165E+017  ...       none      0.0\n",
              "1      5.71917888690393E+017  5.71917888690393E+017  ...       none      0.0\n",
              "2      3.90255841338601E+017  3.90255841338601E+017  ...     sexism      1.0\n",
              "3      5.68208850655916E+017  5.68208850655916E+017  ...     racism      1.0\n",
              "4      5.75596338802373E+017  5.75596338802373E+017  ...       none      0.0\n",
              "...                      ...                    ...  ...        ...      ...\n",
              "16846  5.75606766236475E+017  5.75606766236475E+017  ...       none      0.0\n",
              "16847  5.72333822886326E+017  5.72333822886326E+017  ...       none      0.0\n",
              "16848  5.72326950057845E+017  5.72326950057845E+017  ...       none      0.0\n",
              "16849  5.74799612642357E+017  5.74799612642357E+017  ...       none      0.0\n",
              "16850  5.68826121153684E+017  5.68826121153684E+017  ...       none      0.0\n",
              "\n",
              "[16851 rows x 5 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eUeustzYb3sS",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "3c81c672-bacb-4445-bd38-5ecebaf8ce4e"
      },
      "source": [
        "cyberbully = twitter_df[twitter_df.oh_label==1.0].count()[0]\n",
        "nocyberbully = twitter_df[twitter_df.oh_label==0.0].count()[0]\n",
        "value = [cyberbully,nocyberbully]\n",
        "labels = [\"cyberbully\", \"no-cyberbully\"]\n",
        "plt.bar(np.arange(len(value)),value);\n",
        "plt.title(\"Comparison between number of cyberbully and nocyberbully text detected\")\n",
        "plt.xlabel(\"Labels\")\n",
        "plt.ylabel(\"Count\")\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAdQAAAEWCAYAAADfB2bTAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3deZxcVZn/8c83CWGRJYFkIiSBBI3K4rAYAZdRBhg2lzAziLgREEVHXGZUNnEGfgoKMzooCigDkQQZAkYQFB0M+4+BAIki+xI2kxBIIGGTRQLP/HFOkdtlVXen+3RXdff3/XrVq+8999xTz12fu/UtRQRmZmbWO8NaHYCZmdlg4IRqZmZWgBOqmZlZAU6oZmZmBTihmpmZFeCEamZmVsCgTqiSPibpt/38nQ9L2qM/v3OwkhSS3tii736zpFslPSvpi4XanJSnaUQ7tFddVyUdL+mnJeLqDUnXSPpUq+OoKb0O9qa96ryRdLCk60vF1V9KbwP9pbvzu1sJVdJHJc2X9JykpZJ+I+ndvQ+zb0XEeRGxZ6vj6C4n47ZyJHB1RGwQEae2OhizUkoeJPTlQW/Jg4b+OlDrMqFK+jLwPeBbwDhgc+B0YFrfhtY7A+0IyPpOD9eFLYA7S8dSitfvgcXLa4iIiKYfYCPgOeBDndRZm5RwH82f7wFr52G7AotJR/vLgKXAfsC+wH3ACuBrlbaOB+YAFwDPAr8DtqsMPxp4IA+7C/j7yrCDgf8FTgGeBE7IZdfn4crDlgHPALcD21amcxawHHgE+DowrNLu9cB3gJXAQ8A+ncyPh4FjcnwrgZ8A61SGvx+4FXgKuAH461x+LvAq8EKe50cCM4Gv5OHjgQAOz/1vyPNvWGft5mGbAT/P0/cQ8MW6eX5hnv5nSUlkaifTF8Bngfvzd50GqNLWTyt1J+X6I3L/NXm53JCn8ZfAJsB5eZncAkyq+64vAg8CTwD/UZvePPyTwN15Pl8ObFE37uE5zoeaTMsH8/Q+lWPbKpdfBbwCvJjjfFODcTfOy/bR/P2/yOV3AB+o1Fsrx75DZX4clsdbCny1UncYq9fxJ/Ny2bhuXh4K/BG4rhvtnQOcUOnfFVhct67uUb/sgMuAL9RN721Utre6YT8DHgOeznFtUxfDabnNZ4GbgDdUhv8dcE8e94fAtcCnmnzP8XSyrgJb5eX4VB72wcqwdYHvkrbvp0nb9LpdTSuF18HO2qN728+nqvul3H0a8N26abgU+JcG8/C63OafSOv2h7vYL32YtM/YMPfvk5f12GZt1X3fcNK+84k8zYfXTdNGwNmkdXcJaf8wPC/LF0nb4XPAU5V88x3SNvA48CNg3cr3TcvT8QxpO9obOJGO2/MPc923AHNJ+9F7gQMq7WyS5+EzwM3AN2vzu7NPVwl1b2BVbeKb1PkGMA/4qzyTbwC+WdmAVwH/RtqxfJq0U/9vYANgG1ICmVxZoV4G9s/1v5oX5lp5+IdIyWFYXtB/AjatrGCrgC8AI0gby8GsXun2AhYAo0jJdavKuLOAS3JMk0jJ/tBKuy/n2IcD/0TaeanJ/HiYtFOdSNrp/i95p0baqS4Dds5tTc/1166Mu0fdxvrL3P3RvIJcUBl2SVft5nm1IC+DkcCWpBV7r8o8f5F0kDMc+DYwr5PlHcCv8nzcPC/Pvddgh7CQdDCwEemg4z5gj7zMZgE/qfuuq/N83DzXre1QpuW2tsrjfh24oW7cuXncdRtMx5tI68/fkda1I3N7I+t3Xk3mw2WkA7/Refz35vIja8uoEuftdfPjfOB1wFvz/KsltS+RtqUJedn9GDi/btxZedx1u9HeOfQsoR4A3FSptx0pwY9sMi8+Sdp2agfXt1aGnZPH3Skvp/OA2XnYGFJirG3v/0LahjtLqA3X1Tz+QuBrpPV8t9z2mytJ5xrSgelw4J053k6nlcLrYBftvbYMOtl+GiXUnUj7pGGV+fo8MK6TbfiNlf6u9kvn5eW4Sf6e9zdrq8F3fZZ0wFTbH15dN00Xk9bz15FyyM3AZ+qnsdLeKaREtzFpnfsl8O3KfHiatE0Py8v6LY225/x9i4BD8rLbgZT0t87DZ5MO3l4HbEtK9r1OqB8DHuuizgPAvpX+vYCHKxvwC8Dw3L9Bnpk7V+ovAParrFDzKsOGkY5c/qbJd98KTKvM/D/WDa+udLuRVt5d6HiEORz4c21G5rLPANdU2lhYGbZenobXN4npYeCzlf59gQdy9xnkg43K8HtZvTN+mI4J9Q2kI99hpCOxz5B3iKSz1y931S5pI6mfL8eQE1ee51dUhm0NvNDJ8g7g3ZX+C4Gj12CHcGxl+HeB31T6P0DHnXGQk3Xu/xxwZe7+Dfmgp7KuPE8+Q8jj7tbJdPwrcGHd+EuAXRttgHXjbkq6mjC6wbDNSDvy2hH9HODIuvnxlkr9fwfOzt13A7vXfc/LpA2+Nu6WDeZvs/bOoWcJdR3Sejcl938HOL2z/UClzVE5po0qMZxVtz3ck7sPouP2LtIVrc4SasN1Ffgb0plTdds+P48zjLQf2q5Bm51OK4XXwS7ae20ZdLL9/EVCraw7f5e7Pw/8uottuJpQu9ovjSKdEd4O/Lizthp811V03B/uWZsm0i3El+h4hvkR0rMLjaZRpIPg6hWOd7D67P/HwClN4nht3uX+DwP/v67Oj4HjSDnhZTpuV9+iGwm1q3uoTwJjurj+vxnpMkrNI7nstTYi4pXc/UL++3hl+AvA+pX+RbWOiHiVtIFtBiDpoPzk5VOSniIdOYxpNG69iLiKdEnpNGCZpDMlbZjHX6vBNIyv9D9Waef53FmNuV41jur82AL4Si3+PA0T6Ti/qjE/QFqBtiftMH4FPCrpzaRkeW032t0C2Kxu2NdIK/NfTB9ph7BOF8u8vn5n86Je/bLvbF2Azufl9yvTtIK0wY1vMm69DuttXtcW1Y3fzERgRUSsrB8QEY+Srkr8o6RRpEtk563BNF1cmaa7SZeqxjUZt6v2eiQiXiSdfX9c0jDSTu7cRnUlDZd0kqQHJD1DStLQcbtstr5sRsftPeh8mTVqq7aubgYsysuxprYdjyElzgfqG+vmtJZeB4sur2wm8PHc/XGaLK8mOt0vRcRTpMv625IOgtdEh2VMx/3sFqR979LK9/6YdKbayFjSCc2CSv3/yeXkmP9iGTexBbBz3TR/DHh9bm9EJ3E31VVCvZF0BLFfJ3UezcHVbJ7LempirSOv4BNISWQL4L9IR1+bRMQo0qVVVcaNzhqOiFMj4m2kI9s3AUeQTvNfbjANS0pMAx3nxyLgxIgYVfmsFxHndxL/taRLYiMjYknun0661HhrN9pdRDqCqw7bICL27cX0NfMn0gpf8/oCbXY2Lz9TN13rRsQNlfqdrQ8d1ltJyt/VneW+CNg4J8xGaju3DwE35uVW1dk07VM3TevUjd9ompq115vlMZO0g9kdeD4ibmxS76OkS597kC7jT8rlalK/aikdt3fRcVrWxKPAxLzPqKltx0+QLhW/ocm4XU1r6XWwL5bXT4FpkrYjXYL+xRqM2+l+SdL2pMv65wNr+sR7h2VMmt7q974EjKl874YRsU0eXj/vniAddG9Tqb9RRKxfaa/ZMq5vaxFwbd00rx8R/0S6bbKqk7ib6jShRsTTpHtvp0naT9J6ktaStI+kf8/Vzge+LmmspDG5fm/+n+1tkv4hH3X+M2mGzyNdyw7SxCLpENIRU7dIeruknSWtRVpxXwRezWfPFwInStogJ+4v93IaDpc0QdLGwLGkI2BIBwSfzXFI0uskvU/SBnn446R7nFXXkg4irsv91+T+6ytn/p21ezPwrKSjJK2bzyi2lfT2XkxfM7cC75G0uaSNSJeWe+sISaMlTSTdY6zNyx8Bx0jaBkDSRpI+tAbtXgi8T9LueZ34Cmldu6Hz0SAilpIu952eY1tL0nsqVX4B7JjjndWgiX/N29I2pHs41Wk6Ma+D5G2qO0/TN2vvVmBfSRtLej1pe+qWnFReJZ2RdHa2swFpvj1JSgbf6u53kO5Db1PZ3r9Izw/CbiKdsR6Zl8eupFsIs/NZ6wzgPyVtlreBd0haG7o1raXXwWbt9Xj7iYjFpIf6zgV+HhEvdFK9fj/TdP8haR3SvvBrpHVrvKTPddJWvQuBL+b94WjSQ3e1mJcCvwW+K2lDScMkvUHSeyttT5A0Mtd/Ncd6iqS/ApA0XtJeuf7ZwCF5mx6Wh72lSZy/At4k6RN5fVkr54it8n71IuD4vF1tTTqJ6VKX/zYTEd8lJZivk5LZItIOvXYEdAIwn/Rk3O2kJ3NP6M6XN3EJ6fr2SuATwD9ExMsRcRdphb+RNHPeSrq01l0bkhbGStLp+5OkJ+wgPcj0J9LDOteTHpqa0Ytp+G/SivIg6RLECQARMZ/0cNMPcxwLSfcJar5NOjh5StJXc9m1pJ1WLaFeT9px1fo7bTevHO8nXTZ+iHSUdxbpbKKoiJhL2jncRro3/qsCzV6S27qVtAM+O3/XxcDJwGylS413kC6vdjfWe0lnkT8gzZMPkJ7O/XM3m/gE6crGPaQHOl5LVnln9nNgMmnDrHctaRldCXwnImovH/k+6YGL30p6lnQguXM3YmnW3rnAH0iXYX/L6h13d80ibWedHVzOIm1PS0gPmc3rbuMR8QTpLP4k0vY4hTXbpqtt/Zm0DPchLc/TgYMi4p5c5auk/dMtpEuzJ9Nx/9fZtJZeB5u119vtZ2aehq4u9x4PzMz7mQO62C99m3Qp/YyIeIm0zZwgaUqjthp813+Rnn7+Ayk31G8PB5EeIqv9V8Qc0rMDkO6/3gk8JumJXHZUjm9enudXAG8GiIibSUn/FNLDSdey+irU94H9Ja2UdGpEPEu6n3sg6QrBY6RluXau/3nSrYnHSM8B/KTJvOyg9u8ObUHS8aQb3B/vqq5ZO5P0b6R/txmw67Kkg4DDIqLtX+LSW4NhWvNVkp+SHopqnx37EOJ/NjYrLF/qP5R0FjsgSVqP9ATq6a2Opa8NhmnNty2+RHqi2sm0RQb1u3zN+pukT5Nui/wmIq7rqn47yveklpNurfx3i8PpU4NhWiVtRXohw6ak/wO2FmmrS75mZmYDVUvOUCXNkLRM0h2Vsv+QdI+k2yRdrMq/JEg6RtJCSfdWnuhC0t65bKGkoyvlkyXdlMsvqD0lZmZm1ldacoaab54/B8yKiG1z2Z7AVRGxStLJABFxVH5k+XzSa6U2Iz3V9abc1H2k10zVHhn/SETcJelC4KKImC3pR8AfIuKMruIaM2ZMTJo0qeSkmpkNegsWLHgiIsZ2XXNwa8lDSRFxnaRJdWXV3y2dR3qZAaR/Gp+dH9l+SNJCUnKF9ErABwEkzSb9Y/PdpNcMfjTXmUl6tLvLhDpp0iTmz5/fk0kyMxuyJHXrTUKDXbs+lPRJ0j/OQ3qNV/UVUItzWbPyTUi/TLCqrrwhSYcp/dbr/OXLlxcK38zMhpq2S6iSjiW99qn+/ad9IiLOjIipETF17Nghf8XCzMx6qK3+D1XSwaS3+uxe+V+qJXR8p+IEVr9vtVH5k8AoSSPyWWq1vpmZWZ9omzNUSXuTfkvyg7H6F10gvYrtQElrS5pMej3ZzaSHkKbkJ3pHkl4hdWlOxFez+h7sdNKrvszMzPpMq/5t5nzSO3nfLGmxpENJ75HcAJir9BNtPwKIiDtJL1i+i/RTPYdHxCv57PPzpPdE3k36bcs781ccBXw5P8C0CfldmWZmZn3FL3aomDp1avgpXzOzNSNpQURMbXUcrdY2l3zNzMwGMidUMzOzApxQzczMCmirf5sxs74z6ejLWh2CtamHT3pfq0MYFHyGamZmVoATqpmZWQFOqGZmZgU4oZqZmRXghGpmZlaAE6qZmVkBTqhmZmYFOKGamZkV4IRqZmZWgBOqmZlZAU6oZmZmBTihmpmZFeCEamZmVoATqpmZWQFOqGZmZgU4oZqZmRXghGpmZlaAE6qZmVkBTqhmZmYFOKGamZkV4IRqZmZWgBOqmZlZAU6oZmZmBbQkoUqaIWmZpDsqZRtLmivp/vx3dC6XpFMlLZR0m6QdK+NMz/XvlzS9Uv42SbfncU6VpP6dQjMzG2padYZ6DrB3XdnRwJURMQW4MvcD7ANMyZ/DgDMgJWDgOGBnYCfguFoSznU+XRmv/rvMzMyKaklCjYjrgBV1xdOAmbl7JrBfpXxWJPOAUZI2BfYC5kbEiohYCcwF9s7DNoyIeRERwKxKW2ZmZn2ine6hjouIpbn7MWBc7h4PLKrUW5zLOitf3KC8IUmHSZovaf7y5ct7NwVmZjZktVNCfU0+s4x++q4zI2JqREwdO3Zsf3ylmZkNQu2UUB/Pl2vJf5fl8iXAxEq9Cbmss/IJDcrNzMz6TDsl1EuB2pO604FLKuUH5ad9dwGezpeGLwf2lDQ6P4y0J3B5HvaMpF3y070HVdoyMzPrEyNa8aWSzgd2BcZIWkx6Wvck4EJJhwKPAAfk6r8G9gUWAs8DhwBExApJ3wRuyfW+ERG1B50+R3qSeF3gN/ljZmbWZ1qSUCPiI00G7d6gbgCHN2lnBjCjQfl8YNvexGhmZrYm2umSr5mZ2YDlhGpmZlaAE6qZmVkBTqhmZmYFOKGamZkV4IRqZmZWgBOqmZlZAU6oZmZmBTihmpmZFeCEamZmVoATqpmZWQFOqGZmZgU4oZqZmRXghGpmZlaAE6qZmVkBTqhmZmYFOKGamZkV4IRqZmZWgBOqmZlZAU6oZmZmBTihmpmZFeCEamZmVoATqpmZWQFOqGZmZgU4oZqZmRXghGpmZlaAE6qZmVkBbZdQJf2LpDsl3SHpfEnrSJos6SZJCyVdIGlkrrt27l+Yh0+qtHNMLr9X0l6tmh4zMxsa2iqhShoPfBGYGhHbAsOBA4GTgVMi4o3ASuDQPMqhwMpcfkquh6St83jbAHsDp0sa3p/TYmZmQ0tbJdRsBLCupBHAesBSYDdgTh4+E9gvd0/L/eThu0tSLp8dES9FxEPAQmCnforfzMyGoLZKqBGxBPgO8EdSIn0aWAA8FRGrcrXFwPjcPR5YlMddletvUi1vME4Hkg6TNF/S/OXLl5edIDMzGzLaKqFKGk06u5wMbAa8jnTJts9ExJkRMTUipo4dO7Yvv8rMzAaxtkqowB7AQxGxPCJeBi4C3gWMypeAASYAS3L3EmAiQB6+EfBktbzBOGZmZsW1W0L9I7CLpPXyvdDdgbuAq4H9c53pwCW5+9LcTx5+VURELj8wPwU8GZgC3NxP02BmZkPQiK6r9J+IuEnSHOB3wCrg98CZwGXAbEkn5LKz8yhnA+dKWgisID3ZS0TcKelCUjJeBRweEa/068SYmdmQ0lYJFSAijgOOqyt+kAZP6UbEi8CHmrRzInBi8QDNzMwaaLdLvmZmZgOSE6qZmVkBTqhmZmYFOKGamZkV4IRqZmZWgBOqmZlZAU6oZmZmBTihmpmZFeCEamZmVoATqpmZWQFOqGZmZgU4oZqZmRXghGpmZlaAE6qZmVkBTqhmZmYFOKGamZkV4IRqZmZWQK8TqqR3dafMzMxsMCtxhvqDbpaZmZkNWiN6OqKkdwDvBMZK+nJl0IbA8N4GZmZmNpD0OKECI4H1cxsbVMqfAfbvTVBmZmYDTY8TakRcC1wr6ZyIeKRgTGZmZgNOb85Qa9aWdCYwqdpeROxWoG0zM7MBoURC/RnwI+As4JUC7ZmZmQ04JRLqqog4o0A7ZmZmA1aJf5v5paTPSdpU0sa1T4F2zczMBowSZ6jT898jKmUBbFmgbTMzswGh12eoETG5wafHyVTSKElzJN0j6W5J78hnvXMl3Z//js51JelUSQsl3SZpx0o703P9+yVNb/6NZmZmvdfrM1RJBzUqj4hZPWzy+8D/RMT+kkYC6wFfA66MiJMkHQ0cDRwF7ANMyZ+dgTOAnfMl5+OAqaSz5QWSLo2IlT2MyczMrFMlLvm+vdK9DrA78DtgjROqpI2A9wAHA0TEn4E/S5oG7JqrzQSuISXUacCsiAhgXj673TTXnRsRK3K7c4G9gfPXNCYzM7Pu6HVCjYgvVPsljQJm97C5ycBy4CeStgMWAF8CxkXE0lznMWBc7h4PLKqMvziXNSv/C5IOAw4D2HzzzXsYtpmZDXV98fNtfyIlxp4YAewInBERO+S2jq5WyGej0asIO7Z3ZkRMjYipY8eOLdWsmZkNMSXuof6S1QluOLAVcGEPm1sMLI6Im3L/HFJCfVzSphGxNF/SXZaHLwEmVsafkMuWsPoSca38mh7GZGZm1qUS91C/U+leBTwSEYt70lBEPCZpkaQ3R8S9pPuxd+XPdOCk/PeSPMqlwOclzSY9lPR0TrqXA9+qPQ0M7Akc05OYzMzMuqPEPdRrJY1j9cNJ9/eyyS8A5+UnfB8EDiFdmr5Q0qHAI8ABue6vgX2BhcDzuS4RsULSN4Fbcr1v1B5QMjMz6wslLvkeAPwH6ZKqgB9IOiIi5vSkvYi4lfTvLvV2b1A3gMObtDMDmNGTGMzMzNZUiUu+xwJvj4hlAJLGAleQ7n+amZkNCSWe8h1WS6bZk4XaNTMzGzBKnKH+T34IqPbShA+T7m2amZkNGT1OqJLeSHrhwhGS/gF4dx50I3BeieDMzMwGit6coX6P/K8oEXERcBGApLfmYR/odXRmZmYDRG/udY6LiNvrC3PZpF60a2ZmNuD05gx1VCfD1u1FuwPSpKMva3UI1qYePul9rQ7BzPpBb85Q50v6dH2hpE+RXmpvZmY2ZPTmDPWfgYslfYzVCXQqMBL4+94GZmZmNpD0OKFGxOPAOyX9LbBtLr4sIq4qEpmZmdkAUuJdvlcDVxeIxczMbMDyG43MzMwKcEI1MzMrwAnVzMysACdUMzOzApxQzczMCnBCNTMzK8AJ1czMrAAnVDMzswKcUM3MzApwQjUzMyvACdXMzKwAJ1QzM7MCnFDNzMwKcEI1MzMrwAnVzMysACdUMzOzAtoyoUoaLun3kn6V+ydLuknSQkkXSBqZy9fO/Qvz8EmVNo7J5fdK2qs1U2JmZkNFWyZU4EvA3ZX+k4FTIuKNwErg0Fx+KLAyl5+S6yFpa+BAYBtgb+B0ScP7KXYzMxuC2i6hSpoAvA84K/cL2A2Yk6vMBPbL3dNyP3n47rn+NGB2RLwUEQ8BC4Gd+mcKzMxsKGq7hAp8DzgSeDX3bwI8FRGrcv9iYHzuHg8sAsjDn871XytvME4Hkg6TNF/S/OXLl5ecDjMzG0LaKqFKej+wLCIW9Nd3RsSZETE1IqaOHTu2v77WzMwGmRGtDqDOu4APStoXWAfYEPg+MErSiHwWOgFYkusvASYCiyWNADYCnqyU11THMTMzK66tzlAj4piImBARk0gPFV0VER8Drgb2z9WmA5fk7ktzP3n4VRERufzA/BTwZGAKcHM/TYaZmQ1B7XaG2sxRwGxJJwC/B87O5WcD50paCKwgJWEi4k5JFwJ3AauAwyPilf4P28zMhoq2TagRcQ1wTe5+kAZP6UbEi8CHmox/InBi30VoZma2Wltd8jUzMxuonFDNzMwKcEI1MzMrwAnVzMysACdUMzOzApxQzczMCnBCNTMzK8AJ1czMrAAnVDMzswKcUM3MzApwQjUzMyvACdXMzKwAJ1QzM7MCnFDNzMwKcEI1MzMrwAnVzMysACdUMzOzApxQzczMCnBCNTMzK8AJ1czMrAAnVDMzswKcUM3MzApwQjUzMyvACdXMzKwAJ1QzM7MCnFDNzMwKcEI1MzMroK0SqqSJkq6WdJekOyV9KZdvLGmupPvz39G5XJJOlbRQ0m2Sdqy0NT3Xv1/S9FZNk5mZDQ1tlVCBVcBXImJrYBfgcElbA0cDV0bEFODK3A+wDzAlfw4DzoCUgIHjgJ2BnYDjaknYzMysL7RVQo2IpRHxu9z9LHA3MB6YBszM1WYC++XuacCsSOYBoyRtCuwFzI2IFRGxEpgL7N2Pk2JmZkNMWyXUKkmTgB2Am4BxEbE0D3oMGJe7xwOLKqMtzmXNyht9z2GS5kuav3z58mLxm5nZ0NKWCVXS+sDPgX+OiGeqwyIigCj1XRFxZkRMjYipY8eOLdWsmZkNMW2XUCWtRUqm50XERbn48Xwpl/x3WS5fAkysjD4hlzUrNzMz6xNtlVAlCTgbuDsi/rMy6FKg9qTudOCSSvlB+WnfXYCn86Xhy4E9JY3ODyPtmcvMzMz6xIhWB1DnXcAngNsl3ZrLvgacBFwo6VDgEeCAPOzXwL7AQuB54BCAiFgh6ZvALbneNyJiRf9MgpmZDUVtlVAj4npATQbv3qB+AIc3aWsGMKNcdGZmZs211SVfMzOzgcoJ1czMrAAnVDMzswKcUM3MzApwQjUzMyvACdXMzKwAJ1QzM7MCnFDNzMwKcEI1MzMrwAnVzMysACdUMzOzApxQzczMCnBCNTMzK8AJ1czMrAAnVDMzswKcUM3MzApwQjUzMyvACdXMzKwAJ1QzM7MCnFDNzMwKcEI1MzMrwAnVzMysACdUMzOzApxQzczMCnBCNTMzK8AJ1czMrAAnVDMzswIGdUKVtLekeyUtlHR0q+MxM7PBa9AmVEnDgdOAfYCtgY9I2rq1UZmZ2WA1aBMqsBOwMCIejIg/A7OBaS2OyczMBqkRrQ6gD40HFlX6FwM711eSdBhwWO59TtK9/RBbT40Bnmh1EN00UGLt8zh1cpFmPD/LGyixDoR1dIsCYQx4gzmhdktEnAmc2eo4ukPS/IiY2uo4umOgxOo4yxooccLAiXWgxGmD+5LvEmBipX9CLjMzMytuMCfUW4ApkiZLGgkcCFza4pjMzGyQGrSXfCNilaTPA5cDw4EZEXFni8PqrQFxaTobKLE6zrIGSpwwcGIdKHEOeYqIVsdgZmY24A3mS75mZmb9xgnVzMysACfUNiNpY0lzJd2f/45uUGd7STdKulPSbZI+XBl2jqSHJN2aP9sXjq/T1zlKWlvSBXn4TZImVYYdk8vvlbRXybh6EOeXJd2V59+VkraoDHulMv/6/EG2bsR6sKTllZg+VRk2Pa8r90ua3uI4T6nEeJ+kpyrD+m2eSpohaZmkO5oMl6RT83nGIVsAAAXtSURBVHTcJmnHyrD+nJ9dxfmxHN/tkm6QtF1l2MO5/FZJ8/syTlsDEeFPG32AfweOzt1HAyc3qPMmYEru3gxYCozK/ecA+/dRbMOBB4AtgZHAH4Ct6+p8DvhR7j4QuCB3b53rrw1Mzu0Mb2Gcfwusl7v/qRZn7n+uH5d3d2I9GPhhg3E3Bh7Mf0fn7tGtirOu/hdIDwK2Yp6+B9gRuKPJ8H2B3wACdgFu6u/52c0431n7ftIrVG+qDHsYGNNf89Sf7n18htp+pgEzc/dMYL/6ChFxX0Tcn7sfBZYBY/shtu68zrEa/xxgd0nK5bMj4qWIeAhYmNtrSZwRcXVEPJ9755H+T7kVevOKzL2AuRGxIiJWAnOBvdskzo8A5/dRLJ2KiOuAFZ1UmQbMimQeMErSpvTv/Owyzoi4IccBrV1HrZucUNvPuIhYmrsfA8Z1VlnSTqQzhgcqxSfmS0WnSFq7YGyNXuc4vlmdiFgFPA1s0s1x+zPOqkNJZyw160iaL2mepL84oCmsu7H+Y16mcyTVXljSlvM0Xz6fDFxVKe7PedqVZtPSn/NzTdWvowH8VtKC/PpUawOD9v9Q25mkK4DXNxh0bLUnIkJS0/9rykfV5wLTI+LVXHwMKRGPJP3/2lHAN0rEPRhJ+jgwFXhvpXiLiFgiaUvgKkm3R8QDjVvoF78Ezo+IlyR9hnQFYLcWxtOVA4E5EfFKpazd5umAIelvSQn13ZXid+f5+VfAXEn35DNeayGfobZAROwREds2+FwCPJ4TZS1hLmvUhqQNgcuAY/Nlq1rbS/OlrJeAn1D2smp3Xuf4Wh1JI4CNgCe7OW5/xomkPUgHMR/M8wuAiFiS/z4IXAPs0EdxdivWiHiyEt9ZwNu6O25/xllxIHWXe/t5nnal2bS03etKJf01aZlPi4gna+WV+bkMuJi+u31ia8AJtf1cCtSeLpwOXFJfQelViheT7gPNqRtWS8Yi3X9t+ARhD3XndY7V+PcHroqIyOUH5qeAJwNTgJsLxrZGcUraAfgxKZkuq5SPrl0mlzQGeBdwVx/F2d1YN630fhC4O3dfDuyZYx4N7JnLWhJnjvUtpAd6bqyU9fc87cqlwEH5ad9dgKfzbZb+nJ9dkrQ5cBHwiYi4r1L+Okkb1LpJcZbczq2nWv1UlD8dP6T7jVcC9wNXABvn8qnAWbn748DLwK2Vz/Z52FXA7aQN7KfA+oXj2xe4j3TP9thc9g1SYgJYB/gZ6aGjm4EtK+Mem8e7F9inj+djV3FeATxemX+X5vJ35vn3h/z30H5Y5l3F+m3gzhzT1cBbKuN+Ms/rhcAhrYwz9x8PnFQ3Xr/OU9LZ8dK8jSwmXS79LPDZPFzAaXk6bgemtmh+dhXnWcDKyjo6P5dvmeflH/J6cWxfr6P+dO/jVw+amZkV4Eu+ZmZmBTihmpmZFeCEamZmVoATqpmZWQFOqGZmZgU4oZr1E0nPrUHd4yV9ta/aN7PynFDNzMwKcEI1ayFJH1D63djfS7pCUvXHELZT+t3b+yV9ujLOEZJuyS/L/38N2txU0nX5tzLvkPQ3/TIxZkOcE6pZa10P7BIRO5B+Eu3IyrC/Jr0E/x3Av0naTNKepNc27gRsD7xN0nvq2vwocHlEbA9sR3rLjpn1Mf/ajFlrTQAuyO/rHQk8VBl2SUS8ALwg6WpSEn036d2tv8911icl2OovjdwCzJC0FvCLiHBCNesHPkM1a60fAD+MiLcCnyG9C7mm/r2gQXoP7bcjYvv8eWNEnN2hUvoZr/eQfinlHEkH9V34ZlbjhGrWWhux+ifCptcNmyZpHUmbALuSzjwvBz4paX0ASePzb2K+Jv/A9+MR8V+kF6zv2Ifxm1nmS75m/Wc9SYsr/f9J+nWWn0laSfqloMmV4beRfl1mDPDNiHgUeFTSVsCN6Rf6eI7060PV383dFThC0st5uM9QzfqBf23GzMysAF/yNTMzK8AJ1czMrAAnVDMzswKcUM3MzApwQjUzMyvACdXMzKwAJ1QzM7MC/g868cnBjoSTCgAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eClUPjVDfjTM"
      },
      "source": [
        "twitter_df = twitter_df.drop(['index', 'id', 'Annotation'], axis=1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "d9DTiiV3g1hv",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        },
        "outputId": "9cbf5f40-5b50-4109-9fc0-94a72bb00852"
      },
      "source": [
        "twitter_df"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Text</th>\n",
              "      <th>oh_label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>@halalflaws @biebervalue @greenlinerzjm I read...</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>@ShreyaBafna3 Now you idiots claim that people...</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>RT @Mooseoftorment Call me sexist, but when I ...</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>@g0ssipsquirrelx Wrong, ISIS follows the examp...</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>#mkr No No No No No No</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16846</th>\n",
              "      <td>Feeling so sorry for the girls, they should be...</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16847</th>\n",
              "      <td>#MKR 'pretty good dishes we're happy with' - O...</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16848</th>\n",
              "      <td>RT @colonelkickhead: Deconstructed lemon tart!...</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16849</th>\n",
              "      <td>@versacezaynx @nyazpolitics @greenlinerzjm You...</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16850</th>\n",
              "      <td>And before you protest that you're *not* mad, ...</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>16851 rows × 2 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                    Text  oh_label\n",
              "0      @halalflaws @biebervalue @greenlinerzjm I read...       0.0\n",
              "1      @ShreyaBafna3 Now you idiots claim that people...       0.0\n",
              "2      RT @Mooseoftorment Call me sexist, but when I ...       1.0\n",
              "3      @g0ssipsquirrelx Wrong, ISIS follows the examp...       1.0\n",
              "4                                 #mkr No No No No No No       0.0\n",
              "...                                                  ...       ...\n",
              "16846  Feeling so sorry for the girls, they should be...       0.0\n",
              "16847  #MKR 'pretty good dishes we're happy with' - O...       0.0\n",
              "16848  RT @colonelkickhead: Deconstructed lemon tart!...       0.0\n",
              "16849  @versacezaynx @nyazpolitics @greenlinerzjm You...       0.0\n",
              "16850  And before you protest that you're *not* mad, ...       0.0\n",
              "\n",
              "[16851 rows x 2 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gpd-AlLnbcef"
      },
      "source": [
        "## Aggression Data Cleaning"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "p3N-hly8aYcf"
      },
      "source": [
        "aggression_df = pd.read_csv(\"./aggression_parsed_dataset.csv\")\n",
        "aggression_df['Source'] = aggression_df.shape[0]*['aggression_df']"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xHkmj6UYbA8X",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "206fdd69-e2c5-4748-e795-6ce31b8361fc"
      },
      "source": [
        "aggression_df"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>index</th>\n",
              "      <th>Text</th>\n",
              "      <th>ed_label_0</th>\n",
              "      <th>ed_label_1</th>\n",
              "      <th>oh_label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0</td>\n",
              "      <td>`- This is not ``creative``.  Those are the di...</td>\n",
              "      <td>0.900000</td>\n",
              "      <td>0.100000</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1</td>\n",
              "      <td>`  :: the term ``standard model`` is itself le...</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>2</td>\n",
              "      <td>True or false, the situation as of March 200...</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>3</td>\n",
              "      <td>Next, maybe you could work on being less cond...</td>\n",
              "      <td>0.555556</td>\n",
              "      <td>0.444444</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>4</td>\n",
              "      <td>This page will need disambiguation.</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>115859</th>\n",
              "      <td>115859</td>\n",
              "      <td>`   These sources don't exactly exude a sense ...</td>\n",
              "      <td>0.777778</td>\n",
              "      <td>0.222222</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>115860</th>\n",
              "      <td>115860</td>\n",
              "      <td>The Institute for Historical Review is a pee...</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>115861</th>\n",
              "      <td>115861</td>\n",
              "      <td>:The way you're trying to describe it in this...</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>115862</th>\n",
              "      <td>115862</td>\n",
              "      <td>== Warning ==  There is clearly a protection...</td>\n",
              "      <td>0.500000</td>\n",
              "      <td>0.500000</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>115863</th>\n",
              "      <td>115863</td>\n",
              "      <td>Alternate option=== Is there perhaps enough ne...</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>115864 rows × 5 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "         index  ... oh_label\n",
              "0            0  ...        0\n",
              "1            1  ...        0\n",
              "2            2  ...        0\n",
              "3            3  ...        0\n",
              "4            4  ...        0\n",
              "...        ...  ...      ...\n",
              "115859  115859  ...        0\n",
              "115860  115860  ...        0\n",
              "115861  115861  ...        0\n",
              "115862  115862  ...        0\n",
              "115863  115863  ...        0\n",
              "\n",
              "[115864 rows x 5 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 15
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PhA2SLQZemzS"
      },
      "source": [
        "aggression_df = aggression_df.drop(['ed_label_0', 'ed_label_1', 'index'], axis=1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "97GLbhQ0gUx5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        },
        "outputId": "e9ffa0a9-700b-4f4c-ae77-0ddc16726270"
      },
      "source": [
        "aggression_df\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Text</th>\n",
              "      <th>oh_label</th>\n",
              "      <th>Source</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>`- This is not ``creative``.  Those are the di...</td>\n",
              "      <td>0</td>\n",
              "      <td>aggression_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>`  :: the term ``standard model`` is itself le...</td>\n",
              "      <td>0</td>\n",
              "      <td>aggression_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>True or false, the situation as of March 200...</td>\n",
              "      <td>0</td>\n",
              "      <td>aggression_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>Next, maybe you could work on being less cond...</td>\n",
              "      <td>0</td>\n",
              "      <td>aggression_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>This page will need disambiguation.</td>\n",
              "      <td>0</td>\n",
              "      <td>aggression_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>115859</th>\n",
              "      <td>`   These sources don't exactly exude a sense ...</td>\n",
              "      <td>0</td>\n",
              "      <td>aggression_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>115860</th>\n",
              "      <td>The Institute for Historical Review is a pee...</td>\n",
              "      <td>0</td>\n",
              "      <td>aggression_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>115861</th>\n",
              "      <td>:The way you're trying to describe it in this...</td>\n",
              "      <td>0</td>\n",
              "      <td>aggression_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>115862</th>\n",
              "      <td>== Warning ==  There is clearly a protection...</td>\n",
              "      <td>0</td>\n",
              "      <td>aggression_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>115863</th>\n",
              "      <td>Alternate option=== Is there perhaps enough ne...</td>\n",
              "      <td>0</td>\n",
              "      <td>aggression_df</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>115864 rows × 3 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                     Text  ...         Source\n",
              "0       `- This is not ``creative``.  Those are the di...  ...  aggression_df\n",
              "1       `  :: the term ``standard model`` is itself le...  ...  aggression_df\n",
              "2         True or false, the situation as of March 200...  ...  aggression_df\n",
              "3        Next, maybe you could work on being less cond...  ...  aggression_df\n",
              "4                    This page will need disambiguation.   ...  aggression_df\n",
              "...                                                   ...  ...            ...\n",
              "115859  `   These sources don't exactly exude a sense ...  ...  aggression_df\n",
              "115860    The Institute for Historical Review is a pee...  ...  aggression_df\n",
              "115861   :The way you're trying to describe it in this...  ...  aggression_df\n",
              "115862    == Warning ==  There is clearly a protection...  ...  aggression_df\n",
              "115863  Alternate option=== Is there perhaps enough ne...  ...  aggression_df\n",
              "\n",
              "[115864 rows x 3 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 19
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0F-6qZztgGu6"
      },
      "source": [
        "## Attack Dataset Cleaning"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9Y_EP-kZfC7T"
      },
      "source": [
        "attack_df = pd.read_csv(\"./attack_parsed_dataset.csv\")\n",
        "attack_df['Source'] = attack_df.shape[0]*['attack_df']"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7KAI-Gh1hFnT",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        },
        "outputId": "b6cdb65c-edc7-4d04-bf03-b115080476c4"
      },
      "source": [
        "attack_df"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>index</th>\n",
              "      <th>Text</th>\n",
              "      <th>ed_label_0</th>\n",
              "      <th>ed_label_1</th>\n",
              "      <th>oh_label</th>\n",
              "      <th>Source</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0</td>\n",
              "      <td>`- This is not ``creative``.  Those are the di...</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0</td>\n",
              "      <td>attack_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1</td>\n",
              "      <td>`  :: the term ``standard model`` is itself le...</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0</td>\n",
              "      <td>attack_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>2</td>\n",
              "      <td>True or false, the situation as of March 200...</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0</td>\n",
              "      <td>attack_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>3</td>\n",
              "      <td>Next, maybe you could work on being less cond...</td>\n",
              "      <td>0.555556</td>\n",
              "      <td>0.444444</td>\n",
              "      <td>0</td>\n",
              "      <td>attack_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>4</td>\n",
              "      <td>This page will need disambiguation.</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0</td>\n",
              "      <td>attack_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>115859</th>\n",
              "      <td>115859</td>\n",
              "      <td>`   These sources don't exactly exude a sense ...</td>\n",
              "      <td>0.888889</td>\n",
              "      <td>0.111111</td>\n",
              "      <td>0</td>\n",
              "      <td>attack_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>115860</th>\n",
              "      <td>115860</td>\n",
              "      <td>The Institute for Historical Review is a pee...</td>\n",
              "      <td>0.900000</td>\n",
              "      <td>0.100000</td>\n",
              "      <td>0</td>\n",
              "      <td>attack_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>115861</th>\n",
              "      <td>115861</td>\n",
              "      <td>:The way you're trying to describe it in this...</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0</td>\n",
              "      <td>attack_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>115862</th>\n",
              "      <td>115862</td>\n",
              "      <td>== Warning ==  There is clearly a protection...</td>\n",
              "      <td>0.800000</td>\n",
              "      <td>0.200000</td>\n",
              "      <td>0</td>\n",
              "      <td>attack_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>115863</th>\n",
              "      <td>115863</td>\n",
              "      <td>Alternate option=== Is there perhaps enough ne...</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0</td>\n",
              "      <td>attack_df</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>115864 rows × 6 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "         index  ...     Source\n",
              "0            0  ...  attack_df\n",
              "1            1  ...  attack_df\n",
              "2            2  ...  attack_df\n",
              "3            3  ...  attack_df\n",
              "4            4  ...  attack_df\n",
              "...        ...  ...        ...\n",
              "115859  115859  ...  attack_df\n",
              "115860  115860  ...  attack_df\n",
              "115861  115861  ...  attack_df\n",
              "115862  115862  ...  attack_df\n",
              "115863  115863  ...  attack_df\n",
              "\n",
              "[115864 rows x 6 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 21
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WkWaZnjChIK5"
      },
      "source": [
        "attack_df = attack_df.drop(['ed_label_0', 'ed_label_1', 'index'], axis=1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "F3ABBE0zhICb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        },
        "outputId": "89e17f0c-6ca8-4790-95ec-9c7e9b46e8a9"
      },
      "source": [
        "attack_df"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Text</th>\n",
              "      <th>oh_label</th>\n",
              "      <th>Source</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>`- This is not ``creative``.  Those are the di...</td>\n",
              "      <td>0</td>\n",
              "      <td>attack_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>`  :: the term ``standard model`` is itself le...</td>\n",
              "      <td>0</td>\n",
              "      <td>attack_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>True or false, the situation as of March 200...</td>\n",
              "      <td>0</td>\n",
              "      <td>attack_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>Next, maybe you could work on being less cond...</td>\n",
              "      <td>0</td>\n",
              "      <td>attack_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>This page will need disambiguation.</td>\n",
              "      <td>0</td>\n",
              "      <td>attack_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>115859</th>\n",
              "      <td>`   These sources don't exactly exude a sense ...</td>\n",
              "      <td>0</td>\n",
              "      <td>attack_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>115860</th>\n",
              "      <td>The Institute for Historical Review is a pee...</td>\n",
              "      <td>0</td>\n",
              "      <td>attack_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>115861</th>\n",
              "      <td>:The way you're trying to describe it in this...</td>\n",
              "      <td>0</td>\n",
              "      <td>attack_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>115862</th>\n",
              "      <td>== Warning ==  There is clearly a protection...</td>\n",
              "      <td>0</td>\n",
              "      <td>attack_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>115863</th>\n",
              "      <td>Alternate option=== Is there perhaps enough ne...</td>\n",
              "      <td>0</td>\n",
              "      <td>attack_df</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>115864 rows × 3 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                     Text  oh_label     Source\n",
              "0       `- This is not ``creative``.  Those are the di...         0  attack_df\n",
              "1       `  :: the term ``standard model`` is itself le...         0  attack_df\n",
              "2         True or false, the situation as of March 200...         0  attack_df\n",
              "3        Next, maybe you could work on being less cond...         0  attack_df\n",
              "4                    This page will need disambiguation.          0  attack_df\n",
              "...                                                   ...       ...        ...\n",
              "115859  `   These sources don't exactly exude a sense ...         0  attack_df\n",
              "115860    The Institute for Historical Review is a pee...         0  attack_df\n",
              "115861   :The way you're trying to describe it in this...         0  attack_df\n",
              "115862    == Warning ==  There is clearly a protection...         0  attack_df\n",
              "115863  Alternate option=== Is there perhaps enough ne...         0  attack_df\n",
              "\n",
              "[115864 rows x 3 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 23
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "N_2SPNk4hTsk"
      },
      "source": [
        "## Toxicity Data Cleaning"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "O3vRelkehWRR"
      },
      "source": [
        "toxicity_df = pd.read_csv(\"./toxicity_parsed_dataset.csv\")\n",
        "toxicity_df['Source'] = toxicity_df.shape[0]*['toxicity_df']"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hyFE94yVhdnv",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        },
        "outputId": "b558588c-ffd9-4f8d-f02f-d4af5b226194"
      },
      "source": [
        "toxicity_df "
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>index</th>\n",
              "      <th>Text</th>\n",
              "      <th>ed_label_0</th>\n",
              "      <th>ed_label_1</th>\n",
              "      <th>oh_label</th>\n",
              "      <th>Source</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0</td>\n",
              "      <td>This: :One can make an analogy in mathematical...</td>\n",
              "      <td>0.9</td>\n",
              "      <td>0.1</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1</td>\n",
              "      <td>`  :Clarification for you  (and Zundark's righ...</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>2</td>\n",
              "      <td>Elected or Electoral? JHK</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>3</td>\n",
              "      <td>`This is such a fun entry.   Devotchka  I once...</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>4</td>\n",
              "      <td>Please relate the ozone hole to increases in c...</td>\n",
              "      <td>0.8</td>\n",
              "      <td>0.2</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>159681</th>\n",
              "      <td>159681</td>\n",
              "      <td>`   These sources don't exactly exude a sense ...</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>159682</th>\n",
              "      <td>159682</td>\n",
              "      <td>The Institute for Historical Review is a pee...</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>159683</th>\n",
              "      <td>159683</td>\n",
              "      <td>:The way you're trying to describe it in this...</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>159684</th>\n",
              "      <td>159684</td>\n",
              "      <td>== Warning ==  There is clearly a protection...</td>\n",
              "      <td>0.6</td>\n",
              "      <td>0.4</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>159685</th>\n",
              "      <td>159685</td>\n",
              "      <td>Alternate option=== Is there perhaps enough ne...</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>159686 rows × 6 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "         index  ...       Source\n",
              "0            0  ...  toxicity_df\n",
              "1            1  ...  toxicity_df\n",
              "2            2  ...  toxicity_df\n",
              "3            3  ...  toxicity_df\n",
              "4            4  ...  toxicity_df\n",
              "...        ...  ...          ...\n",
              "159681  159681  ...  toxicity_df\n",
              "159682  159682  ...  toxicity_df\n",
              "159683  159683  ...  toxicity_df\n",
              "159684  159684  ...  toxicity_df\n",
              "159685  159685  ...  toxicity_df\n",
              "\n",
              "[159686 rows x 6 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 25
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cy2Lk6ubrL0r",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "bed6fa07-4c5e-4528-f563-78b46ae26b0c"
      },
      "source": [
        "n = len(pd.unique(toxicity_df['Text']))\n",
        "print(\"Unique -\", n)\n",
        "print(\"Total -\", toxicity_df.shape[0])\n",
        "print(\"Duplicates\", toxicity_df.shape[0] - n)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Unique - 159388\n",
            "Total - 159686\n",
            "Duplicates 298\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AJg2YdchsDMY"
      },
      "source": [
        "check_df = toxicity_df[toxicity_df[['Text', 'oh_label']].duplicated(keep='last')]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5AiCQWpcvAU9",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "d2b0893d-feee-4e9c-b3f0-a314d53bdf0d"
      },
      "source": [
        "checkcyberbully = check_df[check_df.oh_label==1.0].count()[0]\n",
        "checknocyberbully = check_df[check_df.oh_label==0.0].count()[0]\n",
        "print(checkcyberbully, checknocyberbully)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "14 283\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Qs2DAFd7sfm2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 142
        },
        "outputId": "3e916f1c-1a46-411c-a471-df48241a0f4d"
      },
      "source": [
        "toxicity_df[toxicity_df['Text']==\"Hello, welcome to Wikipedia. Here are some useful links in case you haven't already found them:  *How to edit a page *How to write a great article *Naming conventions *Manual of Style  If you have any questions, see the help pages, add a question to the village pump or ask me on . I hope you enjoy editing here and being a Wikipedian!  Tip: you can sign your name with ~~~~    \"]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>index</th>\n",
              "      <th>Text</th>\n",
              "      <th>ed_label_0</th>\n",
              "      <th>ed_label_1</th>\n",
              "      <th>oh_label</th>\n",
              "      <th>Source</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>401</th>\n",
              "      <td>401</td>\n",
              "      <td>Hello, welcome to Wikipedia. Here are some use...</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3016</th>\n",
              "      <td>3016</td>\n",
              "      <td>Hello, welcome to Wikipedia. Here are some use...</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3019</th>\n",
              "      <td>3019</td>\n",
              "      <td>Hello, welcome to Wikipedia. Here are some use...</td>\n",
              "      <td>0.9</td>\n",
              "      <td>0.1</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "      index  ...       Source\n",
              "401     401  ...  toxicity_df\n",
              "3016   3016  ...  toxicity_df\n",
              "3019   3019  ...  toxicity_df\n",
              "\n",
              "[3 rows x 6 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 29
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sqor-mftssiU",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 86
        },
        "outputId": "a6084309-23a5-47b9-f791-ea0c3c074e31"
      },
      "source": [
        "toxicity_df.iloc[1188]['Text']"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'Welcome!  Hello, and welcome to Wikipedia. Thank you for your contributions. I hope you like the place and decide to stay. Here are a few good links for newcomers:  *How to edit a page *Editing tutorial *Picture tutorial *How to write a great article *Naming conventions *Manual of Style  I hope you enjoy editing here and being a Wikipedian! By the way, you can sign your name on Talk and vote pages using three tildes, like this: ~~~. Four tildes (~~~~) produces your name and the current date. If you have any questions, see the help pages, add a question to the village pump or ask me on my Talk page. Again, welcome!   |  '"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 30
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9hpB_t0Whg1W"
      },
      "source": [
        "toxicity_df = toxicity_df.drop(['ed_label_0', 'ed_label_1', 'index'], axis=1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xdEi1iTGhofU",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        },
        "outputId": "772189b0-9e84-443a-86e0-2ca51bb3c490"
      },
      "source": [
        "toxicity_df"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Text</th>\n",
              "      <th>oh_label</th>\n",
              "      <th>Source</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>This: :One can make an analogy in mathematical...</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>`  :Clarification for you  (and Zundark's righ...</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>Elected or Electoral? JHK</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>`This is such a fun entry.   Devotchka  I once...</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>Please relate the ozone hole to increases in c...</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>159681</th>\n",
              "      <td>`   These sources don't exactly exude a sense ...</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>159682</th>\n",
              "      <td>The Institute for Historical Review is a pee...</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>159683</th>\n",
              "      <td>:The way you're trying to describe it in this...</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>159684</th>\n",
              "      <td>== Warning ==  There is clearly a protection...</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>159685</th>\n",
              "      <td>Alternate option=== Is there perhaps enough ne...</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>159686 rows × 3 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                     Text  ...       Source\n",
              "0       This: :One can make an analogy in mathematical...  ...  toxicity_df\n",
              "1       `  :Clarification for you  (and Zundark's righ...  ...  toxicity_df\n",
              "2                               Elected or Electoral? JHK  ...  toxicity_df\n",
              "3       `This is such a fun entry.   Devotchka  I once...  ...  toxicity_df\n",
              "4       Please relate the ozone hole to increases in c...  ...  toxicity_df\n",
              "...                                                   ...  ...          ...\n",
              "159681  `   These sources don't exactly exude a sense ...  ...  toxicity_df\n",
              "159682    The Institute for Historical Review is a pee...  ...  toxicity_df\n",
              "159683   :The way you're trying to describe it in this...  ...  toxicity_df\n",
              "159684    == Warning ==  There is clearly a protection...  ...  toxicity_df\n",
              "159685  Alternate option=== Is there perhaps enough ne...  ...  toxicity_df\n",
              "\n",
              "[159686 rows x 3 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 32
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sDvo34lUhyO_"
      },
      "source": [
        "## Youtube Data Cleaning"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Qy34BUg3hw-H"
      },
      "source": [
        "youtube_df = pd.read_csv(\"./youtube_parsed_dataset.csv\")\n",
        "youtube_df['Source'] = youtube_df.shape[0]*['youtube_df']"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fxWxHlrxh6eH",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 589
        },
        "outputId": "3a042513-e599-434c-c1a5-8b2eeb74a79e"
      },
      "source": [
        "youtube_df"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>index</th>\n",
              "      <th>UserIndex</th>\n",
              "      <th>Text</th>\n",
              "      <th>Number of Comments</th>\n",
              "      <th>Number of Subscribers</th>\n",
              "      <th>Membership Duration</th>\n",
              "      <th>Number of Uploads</th>\n",
              "      <th>Profanity in UserID</th>\n",
              "      <th>Age</th>\n",
              "      <th>oh_label</th>\n",
              "      <th>Source</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0</td>\n",
              "      <td>X1</td>\n",
              "      <td>Does N.e.bodyelse Hear her Crazy ass Screamin ...</td>\n",
              "      <td>10</td>\n",
              "      <td>1</td>\n",
              "      <td>3</td>\n",
              "      <td>3</td>\n",
              "      <td>0</td>\n",
              "      <td>15</td>\n",
              "      <td>0</td>\n",
              "      <td>youtube_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1</td>\n",
              "      <td>X2</td>\n",
              "      <td>There are so many things that are incorrect wi...</td>\n",
              "      <td>3</td>\n",
              "      <td>0</td>\n",
              "      <td>6</td>\n",
              "      <td>5</td>\n",
              "      <td>0</td>\n",
              "      <td>31</td>\n",
              "      <td>0</td>\n",
              "      <td>youtube_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>2</td>\n",
              "      <td>X3</td>\n",
              "      <td>3:26 hahah my boyfriend showed this song to me...</td>\n",
              "      <td>7</td>\n",
              "      <td>0</td>\n",
              "      <td>3</td>\n",
              "      <td>5</td>\n",
              "      <td>0</td>\n",
              "      <td>43</td>\n",
              "      <td>1</td>\n",
              "      <td>youtube_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>3</td>\n",
              "      <td>X2218</td>\n",
              "      <td>dick beyonce fuck y a ass hole you are truely ...</td>\n",
              "      <td>34</td>\n",
              "      <td>0</td>\n",
              "      <td>3</td>\n",
              "      <td>5</td>\n",
              "      <td>0</td>\n",
              "      <td>44</td>\n",
              "      <td>1</td>\n",
              "      <td>youtube_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>4</td>\n",
              "      <td>X5</td>\n",
              "      <td>DongHaeTaemin and Kai ;A; luhansehun and bacon...</td>\n",
              "      <td>11</td>\n",
              "      <td>173</td>\n",
              "      <td>5</td>\n",
              "      <td>5</td>\n",
              "      <td>0</td>\n",
              "      <td>21</td>\n",
              "      <td>0</td>\n",
              "      <td>youtube_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3459</th>\n",
              "      <td>3464</td>\n",
              "      <td>X3465</td>\n",
              "      <td>She pretty I love this song I miss the old kel...</td>\n",
              "      <td>15</td>\n",
              "      <td>2</td>\n",
              "      <td>4</td>\n",
              "      <td>7</td>\n",
              "      <td>0</td>\n",
              "      <td>23</td>\n",
              "      <td>1</td>\n",
              "      <td>youtube_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3460</th>\n",
              "      <td>3465</td>\n",
              "      <td>X3466</td>\n",
              "      <td>Status-Online Im ZxkillergirlzX! I'm Zxkillerg...</td>\n",
              "      <td>4</td>\n",
              "      <td>28</td>\n",
              "      <td>4</td>\n",
              "      <td>23</td>\n",
              "      <td>1</td>\n",
              "      <td>15</td>\n",
              "      <td>0</td>\n",
              "      <td>youtube_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3461</th>\n",
              "      <td>3466</td>\n",
              "      <td>X3467</td>\n",
              "      <td>JR so cute EXO M Better I agree like yeah yeah...</td>\n",
              "      <td>23</td>\n",
              "      <td>0</td>\n",
              "      <td>5</td>\n",
              "      <td>3</td>\n",
              "      <td>0</td>\n",
              "      <td>33</td>\n",
              "      <td>0</td>\n",
              "      <td>youtube_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3462</th>\n",
              "      <td>3467</td>\n",
              "      <td>X3468</td>\n",
              "      <td>! !</td>\n",
              "      <td>5</td>\n",
              "      <td>0</td>\n",
              "      <td>6</td>\n",
              "      <td>5</td>\n",
              "      <td>0</td>\n",
              "      <td>38</td>\n",
              "      <td>0</td>\n",
              "      <td>youtube_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3463</th>\n",
              "      <td>3468</td>\n",
              "      <td>X3469</td>\n",
              "      <td>great video and MERRY CHRISTMAS from greece :*...</td>\n",
              "      <td>27</td>\n",
              "      <td>220</td>\n",
              "      <td>4</td>\n",
              "      <td>30</td>\n",
              "      <td>0</td>\n",
              "      <td>18</td>\n",
              "      <td>0</td>\n",
              "      <td>youtube_df</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>3464 rows × 11 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "      index UserIndex  ... oh_label      Source\n",
              "0         0        X1  ...        0  youtube_df\n",
              "1         1        X2  ...        0  youtube_df\n",
              "2         2        X3  ...        1  youtube_df\n",
              "3         3     X2218  ...        1  youtube_df\n",
              "4         4        X5  ...        0  youtube_df\n",
              "...     ...       ...  ...      ...         ...\n",
              "3459   3464     X3465  ...        1  youtube_df\n",
              "3460   3465     X3466  ...        0  youtube_df\n",
              "3461   3466     X3467  ...        0  youtube_df\n",
              "3462   3467     X3468  ...        0  youtube_df\n",
              "3463   3468     X3469  ...        0  youtube_df\n",
              "\n",
              "[3464 rows x 11 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 34
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CM7dkmjoh7Nq"
      },
      "source": [
        "youtube_df = youtube_df.drop(['index', 'UserIndex', 'Number of Comments', 'Number of Subscribers', 'Membership Duration', 'Number of Uploads', 'Profanity in UserID', 'Age'], axis=1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "p_nT_WijiUTv",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        },
        "outputId": "317b209e-ff96-4aa8-fa6a-474fd5ef9017"
      },
      "source": [
        "youtube_df"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Text</th>\n",
              "      <th>oh_label</th>\n",
              "      <th>Source</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>Does N.e.bodyelse Hear her Crazy ass Screamin ...</td>\n",
              "      <td>0</td>\n",
              "      <td>youtube_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>There are so many things that are incorrect wi...</td>\n",
              "      <td>0</td>\n",
              "      <td>youtube_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>3:26 hahah my boyfriend showed this song to me...</td>\n",
              "      <td>1</td>\n",
              "      <td>youtube_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>dick beyonce fuck y a ass hole you are truely ...</td>\n",
              "      <td>1</td>\n",
              "      <td>youtube_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>DongHaeTaemin and Kai ;A; luhansehun and bacon...</td>\n",
              "      <td>0</td>\n",
              "      <td>youtube_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3459</th>\n",
              "      <td>She pretty I love this song I miss the old kel...</td>\n",
              "      <td>1</td>\n",
              "      <td>youtube_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3460</th>\n",
              "      <td>Status-Online Im ZxkillergirlzX! I'm Zxkillerg...</td>\n",
              "      <td>0</td>\n",
              "      <td>youtube_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3461</th>\n",
              "      <td>JR so cute EXO M Better I agree like yeah yeah...</td>\n",
              "      <td>0</td>\n",
              "      <td>youtube_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3462</th>\n",
              "      <td>! !</td>\n",
              "      <td>0</td>\n",
              "      <td>youtube_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3463</th>\n",
              "      <td>great video and MERRY CHRISTMAS from greece :*...</td>\n",
              "      <td>0</td>\n",
              "      <td>youtube_df</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>3464 rows × 3 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                   Text  oh_label      Source\n",
              "0     Does N.e.bodyelse Hear her Crazy ass Screamin ...         0  youtube_df\n",
              "1     There are so many things that are incorrect wi...         0  youtube_df\n",
              "2     3:26 hahah my boyfriend showed this song to me...         1  youtube_df\n",
              "3     dick beyonce fuck y a ass hole you are truely ...         1  youtube_df\n",
              "4     DongHaeTaemin and Kai ;A; luhansehun and bacon...         0  youtube_df\n",
              "...                                                 ...       ...         ...\n",
              "3459  She pretty I love this song I miss the old kel...         1  youtube_df\n",
              "3460  Status-Online Im ZxkillergirlzX! I'm Zxkillerg...         0  youtube_df\n",
              "3461  JR so cute EXO M Better I agree like yeah yeah...         0  youtube_df\n",
              "3462                                                ! !         0  youtube_df\n",
              "3463  great video and MERRY CHRISTMAS from greece :*...         0  youtube_df\n",
              "\n",
              "[3464 rows x 3 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 36
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "77oHbprujRxQ"
      },
      "source": [
        "## Merging all Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xvFDwH7iimZ6"
      },
      "source": [
        "final_df = pd.concat([kaggle_df, twitter_df, aggression_df, attack_df, toxicity_df, youtube_df])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bictrMCzi0wC",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "5d32b137-4e73-41c6-b711-760269cb2df6"
      },
      "source": [
        "final_df.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(420528, 3)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 38
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "id": "iWdZrjWH9SUf",
        "outputId": "05b0b5b9-2b43-4e36-aee4-48a4e639cd4b"
      },
      "source": [
        "final_df.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>oh_label</th>\n",
              "      <th>Text</th>\n",
              "      <th>Source</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1.0</td>\n",
              "      <td>\"You fuck your dad.\"</td>\n",
              "      <td>kaggle_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0.0</td>\n",
              "      <td>\"i really don't understand your point.\\xa0 It ...</td>\n",
              "      <td>kaggle_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0.0</td>\n",
              "      <td>\"A\\\\xc2\\\\xa0majority of Canadians can and has ...</td>\n",
              "      <td>kaggle_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0.0</td>\n",
              "      <td>\"listen if you dont wanna get married to a man...</td>\n",
              "      <td>kaggle_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0.0</td>\n",
              "      <td>\"C\\xe1c b\\u1ea1n xu\\u1ed1ng \\u0111\\u01b0\\u1edd...</td>\n",
              "      <td>kaggle_df</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   oh_label                                               Text     Source\n",
              "0       1.0                               \"You fuck your dad.\"  kaggle_df\n",
              "1       0.0  \"i really don't understand your point.\\xa0 It ...  kaggle_df\n",
              "2       0.0  \"A\\\\xc2\\\\xa0majority of Canadians can and has ...  kaggle_df\n",
              "3       0.0  \"listen if you dont wanna get married to a man...  kaggle_df\n",
              "4       0.0  \"C\\xe1c b\\u1ea1n xu\\u1ed1ng \\u0111\\u01b0\\u1edd...  kaggle_df"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 39
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vtpxoCy3j368"
      },
      "source": [
        "## EDA"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mmsigKD2j7hL",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "43c89e8b-65b5-466a-ce00-88ab08c15635"
      },
      "source": [
        "final_df.isna().sum()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "oh_label    3\n",
              "Text        1\n",
              "Source      0\n",
              "dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 40
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bM91iWDpj_3c"
      },
      "source": [
        "master_df = final_df.dropna()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "z7omhh4dkf6o",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "d6091769-9183-42f2-c788-3fb3bbfe1186"
      },
      "source": [
        "master_df.isna().sum()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "oh_label    0\n",
              "Text        0\n",
              "Source      0\n",
              "dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 42
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "l0vp-YZhkxrs",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 564
        },
        "outputId": "482c099f-3d17-4ce7-8e0a-e5a81ada5278"
      },
      "source": [
        "cyberbully = master_df[master_df.oh_label==1.0].count()[0]\n",
        "nocyberbully = master_df[master_df.oh_label==0.0].count()[0]\n",
        "value = [nocyberbully, cyberbully]\n",
        "labels = [ \"no-cyberbully\", \"cyberbully\"]\n",
        "plt.figure(figsize=(15,8))\n",
        "plt.bar(np.arange(len(value)),value);\n",
        "plt.title(\"Comparison between number of cyberbully and nocyberbully text detected\")\n",
        "plt.xlabel(\"Labels\")\n",
        "plt.ylabel(\"Count\")\n",
        "plt.show()\n",
        "print(value)\n",
        "print(cyberbully)\n",
        "print(nocyberbully)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1080x576 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "[368221, 52304]\n",
            "52304\n",
            "368221\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HD7NLq-ImnXS",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "c0040f64-517e-4dbf-c667-6817aa82fd15"
      },
      "source": [
        "n = len(pd.unique(master_df['Text']))\n",
        "print(\"Unique -\" , n)\n",
        "print(\"Total -\", master_df.shape[0])\n",
        "print(\"Duplicates\", master_df.shape[0] - n)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Unique - 225346\n",
            "Total - 420525\n",
            "Duplicates 195179\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ozZ5nbtnohe1"
      },
      "source": [
        "nondup_df = master_df.drop_duplicates(subset='Text', keep=\"first\")\n",
        "nondup_df = nondup_df[~(nondup_df['Text'].duplicated(keep=False))]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LUKzGbR3owsm",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "2269327f-f75b-4c75-a377-087481f21321"
      },
      "source": [
        "nondup_df"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>index</th>\n",
              "      <th>oh_label</th>\n",
              "      <th>Date</th>\n",
              "      <th>Text</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>20120618192155Z</td>\n",
              "      <td>\"You fuck your dad.\"</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>20120528192215Z</td>\n",
              "      <td>\"i really don't understand your point.\\xa0 It ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>4.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>20120619094753Z</td>\n",
              "      <td>\"C\\xe1c b\\u1ea1n xu\\u1ed1ng \\u0111\\u01b0\\u1edd...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>5.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>20120620171226Z</td>\n",
              "      <td>\"@SDL OK, but I would hope they'd sign him to ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>6.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>20120503012628Z</td>\n",
              "      <td>\"Yeah and where are you now?\"</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8794</th>\n",
              "      <td>8794.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>20120612003508Z</td>\n",
              "      <td>\"Never really gave it much thought. I just fig...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8795</th>\n",
              "      <td>8795.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>20120619210456Z</td>\n",
              "      <td>\"Nadie se salva de la regla 34 xd\"</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8796</th>\n",
              "      <td>8796.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>20120528234613Z</td>\n",
              "      <td>\"Question: Are you a boy or a girl?\"</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8797</th>\n",
              "      <td>8797.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>20120619153537Z</td>\n",
              "      <td>\"Leave your email or phone number and maybe yo...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8798</th>\n",
              "      <td>8798.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>20120620000237Z</td>\n",
              "      <td>\"From the scenarios you present, I see you bel...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>7503 rows × 4 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "       index  ...                                               Text\n",
              "0        0.0  ...                               \"You fuck your dad.\"\n",
              "1        1.0  ...  \"i really don't understand your point.\\xa0 It ...\n",
              "4        4.0  ...  \"C\\xe1c b\\u1ea1n xu\\u1ed1ng \\u0111\\u01b0\\u1edd...\n",
              "5        5.0  ...  \"@SDL OK, but I would hope they'd sign him to ...\n",
              "6        6.0  ...                      \"Yeah and where are you now?\"\n",
              "...      ...  ...                                                ...\n",
              "8794  8794.0  ...  \"Never really gave it much thought. I just fig...\n",
              "8795  8795.0  ...                 \"Nadie se salva de la regla 34 xd\"\n",
              "8796  8796.0  ...               \"Question: Are you a boy or a girl?\"\n",
              "8797  8797.0  ...  \"Leave your email or phone number and maybe yo...\n",
              "8798  8798.0  ...  \"From the scenarios you present, I see you bel...\n",
              "\n",
              "[7503 rows x 4 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 47
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GR9CqKsQAxaY"
      },
      "source": [
        "# Text Cleaning and Contraction Mapping"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SgpPeIJ7A1I8"
      },
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn import metrics\n",
        "import re"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SFRRqqaHA8ld"
      },
      "source": [
        "final_df = pd.read_csv(\"./Final_merged_file.csv\", index_col=None)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MXiv_yKhBIMr",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        },
        "outputId": "e9b4456b-ca0f-4f29-c2f2-dee7b91c34a5"
      },
      "source": [
        "final_df"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Unnamed: 0</th>\n",
              "      <th>Text</th>\n",
              "      <th>label</th>\n",
              "      <th>Source</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0</td>\n",
              "      <td>@switchfoot http://twitpic.com/2y1zl - Awww, t...</td>\n",
              "      <td>1</td>\n",
              "      <td>k_suspicious_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1</td>\n",
              "      <td>is upset that he can't update his Facebook by ...</td>\n",
              "      <td>1</td>\n",
              "      <td>k_suspicious_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>2</td>\n",
              "      <td>@Kenichan I dived many times for the ball. Man...</td>\n",
              "      <td>1</td>\n",
              "      <td>k_suspicious_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>3</td>\n",
              "      <td>my whole body feels itchy and like its on fire</td>\n",
              "      <td>0</td>\n",
              "      <td>k_suspicious_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>4</td>\n",
              "      <td>@nationwideclass no, it's not behaving at all....</td>\n",
              "      <td>1</td>\n",
              "      <td>k_suspicious_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>278587</th>\n",
              "      <td>480523</td>\n",
              "      <td>`   These sources don't exactly exude a sense ...</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>278588</th>\n",
              "      <td>480524</td>\n",
              "      <td>The Institute for Historical Review is a pee...</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>278589</th>\n",
              "      <td>480525</td>\n",
              "      <td>:The way you're trying to describe it in this...</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>278590</th>\n",
              "      <td>480526</td>\n",
              "      <td>== Warning ==  There is clearly a protection...</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>278591</th>\n",
              "      <td>480527</td>\n",
              "      <td>Alternate option=== Is there perhaps enough ne...</td>\n",
              "      <td>0</td>\n",
              "      <td>toxicity_df</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>278592 rows × 4 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "        Unnamed: 0  ...           Source\n",
              "0                0  ...  k_suspicious_df\n",
              "1                1  ...  k_suspicious_df\n",
              "2                2  ...  k_suspicious_df\n",
              "3                3  ...  k_suspicious_df\n",
              "4                4  ...  k_suspicious_df\n",
              "...            ...  ...              ...\n",
              "278587      480523  ...      toxicity_df\n",
              "278588      480524  ...      toxicity_df\n",
              "278589      480525  ...      toxicity_df\n",
              "278590      480526  ...      toxicity_df\n",
              "278591      480527  ...      toxicity_df\n",
              "\n",
              "[278592 rows x 4 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6EF-7Y6QBJ3B"
      },
      "source": [
        "final_df = final_df.drop(['Source', 'Unnamed: 0'], axis=1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cgd36psKCB34",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        },
        "outputId": "4437f851-c3f2-4351-e19f-20995bb8aede"
      },
      "source": [
        "final_df"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Text</th>\n",
              "      <th>label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>@switchfoot http://twitpic.com/2y1zl - Awww, t...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>is upset that he can't update his Facebook by ...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>@Kenichan I dived many times for the ball. Man...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>my whole body feels itchy and like its on fire</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>@nationwideclass no, it's not behaving at all....</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>278587</th>\n",
              "      <td>`   These sources don't exactly exude a sense ...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>278588</th>\n",
              "      <td>The Institute for Historical Review is a pee...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>278589</th>\n",
              "      <td>:The way you're trying to describe it in this...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>278590</th>\n",
              "      <td>== Warning ==  There is clearly a protection...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>278591</th>\n",
              "      <td>Alternate option=== Is there perhaps enough ne...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>278592 rows × 2 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                     Text  label\n",
              "0       @switchfoot http://twitpic.com/2y1zl - Awww, t...      1\n",
              "1       is upset that he can't update his Facebook by ...      1\n",
              "2       @Kenichan I dived many times for the ball. Man...      1\n",
              "3         my whole body feels itchy and like its on fire       0\n",
              "4       @nationwideclass no, it's not behaving at all....      1\n",
              "...                                                   ...    ...\n",
              "278587  `   These sources don't exactly exude a sense ...      0\n",
              "278588    The Institute for Historical Review is a pee...      0\n",
              "278589   :The way you're trying to describe it in this...      0\n",
              "278590    == Warning ==  There is clearly a protection...      0\n",
              "278591  Alternate option=== Is there perhaps enough ne...      0\n",
              "\n",
              "[278592 rows x 2 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dxlRUfRwNk8H"
      },
      "source": [
        "## Contraction Mapping"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ztSao7fyJgmu"
      },
      "source": [
        "CONTRACTION_MAP = {\n",
        "\"ain't\": \"is not\",\n",
        "\"aren't\": \"are not\",\n",
        "\"can't\": \"cannot\",\n",
        "\"can't've\": \"cannot have\",\n",
        "\"'cause\": \"because\",\n",
        "\"could've\": \"could have\",\n",
        "\"couldn't\": \"could not\",\n",
        "\"couldn't've\": \"could not have\",\n",
        "\"didn't\": \"did not\",\n",
        "\"doesn't\": \"does not\",\n",
        "\"don't\": \"do not\",\n",
        "\"hadn't\": \"had not\",\n",
        "\"hadn't've\": \"had not have\",\n",
        "\"hasn't\": \"has not\",\n",
        "\"haven't\": \"have not\",\n",
        "\"he'd\": \"he would\",\n",
        "\"he'd've\": \"he would have\",\n",
        "\"he'll\": \"he will\",\n",
        "\"he'll've\": \"he he will have\",\n",
        "\"he's\": \"he is\",\n",
        "\"how'd\": \"how did\",\n",
        "\"how'd'y\": \"how do you\",\n",
        "\"how'll\": \"how will\",\n",
        "\"how's\": \"how is\",\n",
        "\"I'd\": \"I would\",\n",
        "\"I'd've\": \"I would have\",\n",
        "\"I'll\": \"I will\",\n",
        "\"I'll've\": \"I will have\",\n",
        "\"I'm\": \"I am\",\n",
        "\"I've\": \"I have\",\n",
        "\"i'd\": \"i would\",\n",
        "\"i'd've\": \"i would have\",\n",
        "\"i'll\": \"i will\",\n",
        "\"i'll've\": \"i will have\",\n",
        "\"i'm\": \"i am\",\n",
        "\"i've\": \"i have\",\n",
        "\"isn't\": \"is not\",\n",
        "\"it'd\": \"it would\",\n",
        "\"it'd've\": \"it would have\",\n",
        "\"it'll\": \"it will\",\n",
        "\"it'll've\": \"it will have\",\n",
        "\"it's\": \"it is\",\n",
        "\"let's\": \"let us\",\n",
        "\"ma'am\": \"madam\",\n",
        "\"mayn't\": \"may not\",\n",
        "\"might've\": \"might have\",\n",
        "\"mightn't\": \"might not\",\n",
        "\"mightn't've\": \"might not have\",\n",
        "\"must've\": \"must have\",\n",
        "\"mustn't\": \"must not\",\n",
        "\"mustn't've\": \"must not have\",\n",
        "\"needn't\": \"need not\",\n",
        "\"needn't've\": \"need not have\",\n",
        "\"o'clock\": \"of the clock\",\n",
        "\"oughtn't\": \"ought not\",\n",
        "\"oughtn't've\": \"ought not have\",\n",
        "\"shan't\": \"shall not\",\n",
        "\"sha'n't\": \"shall not\",\n",
        "\"shan't've\": \"shall not have\",\n",
        "\"she'd\": \"she would\",\n",
        "\"she'd've\": \"she would have\",\n",
        "\"she'll\": \"she will\",\n",
        "\"she'll've\": \"she will have\",\n",
        "\"she's\": \"she is\",\n",
        "\"should've\": \"should have\",\n",
        "\"shouldn't\": \"should not\",\n",
        "\"shouldn't've\": \"should not have\",\n",
        "\"so've\": \"so have\",\n",
        "\"so's\": \"so as\",\n",
        "\"that'd\": \"that would\",\n",
        "\"that'd've\": \"that would have\",\n",
        "\"that's\": \"that is\",\n",
        "\"there'd\": \"there would\",\n",
        "\"there'd've\": \"there would have\",\n",
        "\"there's\": \"there is\",\n",
        "\"they'd\": \"they would\",\n",
        "\"they'd've\": \"they would have\",\n",
        "\"they'll\": \"they will\",\n",
        "\"they'll've\": \"they will have\",\n",
        "\"they're\": \"they are\",\n",
        "\"they've\": \"they have\",\n",
        "\"to've\": \"to have\",\n",
        "\"wasn't\": \"was not\",\n",
        "\"we'd\": \"we would\",\n",
        "\"we'd've\": \"we would have\",\n",
        "\"we'll\": \"we will\",\n",
        "\"we'll've\": \"we will have\",\n",
        "\"we're\": \"we are\",\n",
        "\"we've\": \"we have\",\n",
        "\"weren't\": \"were not\",\n",
        "\"what'll\": \"what will\",\n",
        "\"what'll've\": \"what will have\",\n",
        "\"what're\": \"what are\",\n",
        "\"what's\": \"what is\",\n",
        "\"what've\": \"what have\",\n",
        "\"when's\": \"when is\",\n",
        "\"when've\": \"when have\",\n",
        "\"where'd\": \"where did\",\n",
        "\"where's\": \"where is\",\n",
        "\"where've\": \"where have\",\n",
        "\"who'll\": \"who will\",\n",
        "\"who'll've\": \"who will have\",\n",
        "\"who's\": \"who is\",\n",
        "\"who've\": \"who have\",\n",
        "\"why's\": \"why is\",\n",
        "\"why've\": \"why have\",\n",
        "\"will've\": \"will have\",\n",
        "\"won't\": \"will not\",\n",
        "\"won't've\": \"will not have\",\n",
        "\"would've\": \"would have\",\n",
        "\"wouldn't\": \"would not\",\n",
        "\"wouldn't've\": \"would not have\",\n",
        "\"y'all\": \"you all\",\n",
        "\"y'all'd\": \"you all would\",\n",
        "\"y'all'd've\": \"you all would have\",\n",
        "\"y'all're\": \"you all are\",\n",
        "\"y'all've\": \"you all have\",\n",
        "\"you'd\": \"you would\",\n",
        "\"you'd've\": \"you would have\",\n",
        "\"you'll\": \"you will\",\n",
        "\"you'll've\": \"you will have\",\n",
        "\"you're\": \"you are\",\n",
        "\"you've\": \"you have\"\n",
        "}\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xfkcZH9DJhov"
      },
      "source": [
        "def expand_contractions(text, contraction_mapping=CONTRACTION_MAP):\n",
        "    text = text.split() # to convert have'nt -> have not\n",
        "    for i in range(len(text)):\n",
        "        word = text[i]\n",
        "        if word in contraction_mapping:\n",
        "            text[i] = contraction_mapping[word]\n",
        "    text = \" \".join(text)\n",
        "    #text = text.split()\n",
        "    #newtext = []\n",
        "    #for word in text:\n",
        "    #    if word not in stop_words:\n",
        "    #        newtext.append(word)\n",
        "    #text = \" \".join(newtext)\n",
        "    text = text.replace(\"'s\",'') # to convert your's -> your\n",
        "    #text = re.sub(r'\\(.*\\)','',text) # remove (words)\n",
        "    #text = re.sub(r'[^a-zA-Z0-9. ]','',text) # remove punctuations\n",
        "    #text = re.sub(r'\\.',' . ',text)\n",
        "    return text"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CAKnmLoDL3EK",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "70b79ad0-76fd-47e9-d0e8-b522387f88fe"
      },
      "source": [
        "new_list_1 = []\n",
        "# f = re.sub(r'[^a-zA-Z0-9. ]','',merged_df_2['Text'])\n",
        "for text in final_df['Text']:\n",
        "    new_list_1.append(expand_contractions(re.sub(r'[^a-zA-Z0-9. ]','',text)))\n",
        "new_list_1"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['switchfoot httptwitpic.com2y1zl Awww thats a bummer. You shoulda got David Carr of Third Day to do it. D',\n",
              " 'is upset that he cant update his Facebook by texting it... and might cry as a result School today also. Blah',\n",
              " 'Kenichan I dived many times for the ball. Managed to save 50 The rest go out of bounds',\n",
              " 'my whole body feels itchy and like its on fire',\n",
              " 'nationwideclass no its not behaving at all. im mad. why am i here because I cant see you all over there.',\n",
              " 'Kwesidei not the whole crew',\n",
              " 'Need a hug',\n",
              " 'LOLTrish hey long time no see Yes.. Rains a bit only a bit LOL Im fine thanks hows you',\n",
              " 'TatianaK nope they didnt have it',\n",
              " 'twittera que me muera',\n",
              " 'spring break in plain city... its snowing',\n",
              " 'I just repierced my ears',\n",
              " 'caregiving I couldnt bear to watch it. And I thought the UA loss was embarrassing . . . . .',\n",
              " 'octolinz16 It it counts idk why I did either. you never talk to me anymore',\n",
              " 'smarrison i wouldve been the first but i didnt have a gun. not really though zac snyders just a doucheclown.',\n",
              " 'iamjazzyfizzle I wish I got to watch it with you I miss you and iamlilnicki how was the premiere',\n",
              " 'Hollis death scene will hurt me severely to watch on film wry is directors cut not out now',\n",
              " 'about to file taxes',\n",
              " 'LettyA ahh ive always wanted to see rent love the soundtrack',\n",
              " 'FakerPattyPattz Oh dear. Were you drinking out of the forgotten table drinks',\n",
              " 'alydesigns i was out most of the day so didnt get much done',\n",
              " 'one of my friend called me and asked to meet with her at Mid Valley today...but ive no time sigh',\n",
              " 'angrybarista I baked you a cake but I ated it',\n",
              " 'this week is not going as i had hoped',\n",
              " 'blagh class at 8 tomorrow',\n",
              " 'I hate when I have to call and wake people up',\n",
              " 'Just going to cry myself to sleep after watching Marley and Me.',\n",
              " 'im sad now Miss.Lilly',\n",
              " 'ooooh.... LOL that leslie.... and ok I wont do it again so leslie wont get mad again',\n",
              " 'Meh... Almost Lover is the exception... this track gets me depressed every time.',\n",
              " 'some1 hacked my account on aim now i have to make a new one',\n",
              " 'alielayus I want to go to promote GEAR AND GROOVE but unfornately no ride there I may b going to the one in Anaheim in May though',\n",
              " 'thought sleeping in was an option tomorrow but realizing that it now is not. evaluations in the morning and work in the afternoon',\n",
              " 'julieebaby awe i love you too 1 am here i miss you',\n",
              " 'HumpNinja I cry my asian eyes to sleep at night',\n",
              " 'ok Im sick and spent an hour sitting in the shower cause I was too sick to stand and held back the puke like a champ. BED now',\n",
              " 'cocomix04 ill tell ya the story later not a good day and ill be workin for like three more hours...',\n",
              " 'MissXu sorry bed time came here GMT1 httpis.gdfNge',\n",
              " 'fleurylis I dont either. Its depressing. I dont think I even want to know about the kids in suitcases.',\n",
              " 'Bed. Class 812. Work 123. Gym 35 or 6. Then class 610. Another day thats gonna fly by. I miss my girlfriend',\n",
              " 'really dont feel like getting up today... but got to study to for tomorrows practical exam...',\n",
              " 'Hes the reason for the teardrops on my guitar the only one who has enough of me to break my heart',\n",
              " 'Sad sad sad. I dont know why but I hate this feeling I wanna sleep and I still cant',\n",
              " 'JonathanRKnight Awww I soo wish I was there to see you finally comfortable Im sad that I missed it',\n",
              " 'Falling asleep. Just heard about that Tracy girls body being found. How sad My heart breaks for that family.',\n",
              " 'Viennah Yay Im happy for you with your job But that also means less time for me and you...',\n",
              " 'Just checked my user timeline on my blackberry it looks like the twanking is still happening Are ppl still having probs w BGs and UIDs',\n",
              " 'Oh man...was ironing jeancjumbes fave top to wear to a meeting. Burnt it',\n",
              " 'is strangely sad about LiLo and SamRo breaking up.',\n",
              " 'tea oh im so sorry i didnt think about that before retweeting.',\n",
              " 'Broadband plan a massive broken promise httptinyurl.comdcuc33 via www.diigo.comtautao Still waiting for broadband we are',\n",
              " 'localtweeps Wow tons of replies from you may have to unfollow so I can see my friends tweets youre scrolling the feed a lot.',\n",
              " 'our duck and chicken are taking wayyy too long to hatch',\n",
              " 'I need a hug',\n",
              " 'andywana Not sure what they are only that they are PoS As much as I want to I dont think can trade away company assets sorry andy',\n",
              " 'Put vacation photos online a few yrs ago. PC crashed and now I forget the name of the site.',\n",
              " 'I have a sad feeling that Dallas is not going to show up I gotta say though youd think more shows would use music from the game. mmm',\n",
              " 'Ugh....92 degrees tomorrow',\n",
              " 'Where did u move to I thought u were already in sd. Hmmm. Random u found me. Glad to hear yer doing well.',\n",
              " 'BatManYNG I miss my ps3 its out of commission Wutcha playing Have you copped Blood On The Sand',\n",
              " 'just leaving the parking lot of work',\n",
              " 'The Life is cool. But not for Me.',\n",
              " 'Sadly though Ive never gotten to experience the post coitus cigarette before and now I never will.',\n",
              " 'I had such a nice day. Too bad the rain comes in tomorrow at 5am',\n",
              " 'Starrbby too bad I wont be around I lost my job and cant even pay my phone bill lmao aw shucks',\n",
              " 'Damm back to school tomorrow',\n",
              " 'Mo jobs no money. how in the hell is min wage here 4 fn clams an hour',\n",
              " 'katortiz Not forever... See you soon',\n",
              " 'LtAlgonquin agreed I saw the failwhale allllll day today.',\n",
              " 'jdarter Oh Haha... dude I dont really look at em unless someone says HEY I ADDED YOU. Sorry Im so terrible at that. I need a pop up',\n",
              " 'ninjen Im sure youre right... I need to start working out with you and the Nikster... Or Jared at least',\n",
              " 'i really hate how people diss my bands Trace is clearly NOT ugly',\n",
              " 'Gym attire today was Puma singlet Adidas shorts.......and black business socks and leather shoes Lucky did not run into any cute girls.',\n",
              " 'Why wont you show my location httptwitpic.com2y2es',\n",
              " 'No picnic my phone smells like citrus.',\n",
              " 'ashleyac My donkey is sensitive about such comments. Nevertheless hed and med be glad to see your mug asap. Charger is still awol.',\n",
              " 'No new csi tonight. FML',\n",
              " 'i think my arms are sore from tennis',\n",
              " 'wonders why someone that u like so much can make you so unhappy in a split seccond . depressed .',\n",
              " 'sleep soon... i just hate saying bye and see you tomorrow for the night.',\n",
              " 'statravelAU just got ur newsletter those fares really are unbelievable shame I already booked and paid for mine',\n",
              " 'missin the boo',\n",
              " 'markhardy1974 Me too itm',\n",
              " 'Damn... I dont have any chalk MY CHALKBOARD IS USELESS',\n",
              " 'had a blast at the Getty Villa but hates that shes had a sore throat all day. Its just getting worse too',\n",
              " 'msdrama hey missed ya at the meeting sup mama',\n",
              " 'My tummy hurts. I wonder if the hypnosis has anything to do with it If so its working I get it STOP SMOKING',\n",
              " 'why is it always the fat ones',\n",
              " 'januarycrimson Sorry babe My fam annoys me too. Thankfully theyre asleep right now. Muahaha. evil laugh',\n",
              " 'Hollywoodheat I should have paid more attention when we covered photoshop in my webpage design class in undergrad',\n",
              " 'wednesday my bday dont know what 2 do',\n",
              " 'Poor cameron the hills',\n",
              " 'pray for me please the ex is threatening to start sh at myour babies 1st Birthday party. what a jerk. and I still have a headache',\n",
              " 'makeherfamous hmm do u really enjoy being with him if the problems are too constants u should think things more find someone ulike',\n",
              " 'Strider is a sick little puppy httpapps.facebook.comdogbookprofileview5248435',\n",
              " 'so ryleegrace...wana go steves party or not SADLY SINCE ITS EASTER I WNT B ABLE 2 DO MUCH BUT OHH WELL.....',\n",
              " 'hey I actually won one of my bracket pools Too bad it wasnt the one for money',\n",
              " 'stark YOU dont follow me either and i work for you',\n",
              " 'Behind on my classes for work',\n",
              " 'watching quotHousequot',\n",
              " 'kpreyes Remember my bum leg Strikes back this time its serious',\n",
              " 'mangaaa I hope they will increase the capacity fast yesterday was such a pain. Got the fail whale 15 times in 2 hours....',\n",
              " 'would rather the first party send bad messages than the 3rd party send mixed ones Sophmore year all over again',\n",
              " 'Henkuyinepu its overrated',\n",
              " 'marykatherineq i know I heard it this afternoon and wondered the same thing. Moscow is so behind the times',\n",
              " 'laying in bed with no voice..',\n",
              " 'Im sooo sad they killed off Kutner on House whyyyyyyyy',\n",
              " 'jacobsummers Sorry tell them mea culpa from me and that I really am sorry.',\n",
              " 'Alliana07 it didnt make any sense to me the suicide thing. I refuse to believe that that is actually what happened.',\n",
              " 'salancaster hope your ok.',\n",
              " 'mercedesashley Damn The grind is inspirational and saddening at the same time. Dont want you to stop cuz I like what u do Much love',\n",
              " 'HibaNick yeah aw but i know i wudnt stand a chance',\n",
              " 'ugh. cant sleep. its 130am.',\n",
              " 'Hanging in Crooners. Wanna sing. Cant. Sucks.',\n",
              " 'eRResC aaw i miss ya all too.. im leaving to BH tomorrow quotmorningquot i think.. aww i wanna go to the beach w u girls',\n",
              " 'Is pissed off that theres no ASBAs for a radio station.',\n",
              " 'wednesday my bday n dont know what 2 do',\n",
              " 'I know my life has been flipped upside down when I just thought in my head that some Ramen sounds good.',\n",
              " 'I am in pain. My back and sides hurt. Not to mention crying is made of fail.',\n",
              " 'Late night snack glass of OJ bc Im quotdown with the sicknessquot then back to sleep...ugh I hate getting sick...',\n",
              " 'allyheman but.. but.. but.. Im not a big fan on camilla belle',\n",
              " 'grum WAH I cant see clip must be elstupido work filters. Cant wait till I get a puter. Something else 2 blame ex 4. He broke mine',\n",
              " 'This week just seems to get longer and longer in terms of how much I need to do.. and how much Im actually going to get done',\n",
              " 'Im so cold',\n",
              " 'thecoolestout Ehhh dont. Weathers gonna take a turn for the ugly tomorrow',\n",
              " 'chelserlynn haha its so cooooold in the d and no but you should still go to the show. they do some incredible stuff',\n",
              " 'Hoping the tummy rumbles go away soon',\n",
              " 'Knights No no notice. They told me Id be working tomorrow and then I called the agency to follow up and they said it was over.',\n",
              " 'almost bedtime',\n",
              " 'Im missing you babe.. but as long as your alive Im happy.. Yawwwnn.. Im tired my love imma try to sleep hopefully you had a headstart',\n",
              " 'Agh...snow',\n",
              " 'i miss kenny powers',\n",
              " 'BridgetsBeaches Thank you for letting people know but now Im sad that the direct message I got wasnt actually from Bridget',\n",
              " '...and India missed out its 100th Test victory n 10th consecutive win without a loss.',\n",
              " 'JonathanRKnight I guess thats a no then.',\n",
              " 'sadly is going to bed.',\n",
              " 'ozesteph1992 Shame to hear this Stephan',\n",
              " 'mrsaintnick hey im leavin in the morning...',\n",
              " 'Was intending to finish editing my 536page novel manuscript tonight but that will probably not happen. And only 12 pages are left',\n",
              " 'laid around too much today... now my head hurts',\n",
              " 'twista202 I still havent read the 9thamp10th Princess diaries Saving Francesca made me cry at the end. Hmm those are easy books.',\n",
              " 'my nokia 1110 died..',\n",
              " 'My mom might have breast cancer wont find out anything for. Like a week Im so worried',\n",
              " 'Going to sleep. Hoping tomorrow is a better day.',\n",
              " 'rumblepurr lol.. wish they understood daylight savings has ended though and breakfast is an hour later They keep waking the kids up too',\n",
              " 'onemoreproject that is lame',\n",
              " 'I dont understand... I really dont',\n",
              " 'HEROES just isnt doing it for me this season...',\n",
              " 'Living not downtown sure isnt much fun.',\n",
              " 'jonathanchard Not calorie wise I wish junk food was calorie free I ate a thing of sour skittles and a big ass cherry coke',\n",
              " 'Man Work is Hard',\n",
              " 'getting sick time for some hot tea studying and then sleeeep',\n",
              " 'Getting eyebrows waxed. More pain',\n",
              " 'No phantasy star yesterday going to work...',\n",
              " 'Oh Just got all my MacHeist 3.0 apps sweet. Didnt get the Espresso serial no though although they said they sent it oh well',\n",
              " 'Picked Mich St to win it all from the get go. Was feeling pretty good about that pick all the way up until......tonight. As lost too',\n",
              " 'is alone downstairs...working',\n",
              " 'I feel bad for doing it',\n",
              " 'RyanSeacrest is it just me or she hates anoop. i mean seriously shes kinda mean to him.',\n",
              " 'pinkserendipity yes sprint has 4g only in baltimore and chicago so far',\n",
              " 'Im stuck awake in the middle of the night for the second day in a row. And I felt terrible yesterday.',\n",
              " 'Thanks for bursting my bubble',\n",
              " 'Going to school soon... cant find anything to wear gosh its so hard',\n",
              " 'marieclr I was serious LOL',\n",
              " 'naughtyhaughty I had on my page for sooooo long Until it got deleted Sad day in history',\n",
              " 'Crazy wind today no birding httpff.im1XTTi',\n",
              " 'currently at work..',\n",
              " 'Grrr.. my ipods acting weird too Jai ho and thinking of you arent playing the full songs.. ughh.',\n",
              " 'penndbad send me the DVD cos I have missed out on heaps NOT happy about that',\n",
              " 'i dont see the big deal with this website',\n",
              " 'machineplay Im so sorry youre having to go through this. Again. therapyfail',\n",
              " 'ColinDeMar Far too out of the way for rail. any other tips',\n",
              " 'Im not still up I swear. Why do I keep losing. gaining. losing. gaining. tweeps So heart wrenching.',\n",
              " 'Today I realized I am too good at hiding things even I cant find it',\n",
              " 'Staying at a friends house...house sitting neighbors are SO loudhaving a party',\n",
              " 'dannyvegasbaby danny im upset that i wasnt here to watch the live chat i was in a car. for 3 hours on a trip im soooo upset',\n",
              " 'Check out my mug httpwww.erikaobscura.blogspot.com',\n",
              " 'Borders closed at 10',\n",
              " 'Downloading NINs new album quotthe slipquot when the hell did this come out Im so behind the times these days',\n",
              " 'Just woke up an already have written some email... Ive to go early at University today as I have to teach at 830 AM',\n",
              " 'is watching the hill . . .and its making me sad',\n",
              " 'So many channels.... yet so so boring... lazy day again may have to find a hobby',\n",
              " 'supersport I miss my buddy ill be in NY on the 25th',\n",
              " 'robluketic love the french. I tell people here in the south im qtr. french and they snarl at me. french are beautiful people',\n",
              " 'Opps.. As I said.. I still got one day remain and now problem come',\n",
              " 'I activated my Selfcontrol block early meaning I cant check out the new QC. Regularizing my internal clock is might be difficult. fb',\n",
              " 'HillyDoP oH NO',\n",
              " 'spencer is not a good guy.',\n",
              " 'goodlaura What about Reese dying on TTSC And season finale next week. 24 boring Madame President is a crazy woman.',\n",
              " 'JonathanRKnight I hate the limited letterstoo.Hope you and the guys are fineI pray for my dogshes not well',\n",
              " 'didnt get shit done today im so screwed',\n",
              " 'wanttss to go out',\n",
              " 'Is not going to sleep tonite.',\n",
              " 'too worried and tired to post tonight',\n",
              " 'couldnt get shit done today im so screwed',\n",
              " 'Job Interview in Cardiff today wish me luck Got about 3 hours sleep',\n",
              " 'eloquentembrace Youre going to kill me but Ive not seen DS9. Ive been waiting till I can do it in one solid week sitting.',\n",
              " 'cheechbud i think ur right hahaha 4.5 hrs now',\n",
              " 'i hate to see the spartans so sad',\n",
              " 'My mind and body are severely protesting this quotgetting upquot thing. Had nightmares to boot',\n",
              " 'MySteezRadio Im goin to follow u since u didnt LOL GO ANGELS',\n",
              " 'mandayyy',\n",
              " 'twista202 I think I want to read some books but the library doesnt have them',\n",
              " 'My nap was interrupted so many times today Going out for Japanese with the rents again...',\n",
              " 'Kind of longs for the bus that shows up at the end of Ghost World right now. Ugh.',\n",
              " 'GuruMN but this is canada canada is weird. were supposed to get snow through wednesday. ugh.',\n",
              " 'sad that the feet of my macbook just fell off',\n",
              " 'Im gonna get up late tomorrow and its 132am here. I gonna get tipsy by my lonesome. Thats...thats just sad',\n",
              " 'Im sweating my forthcoming trip to E3. If I cant find someone to crash with while Im out there I may be screwed.',\n",
              " 'Has now gotten somebody to read his tweets but cant get them to make an account.',\n",
              " 'infinitydefines omgawd i couldnt handle my cat being in heat all the time D D',\n",
              " 'DonnieWahlberg I hope i can make it to the auburn show. but its not looking good for me',\n",
              " 'DavidHenrie thats people mag haha i couldnt fit it all in.. i dont think those pictures ever made it in the magazine tho haha',\n",
              " 'breadandbadger Congrats i totally forgot to submit photos',\n",
              " 'PaulaAbdul awww Good luck Paula Please dont work too hard but I hope you have fun Your new album is gonna be amazing xxx',\n",
              " 'imaginarypeach now your leaving me .... gets sad',\n",
              " 'I miss you twitter. My phone broke now Im using a stupid Nokia phone. Ughhh I miss my advance phone.',\n",
              " 'SHOOTING OUTSIDE MY HOUSE O NOT KIDDING So SCARED',\n",
              " 'Tuesdayll start with reflection n then a lecture in Stress reducing techniques. That sure might become very useful for us accompaniers',\n",
              " 'What tragedy and disaster in the news this week',\n",
              " 'yes yes still trying to find a picture that will upload correclty..',\n",
              " 'Why oh why was the Red Sox game rained out I was so looking forward to opening day',\n",
              " 'I still cant find my keys.',\n",
              " 'charlietm I know right. I dunno what is going on with twitter.',\n",
              " 'might be getting a sore throat again',\n",
              " 'labelsnotlove my home town. My mammy called all depressd. Pls explain y a parent let their 8yr old child walk alone Hello Its 2009',\n",
              " 'ircmaxell I think I need to find better antidepressants. I think this PaxilWellbutrin combo is losing its efficacy.',\n",
              " 'MyConnecticut Restaurant called WoodNTap has competitive eating tourney round tourney time. We place 2nd',\n",
              " 'is in the bathroom... wake up lakin.',\n",
              " 'CaitlinOConnor i want tacos and margarhitas telll gay i say hellolt3',\n",
              " 'bad day at the betfair office',\n",
              " 'I miss him. Cant wait to celebrate the Tar Heel win this weekend though',\n",
              " 'Im really cold. I dont want to go to sleep yet but theres nothing to do',\n",
              " 'kristencampisi is this it 4 u Its officially over 4 me this go round',\n",
              " 'monkeys i just found out you my twin and you wont even write back. im heartbroken',\n",
              " 'missom Aww I know I felt like that yesterday at work',\n",
              " 'SarahReedSC treaty isnt defined',\n",
              " 'Missing My BFF watching home and away it reminds me of her and me we lt3 It. shout out to u courts',\n",
              " 'mandayyy',\n",
              " 'New video card is DOA.',\n",
              " 'feeling lost naked and confused JK sort of.....no iphone for me.',\n",
              " 'Damn i am so late at filling this appraisal form people have almost sent it .. i was so occupied in work',\n",
              " 'missed Brent at praise band. No fun to not have your lead guitarist. ltpoutgt',\n",
              " 'i think to much on the past. i cant change it. i deserved so much more then wat i got. ..but why am i still thinking about him gah',\n",
              " 'has lost his ring. Its nowhere to be seen',\n",
              " 'burgaw Ooooooh sealclap See I download shitloads of zip folders off 4chan. I have no internet moneys. FUCK YEAH ALICIA amp MIKEY.',\n",
              " 'Is still nursing my Nile but glad he is feeling better. I hate when my baby is sick',\n",
              " 'Is fucked to go back to IC',\n",
              " 'Brandizzzle08 yoyoyo. my internet has been rude tonight. it JUST reconnected and Im about to go to bed.',\n",
              " 'jemcam well i have uni stuff and netball but after netbal if ive done uni stuff we can',\n",
              " 'kissability Me too I is poor',\n",
              " 'help me forget 8th april amp 13th july',\n",
              " 'dierks bentley is comin to columbus OH i wanna go so bad',\n",
              " 'i have to take my sidekick back.',\n",
              " 'chriscantore congrats Im totally jealous only wish my XM was working',\n",
              " 'gr8t my face is very itchy',\n",
              " 'poor socks luvvvvv the golden retriever I want one sighhhh',\n",
              " 'I just saw that they found that Tracy girl in a piece of luggage... How fucking terrible',\n",
              " 'Aaaaand the nausea is back.',\n",
              " 'DonnieWahlberg ooh Im excited and not even going 2 be there long love YOUTUBE',\n",
              " 'spent 1 hour to reach to Axis bank only to find out today is holiday for Mahavir Jayanti contd..',\n",
              " 'mathewsmichael i agree... the jobros dont update theres very often',\n",
              " 'fishmouse it is hilarious and I linked the clip from LJ some time ago but when I went back just now it was a dead link',\n",
              " 'DiannePulham OOOOOOOO who with im not neither but thats because i need to study',\n",
              " 'bostongarden I miss bentley',\n",
              " 'seriously needs to finish these job applications',\n",
              " 'My son Vincas is sick so I stay at home Just three tense days at work and I am back on holiday with kids',\n",
              " 'Whinging. My clientampboss dont understand English well. Rewrote some text unreadable. Its written by v. good writerampreviewed correctly.',\n",
              " 'huntermoore I dont want him to ever punch me.',\n",
              " 'sooo sick of the snow ughh',\n",
              " 'ITSNEMESIS',\n",
              " 'LoveMeagan it doesnt work your fan is upset',\n",
              " 'I would like to apologize for the repeated Video Games Live related tweets. I am going to have a stern discussion with Koodo soon. Stern.',\n",
              " 'zaydia but i cant figure out how to get there back pay for a hotel etc',\n",
              " 'FML So much for seniority bc of technological ineptness I now have to quotregisterquot for classes AGAIN',\n",
              " 'feels like she slept the day away. Not looking forward to any more bouts with my gallbladder. At least I have pills now for the pain.',\n",
              " 'ThaStevieG but what I really want is my old bass back',\n",
              " 'Mad tired today... Callin it in early tonight nighty night twittas',\n",
              " 'LevenRambin Take it easy and be good to you.',\n",
              " 'Im afraid I had bad code.',\n",
              " 'Think Im going to bed. Goodniight. I hate this',\n",
              " 'riancurtis im here friend and i love you.',\n",
              " 'kariajay All this time you didnt notice I was gone...just needed DB2 is it',\n",
              " 'aaronrva is in the bathroom and i have to pee',\n",
              " 'ashleyskyy but I wanted a margarita too',\n",
              " 'KourtneyKardash yup night workouts r the worst but unfortunetly my work schedule only allows me to go at night its tough',\n",
              " 'Just called Hillsong again they said they couldnt tell me where I was on the waiting list I dont know if its looking so good',\n",
              " 'has 0g of milky bar left and around 200ml of coke',\n",
              " 'jokerrrr It stillllll hasnt arrived',\n",
              " 'Mraow I feel like dancing but first art school wants to rape me some more.',\n",
              " 'karenucol ah I bet it does Ive been like craving to play softball or basketball But I have noo time',\n",
              " 'Is terrified she accidentally deleted a reference in her management assignment and hopes she doesnt get in trouble when she gets back',\n",
              " 'amazon s3 plugin not worked in my website . it need PHP 5 . but my hosting provider has only PHP 4.4 . very sad.',\n",
              " 'Argh I was suuuper sleepy an hour ago now Im wide awake. Hope I dont stay up all night.',\n",
              " 'austinhill I wish I was Sold out If you have some time let me know I would love to hear more about what you are up to',\n",
              " 'Just watched tropic thunder. Dreading tomorrow',\n",
              " 'qweendassah no...hes still missn...',\n",
              " 'ChiefDelphi is down',\n",
              " 'httptwitpic.com2y2wr according to my bro our new puppy had a poo fight and was covered in poop picture stolen from him',\n",
              " 'playing game at home my new boss didnt call me yet...',\n",
              " 'Su yin Huen tweeted I feel unbearable guilt. I made my staff cry httptinyurl.comcw2l9t',\n",
              " 'Poor Joshy is sick those damn tejanos',\n",
              " 'My thoughts are with Sandra Cantus family at this difficult and sad time',\n",
              " 'FranzGlaus I know just wanted to let everybody know what was happening',\n",
              " 'Michigan who I dont know them. sigh Shameful',\n",
              " 'i have to fill two hours',\n",
              " 'alicayaba so cuuute hey i miss you na its not the same not seeing you girls everyday',\n",
              " 'JonathanRKnight Oh Did I mention it quotGooooood Moooorniiiiiiingquot from Germany Im back in my cage....or better...my office',\n",
              " 'throat is closing up and i had some string cheese. not a good idea',\n",
              " 'Aww Sandra Cantu is found dead in a suitcase Missing children stories never seem to go good. Poor family.',\n",
              " 'wishing i was home underneath my covers',\n",
              " 'walking home from school drains all my energy',\n",
              " 'ridley1013 I agree. The shapeshifting is a copout. I was so excited for Angelas ep I thought it was this week. Noah was awesome tho',\n",
              " 'SaveTheStrib Dang I would have done this if I knew it was there earlier. Any other outings planned',\n",
              " 'I wish I was better at writing. Its taking me so long to write this paper.',\n",
              " '..and of course... I have access to my Halo 3 Mythic Map Pack redownload.. but bad news.. not the Legendary Map Pack... ugh.. 600MS...',\n",
              " 'vene2ia ...not yet unfortunately another few weeks Ive been told How are you Ive been a little bit busy with my latest project.',\n",
              " 'thecoolestout Ha the suns already gone',\n",
              " 'day 2. a lot harder than day 1. lets see how day 3 goes tomorrow.',\n",
              " 'My man crush Jake Peavy let me down gayforpeavy',\n",
              " 'trishzw megafast trip you have time during day tomweds Or you going out tonight Want to see you bad. Tried to warn you tech fail',\n",
              " 'michigan state you make me sad',\n",
              " 'omg. my mouth is in so much pain i just wanna sleep untill its time to take my braces off.',\n",
              " 'getting annoyed easily today gtgtgt biofuel proposal getting annoyed easily today gtgtgt biof.. httptinyurl.comceprvs',\n",
              " 'MizzChievouz Hey girl. The site is back. Girlyvue is back and they have even more videos.',\n",
              " 'a5hleyf im spending time with my grandma early tomorrow and i cant leave skittles by herself.',\n",
              " 'just got home from watching Michigan State get tromped',\n",
              " 'Oh jewbus Two years worth of classes work is ALOT to organize gtlt im not finishing tonight its not possible...my bones ache',\n",
              " 'griffmiester no exchanging for me my laptop hasnt arrived',\n",
              " 'hyperbets i hope this doesnt last too long. i feel miserable',\n",
              " 'rcompo RACHEL hang outage is neccessary i was gonna be home this wekend but dumb folk make me work so im free F and Sat until 4...',\n",
              " 'TheLeagueSF Not Fun amp Furious The new mantra for the Bay 2 Breakers It was getting 2 rambunctiousthe city overreacted amp clamped down',\n",
              " 'amberbenson my hubby for some reason thinks its more important he has a good spot for red carpet for the Trek Movie Premiere. Sod',\n",
              " 'I swear no matter how long Ive been getting up at 5am it never gets any easier. Man my eyes hurts wah',\n",
              " 'JeffreeStar Jeffree How do you keep your hair one color My PERMANENT dye fades within the week',\n",
              " 'otakusecret my reaction too whoa didnt see that coming',\n",
              " 'still sick. feeling a bit better got some new medicine.... hope I feel good after a night of sleep. ohh and its suppose to snow WTF',\n",
              " 'Dangerm0use I think maybe you should get a couple more hours of sleep hon. How productive can you be right now if ur dog tired I worry',\n",
              " 'maddyVA Thanks a lot. Learned a lot of new words but didnt find what I was looking for.',\n",
              " 'Wish I had all the XBLM downloads all on the gamertag Hitokyri. This 5 gamertag thing is nonsense... and Im tired of it',\n",
              " 'New Testament Test at 930 am',\n",
              " 'ScoutBuck tons no hay Troll ahhhh',\n",
              " 'stupid movies we watched... mirrors ugggggh... stooopeeed rip off',\n",
              " 'httptwitpic.com2y2yi I love you Buck.',\n",
              " 'last one but still not done',\n",
              " 'homework....',\n",
              " 'I dont understand why things get taken away. I didnt have enough time with him its unfair. I want him back He was like my baby',\n",
              " 'Scratch that du in heif two in a half cookies. My tum just cant take anymore shoog. Poots',\n",
              " 'cococourtney i was just listening to the sweets for the first time in forever i miss them so much can we go to chi town for visits plz',\n",
              " 'Geez what a busy afternoon. Meetings emails meetings and email and more meetings. 6pm and the day is still going Ah its dark and',\n",
              " 'DANG ANOTHER SPRING BREAK HOW COOL IS THAT TOO BAD I HAVE TO WORK ALL WEEK OH WELL MORE MONEY FOR A PHONE. ttyl EVERBODY',\n",
              " 'SaraKate Im afraid too ur reply about uni from ages ago',\n",
              " 'CarVin1 lol they are some emotional ass men Omg all this late night eating.. Both of us are broke',\n",
              " 'nickcarter Aww Nick I like your hair longer why did you cut it off Break. My. Heart.',\n",
              " 'wonder if Jon lost the net',\n",
              " 'Wow. The most depressing thing in the world is losing a video that youve created in a matter of seconds. FML',\n",
              " 'nchokkan httpswww.mycomicshop.comsearchTID395031 But all says not in stock',\n",
              " 'Poohpot lmao im sorry poohpot i ate it all',\n",
              " 'Any chance Softbelly for MTUB isnt loud',\n",
              " 'Time to move my posterior and lose some fat. My articulation are creaking so no more running but Im drool for some swimming',\n",
              " 'listening to nathan cry....',\n",
              " 'today sucked. Im gonna die without chris and callum wahhh',\n",
              " 'STOU site not update no info about 22552 exam and open course I must wait for mail notices and doc. um...',\n",
              " 'ohsbrat30...Im sorry. Im feeling kinda yucky myself. 5am is going to come too quick',\n",
              " 'is in love with scrappy and is missin him already',\n",
              " 'wants to hear seblefebvre s song but myspace is messing with me',\n",
              " 'Bad news was Dad has cancer and is dying Good news new business started and I am now a life coach practising holistic weight management',\n",
              " 'duchessrebecca Man... intervention is soo sad',\n",
              " 'HELP I need a new boyfriend... Im stuck in a rut',\n",
              " 'I feel like a complete idiot. Im the only one who doesnt get how this shit works help me',\n",
              " 'Saw an ad on Craigslist for a casting call for a female host on G4. I was totally psyched But realized it was a hoax',\n",
              " 'Pepperoni rolls in L.A. I called Valentinos they said that they had sausage rolls but no pepperoni rolls httptinyurl.comcec5ka',\n",
              " 'soillodge yes it will be. its only Monday',\n",
              " 'i want the new GG episode already',\n",
              " 'I miss watching Rockos Modern Life.',\n",
              " 'RIP Sandra....its so sad how can someone do something like that',\n",
              " 'Keeping my fingers crossed for my buddy He is not feeling well.',\n",
              " 'Wait should I eat Or be skinny for vegas Im hungry',\n",
              " 'MargeInovera I tried tweetdeck once and I hated it with a passion. Or it hated me Im not sure...',\n",
              " 'benackerman BTW my iPhone is acting funny dying quickly and freezing and BS. Im not admitting anything just sayin',\n",
              " 'Sitting here wondering why quotEDquot still has such a strong hold on me',\n",
              " 'bananaface IM SORRY I GOT YOU SICK. lol. going to bed too. NIGHT',\n",
              " 'carolrainbow no Internet at home have to drive into Ox to use Internet because of builders',\n",
              " 'SteveBuscemi the weather in Canada is freezing',\n",
              " 'twiggassssssss I been out of range all day Im back now and hopefully for good.',\n",
              " 'danadearmond',\n",
              " 'anyone who reads this pray for my Grandma. Shes in pain',\n",
              " '3 woke up and was having an accident quotIts pushing its pushingquot he was crying because he couldnt stop from wetting his pants.',\n",
              " 'Home from Franklin street I almost jumped over a fire. Someone kicked it before I could. Jenny lost her shoe.',\n",
              " 'johnnybeane hey you just changed your default.',\n",
              " 'Finally gave in and was bored enough to start this thing. I think the 140 is going to be a problem for me tho. its like a myspace status.',\n",
              " 'ellievolia If oooonly we were really so lucky eh And awh Definitely too early for work.',\n",
              " 'buttload of homework',\n",
              " 'needs another copy of visual studio 2008',\n",
              " 'why to we the only school in the world be in this week.My school is silly anyway... look on the bright side. When im off u will be in',\n",
              " 'omg ive an economics test. and i dont know all the things i have to know and omg im gonna fail',\n",
              " 'robcthegeek Once upon a time Hundred Islands was the pride of Philippine tourism but coralclamrare fish poaching did it in. Kinda sad',\n",
              " 'Henkuyinepu yeah apparently i have really bad taste',\n",
              " 'hummiemd I KNOW I was really surprised since everyone recommends them on youtube. I have to call them tomorrow and figure it all out.',\n",
              " 'Waiting at the airport for my ride while I get harassed by 2 men trying to sell me ugly hats.. Why me I just want to sleep..',\n",
              " 'i left my IDs at the location and now i can not go out to bars FUCKKKKKKKKKKKKK',\n",
              " 'celycarmo im sorry I dont understand your last comment',\n",
              " 'OfficialRandL when is the announcement i stayed up late last night',\n",
              " 'KishoreK this is strange illegal torrents avlbl everywhere legal DVD not in stock What do I do now',\n",
              " 'ITS NOT ENOUGH TO SAY THAT IMISS U',\n",
              " 'This trolley has up packed in like sardines Padre game and they REMOVE a car Good call MTS',\n",
              " 'RandomlyNat Jeez I was just trying to help Haha.',\n",
              " 'SupaMagg that happened to me saturday night. along with my glittery green lighter',\n",
              " 'trying to reduce the number of cigarettes in each day',\n",
              " 'yea it is so quiet around here cuz everyone has to work im bored to death with nobody to talk to',\n",
              " 'gotta do my cooking assignment its too hard',\n",
              " 'treesahquiche okay about the applepears Ive talked to a few people and theyve ALL had them before. No one ever told me never',\n",
              " 'jennifermf I know Im a night owl by nature hahaha Am I a time zone behind you Its almost 1 here',\n",
              " 'mamasvan lol nope but i did have complete camerafail',\n",
              " 'maxime68 megelder But Im useless if I dont sleep Its not fair I want all the fun as well',\n",
              " 'My Pookie has a UTI. I have to be Nurse Gabbie and get him back to 100 No more sodas',\n",
              " 'jillianfish tweet something damn it and hang out with me please',\n",
              " 'anistorm',\n",
              " 'Im SO behind in video games and everything for that matter. Im yesterdays news',\n",
              " 'sleeptilnoon I did it was the only one left but I got tempted and ate it soon as I got home I had a chocolate velvet amp ate d mocha',\n",
              " 'Is stressing out because my blackberry keeps flashing a red alert status for no reason Its faking me out and I hate it',\n",
              " 'httptwitpic.com2y34e I wanna wear my Doc Martens out Havent worn them since December.',\n",
              " 'When to the shoe repair shop and the guy could not find my shoes So I have to go back and see if he can find them in a couple days.',\n",
              " 'BBarnett I did not really see that coming',\n",
              " 'just woke up from the most vivid sketchy notcool dreams of my life. time to stare at the wall with the lights on.',\n",
              " 'itll take 3 days for my sister to get her passport',\n",
              " 'nicolerichie OH MY YES i miss',\n",
              " 'Uh oh... I think I am getting sick',\n",
              " 'Good GOD they ruined my belly button',\n",
              " 'Watching Who Framed Roger Rabbit. Makes me miss Toon Town.',\n",
              " 'akianz renting.. but very cheap renting and awesome house. We had 3 buying attempts fall through',\n",
              " 'running nose spinning head not a good combination for a meeting',\n",
              " 'EazyDoesIt87 NEGATIVE.. you lost my vote of confidence',\n",
              " 'Nasty budget due and my iphone is being sent to Apple today.',\n",
              " 'SofiiNoel thats bad',\n",
              " 'Dammit episode 3 of Kings wont play for some reason stopped in the middle now wont do anything',\n",
              " 'Im so upset that I missed my chat and quiz online because my free internet has ceased',\n",
              " 'cant believe it Im in disbelief of it all in a way really. How much can I take from people and always get hurt by others',\n",
              " 'Car show season has started without me',\n",
              " 'my throat is raw.',\n",
              " 'KevChoice I just dont get it. What in a persons mind could even have them on some ish like this My prayers are with her fam for real',\n",
              " 'frago I hate you. I didnt need to see that. I need to call my therapist now.',\n",
              " 'Twilight didnt come yesterday fingers crossed for today',\n",
              " 'NileLegania glad I was able help you feel better. I hate to see you sick Love you',\n",
              " 'is tireddddddd. want to sleep but i have an assignment to finish and an exam tomorrow to study for',\n",
              " 'You know I was thinking... I need money',\n",
              " 'omg..its jst teusday',\n",
              " 'ddlovato DavidHenrie ummmmm i cant find it.',\n",
              " 'Just heard that they found Sandra Cantu... She was only 8 yrs old',\n",
              " 'corrosivecandy Ive forgotten how to snoosnoo I wonder if snoosnoo and I will ever meet again...',\n",
              " 'AmaNorris wow that last tweet made me seem like a giant sexist...sorry about that',\n",
              " 'My bathtub drain is fired it haz 1 job 2 do amp it iz FAIL. I got all Drano on its ass amp iz STILL NOT DRAINING. I wanna shower dangit',\n",
              " 'stuck at home',\n",
              " 'nomadicmatt mine is 110 how do you get it up',\n",
              " 'Duckling in famous childrens book stolen from Bostons Public Garden The Boston Globe httptinyurl.comdc2htx via ShareThis OH NO',\n",
              " 'AppomattoxNews Thank you However I hate to be on the same list that includes 2 convicts.',\n",
              " 'Heading to Altrincham again Out of hours install',\n",
              " 'MORNING Good im bloody knackered Work is not for me today HELPPP xxx',\n",
              " 'jeffreecuntstar I dont have a garage. But you can park in my driveway',\n",
              " 'lauredhel What happened',\n",
              " 'quotOn popular musicquot by T.W.Adorno is probably the most difficult reading ever prescribed Im actually struggling to continue',\n",
              " 'httptwitpic.com2y36e cant see the flowers falling i dont have a camera just my cellphone',\n",
              " 'Weve been good. Im not liking the snow right now was getting used to the nice Springlike weather. How about you',\n",
              " 'brianarmstrong What a pity YouTube is currently blocked in China. I cant see the vids.',\n",
              " 'margaretcho what happened to your show it was the hightlight of my life',\n",
              " 'This is the best leather sofa in the world Its in the office though which means Im still here working',\n",
              " 'Still 3 more days until my Internet gets uncapped',\n",
              " 'If he doesnt get better in a few days he could have something lodged in his belly',\n",
              " 'i Love my atekinzz so much. ampamp i soo miss her..',\n",
              " 'I wanna sneak into the zoo and play with the kitties',\n",
              " 'lolitariot oh no hope youre not getting sick too',\n",
              " 'alejandralei i dont think i can cause its my cousins birthday party',\n",
              " 'Our sweet little man just fell asleep while waiting up for daddy and big brother to get home Poor baby he missed them so much today',\n",
              " 'the angel is going to miss the athlete this weekend',\n",
              " 'trust is hard',\n",
              " 'have watched that considering today. yaknow. shawnna tomomorrowi need my bestfriend',\n",
              " 'Grr i want to keep reading but if i do i wont have anything left to read',\n",
              " 'atekinzz where in the world are you',\n",
              " 'RumLover No u supposed to be my date',\n",
              " 'pears amp Brie bottle of Cabernet and quotWin a Date With Tad Hamiltonquot... oh gawwd my life flashed forward to when Im 40 with my 75 cats',\n",
              " 'Crap. I need more dresses too.',\n",
              " 'kaeeeep yeah i know it was horrible ugh. saddening.',\n",
              " 'Brodhe geez ur no fun are you',\n",
              " 'I wanna be in a punk rock band again',\n",
              " 'jyesmith Thats a lot of angst for a Tuesday afternoon',\n",
              " 'Im very glad Britney isnt crazy anymore That was one hell of a show Now Im sad that its over. Next up...AP at Cook County',\n",
              " 'has a huuuge headache omg I feel like crap',\n",
              " 'Anistorm Sorry',\n",
              " 'cleders sorry...i was rooting for them too...',\n",
              " 'Man that took forever.',\n",
              " 'shandasaurus I see.',\n",
              " 'I am listing more items on ebay to sell Takes forever..meanwhile Coleman is watching The Hills season premier without me..',\n",
              " 'Just got my presentation done 23 slides done... Im crying for this week will be the hardest of all weeks.',\n",
              " 'Just lost 160',\n",
              " 'I dont like the previously on skins thing that starts at season two. I like it when it went straight into the intro like before',\n",
              " 'Houndour ...i wish i was there...im pretty good at scaring the shit out of people',\n",
              " 'Going to sound vain... but running out of my fav lip gloss',\n",
              " 'stuiy never again will I click on a link that screams quotIm a spider Im a spiderquot. I should have known better. yucky',\n",
              " 'Why is that when you have time off from work you get sick',\n",
              " 'pratama Same iMac came out 320 more in Indonesia than the States.',\n",
              " 'Dammit... I need to stop buying furniture',\n",
              " 'MissSil no i was half asleep and turned off the laptop after that',\n",
              " 'Death is peaceful life however bites',\n",
              " 'i think JonathanRKnight lost the net sure hope not',\n",
              " 'httpis.gdr8Zf httpis.gdr8Zy and httpis.gdr8ZG test footage with my girlfriend in HD The dark one is underxposed',\n",
              " 'Such a tough game to watch tonight for State Finally going to bed after also staying up to watch the season premiere of the Hills',\n",
              " 'Seems jRuby support for hpricot is now two versions behind',\n",
              " 'JinxCat Unlike my sister...I still dont have one',\n",
              " 'Fraking app store is pissing me off httptinyurl.comc4ooho',\n",
              " 'Logging out. I need to study....',\n",
              " 'anyone else having problems accessing TTB I cant get on',\n",
              " 'I strongly dislike people who make stupid comments and dont know the whole story',\n",
              " 'emmaketurah im sorry Emma is swarley a goldfish im sad for u.',\n",
              " 'All my tweets are already gone arent they Missed you guys tonight.',\n",
              " 'Cant sleep again. Face is kinda swollen. Dont let me be allergic to the thing thatll get me to Thursday. School tomorrow Doubtful.',\n",
              " 'SnicktheDog If Izzys on the cat tree shell stare him down but if shes on the floor he chases and she runs.',\n",
              " 'I am officially banning godaddy.com from my comp. My head hurts from the small print AND I wasted 10 that couldve happily gone to Boba',\n",
              " 'HumanOpium pretty much just scary for me',\n",
              " 'ilovepie mines too... Im finding it well hard to get fit...',\n",
              " 'Good morning Ready 2 go but I want 2 go back 2 bed',\n",
              " 'hates waiting for mails',\n",
              " 'wheat bread from the dollar store just doesnt toast nicely going over my taxes and calling it an early night',\n",
              " 'Why Kutner I mean I knew something was going to happen to someone but it was so sad. I lt3 House and it was well done but Im still sad',\n",
              " 'JonathanRKnight aw ok goonite....',\n",
              " 'ill so i cant go to the cinema',\n",
              " 'Fell asleep . . . Really didnt mean too christina',\n",
              " 'KelleyRowe hey you remember that time we used to be friends ugh. vomit. sick. i need sleep',\n",
              " 'my lymph nodes are as massive as rahm emanuels balls right now',\n",
              " 'Oh no my computer sucks. I dont think ill be able to listen to the Xbox 360 Fancast tonite',\n",
              " 'is missing playing my trumpet',\n",
              " 'My throat is still really sore. I was meant to be going on a 5 day camp from Friday but not so sure now',\n",
              " 'sasii I know exactly how you feel',\n",
              " 'Well bed time now. 1 am. sighs Back to 6 am mornings for a week on Wednesday.',\n",
              " 'hert Jesus Camp Yeah...',\n",
              " 'MaryBethune oh no with everything that happened today i forgot we were going dress shopping for the opera tomorrow.',\n",
              " 'Turtles are better than my Mac lasts longer and moves faster',\n",
              " 'I am soaked. This is not pleasant',\n",
              " 'ginayates Sorry to hear about Maggie. Thoughts to your mum.',\n",
              " 'ewarden you may have to email this one to me I hope you feel better.',\n",
              " 'miamiiboii dead yu gettin on wen im leavin',\n",
              " 'arlenecd PLEASE TELL ME THATS SOMEWHERE CLOSE TO CALIFORNIA LOL',\n",
              " 'phlaimeaux where are you',\n",
              " 'stupid arranged marriages... ill convert so you can marry me...love you',\n",
              " 'emilyruppe Well he said that he is a looser and that is what the show implies',\n",
              " 'Poor Sandra Cantu amp the Cantu family My prayers go out to them What a sick world we live in. She was only 8',\n",
              " 'fuck omg austins always there though man lt3 love you',\n",
              " 'nicolerichie I cried so hard when Matthew died',\n",
              " 'sad that the feet of my macbook just fell off sad that the feet of my macbook just fell off',\n",
              " 'chordsy why am I the last to find out about these things Like that youre on the twitter too.',\n",
              " 'lovebscott absolutely not',\n",
              " 'httptwitpic.com2y3cf Filled with curry the true indian in me is coming out',\n",
              " 'im soooo cold right now',\n",
              " 'heidimontag lol I kant believe cam got beat upp sooo embarasssiiiin',\n",
              " 'JonathanRKnight Good Knight hun Looking forward to ur tweets again Hate that i keep missing out on the fun cuz of the time difference',\n",
              " 'Tumblr This is exactly how it feels wearing a tie httptinyurl.comc8bvqh',\n",
              " 'daniela95616 hahaa i just realized quotimpunequot definitely isnt the word i wanted... dang. im so unclever.',\n",
              " 'watching old videos of dance team and such... makes me miss it',\n",
              " 'Wahh no CSI anywhere DVRed I Love Money 2 though gonna watch it then sleep Goodnight Tweeterslt3',\n",
              " 'Struggling hard with Inventory',\n",
              " 'ddlovato Do you hate us Please dont',\n",
              " 'ballinbitch haha im not THAT ballin. I still got bills to paaaaay',\n",
              " 'Im in one of those days when I really just want to work from home. Crap.',\n",
              " 'paulewog Wait...is it a game or just episodes Im so confused.',\n",
              " 'carlyw66 HAHA. I cried... I guess he did....',\n",
              " 'I dont get statistics Its all a bunch of mumbo jumbo for me.',\n",
              " 'sloanb Got it sadly only works for UK amp US at the moment',\n",
              " 'Evicted',\n",
              " 'doesnt enjoy learning CSS.',\n",
              " 'nachojohnny Brian dont make me fuck u up LOL I replied 2 ur message...did u get my I miss u',\n",
              " 'heidimontag I dvrd it and went to watch it apparently the dish network went down SO I MISSED IT I will have to catch a rerun',\n",
              " 'JessDubb lmao. U may have to wait for another season to come thru. its summe boo. Hot nd dry ass weather',\n",
              " 'LizChavez I cant set my feet on the ground. The missing eyelid people might get me',\n",
              " 'daniela95616 my english professor would be ashamed.',\n",
              " 'O I wish the frogs werent becoming extinct httptinyurl.comcxe8w7',\n",
              " 'Came home from Cross County Tired as',\n",
              " 'I dont want to walk home in this snow. Who wants to pick me up',\n",
              " 'heidimontag Lauren is being stubborn. she loves you you love her. I feel so bad for you though. it made me cry lol',\n",
              " 'is grouchy. and wants bmar',\n",
              " 'my heart is broken every morning dropping Foo at pre school now i understand when moms say quothe has my heart brokenquot.',\n",
              " 'Not feeling well and back hurts',\n",
              " 'Im in the mood for some chocolate. I want..... Miniature Reeses cups. Now',\n",
              " 'I cant take this heat Its like an oven in here. I feel sick nwo',\n",
              " 'nickcarter Come to the chat just 15 minutes please httpfanclub.backstreetboys.comchat.php',\n",
              " 'Boredd. Colddd Internet keeps stuffing up',\n",
              " 'I had a horrible nightmare last night which affected my sleep now Im really tired',\n",
              " 'Born4Broadway Lost and it was St. Ignacius Prepatory School. Haha.',\n",
              " 'birutagme yeah it was really depressing... stuff like that gets to me',\n",
              " 'whoisariston No problem tongueincheek remark. Lots of acronyms but no NZ',\n",
              " 'greggrunberg hey you said matt was gonna go all ballistic... i was disappointed good ep tho mostly',\n",
              " 'ugh horrible ending to the sandra kantu story prayers go out.',\n",
              " 'danielledeleasa httptwitpic.com2y1pe I hate 2 B allergic i want a puppy soo bad that ive already thought some cool names',\n",
              " 'Morning folks 7.00 am ..yawn Up amp away to see to horses hope Twitter is better behaved today Last night was a travesty',\n",
              " 'QuanDotCom I miss mine too',\n",
              " 'AmandaEnglund Sorry to hear about your loss. There have been many this year so far.',\n",
              " 'nicolerichie Gossip Girl was a repeat',\n",
              " 'InYourEyes2410 I reckon',\n",
              " 'finding it really hard to use twitter',\n",
              " 'Ive been stuck in this house for two days',\n",
              " 'is really sad and doesnt really know y',\n",
              " 'thousandmiles no',\n",
              " 'Maybe one of these years Ill get a tax return. A girl can dream right',\n",
              " 'Back at work tired as hell and i feel a cold coming',\n",
              " 'clarianne Knot2serious APRIL 9TH ISNT COMING SOON ENOUGH',\n",
              " 'hannahsix cream for his eye and he may have herpes not ocular herpes but a different strain. Hes doing okay though.',\n",
              " 'gotta repeat whole art folio cuz old one with a whole terms work got wet',\n",
              " 'Im ready for the weekend already. Its only Monday.',\n",
              " 'lovebscott umm nope..think im an insomniac ..plus i got the flu...illl be sleepin like a phuckin fish outta water any minute now',\n",
              " 'has Finland sleeping for the night not really enough time',\n",
              " 'lunafiko cant wait to try em but prolly have to wait until next weekend at the earliest',\n",
              " 'Shiner is taking up all my bed and blankets',\n",
              " 'clarianne APRIL 9TH ISNT COMING SOON ENOUGH',\n",
              " 'wants to cry',\n",
              " 'I need to go to the bank tomorrow before I go broke...',\n",
              " 'This earthquake in italy has me sadden. its only three hours away from naples where my family is',\n",
              " 'i should be sleeping i have a stressed out week coming to me.',\n",
              " 'dragoneer Yea I am working tp hack my server to do the 12player thing for folks',\n",
              " 'has hurt her ankle and is going to the dr',\n",
              " 'raymondroman oh noes. how did you manage to send something to the trash can AND empty it without noticing i feel for ya',\n",
              " 'Lost my free copy of radioactive so cant put it on my ipod grrAnd i cant go out buy nother coz it came with that paper',\n",
              " 'doing make up shit',\n",
              " 'Photoshop I hate it when you crash',\n",
              " 'vacations makes me feel sick',\n",
              " 'you shouldnt have gone',\n",
              " 'listening to Murd and 9th wonder just chillen out missing my crazy sex life',\n",
              " 'louie094 shooting be careful luisa',\n",
              " 'hillary006 Im sure everyone has ruined my gift to you Whitney has my serious Cell ...We doing easter as well',\n",
              " 'is at Beso in Hollywood... so tired',\n",
              " 'I dont want to wake up early tomorrow. Damn you work',\n",
              " 'just finished watching the movie Prayers for Bobby. Broke my heart',\n",
              " 'syekr is myki really horrible',\n",
              " 'stephenkruiser i had 2 put my dog down today too someone ran her over',\n",
              " 'NevadaWolf Sorry to hear about your DNF run tonight. Always a risk to get bad coords on an FTF',\n",
              " 'Honey3223 I did get some disgruntled people when I wrote my ESN but the 1000 hits in one day made up for it. Bill pulls in 250 tops',\n",
              " 'moony394 OHH OMG. LMAO. Im crying right now LOL KUTNERRRR was the best',\n",
              " 'came back from running and took a shower...why does my lower stomach still hurt after exercising',\n",
              " 'watching Fallon with mom and working on a project...school is hard',\n",
              " '56 min till home time and counting I have a craving to visit the MAC store but they will have closed by the time I get there',\n",
              " 'stephenkruiser I am so sorry to hear that Take care',\n",
              " 'PaulColes hmmm...greed is good when it motivates the individual to do betternot so much when it takes away from others',\n",
              " 'GABBYiSACTiVE brat You have to rub it in dont you.... I want a mocha',\n",
              " 'KUTNERRRR WHY WHY And to think that 13 is still on the show UGH. KutnerKal Penn youve been the bright star in Ho ...',\n",
              " 'redvinylgirl my mom has it. I wish you the best of luck',\n",
              " 'I so hate homeworks . My head hurts so bad',\n",
              " 'got the ebay blues item i want jumped from no bidders to over 100 in an hour.. still has 3 hours to go. id better not get my hopes up',\n",
              " 'gigdiary I know was a little depressed that we ate so much last night there were no leftovers today',\n",
              " 'well the pc in my living room suffered a fatal partion lost all my itunes library is gone.. and some docs with pics....',\n",
              " 'sad day manu out for the season',\n",
              " 'stephenkruiser You poor spunky...its so sad...my Rotti made it to 10 and I had to go down the same path....I know how you feel',\n",
              " 'im sooo sad right now i need a hug',\n",
              " 'wants her step brother home im so worried you dont even know. i hope he knows how much i care for him...',\n",
              " 'heidimontag just got done watching the hills loved it excited for the next episode and im sad this is the last season',\n",
              " 'ryanmwilson aww that stinks hugs',\n",
              " 'Im tired. I feel like crap. And the world feels all crummy. Make me happy USB disco mouse.',\n",
              " 'i dont want to believe what im reading...buuso sad',\n",
              " 'Wondering why gamebattles is down',\n",
              " 'White trash vs us... We were outnumbered. httptwitpic.com2y3jp',\n",
              " 'emitstop So true I have it and dont use it',\n",
              " 'Spring break is here at last but no one is here to celebrate',\n",
              " 'Morning Bah Car wont start.. waiting for the ANWB....',\n",
              " 'katebornstein which is pretty anti memorial tattoos but for all but the strictest theres no official BAN just disapproving family',\n",
              " 'Balamurugan a longterm member of the team is leaving this month. Well miss working with him hope he does well in his next venture',\n",
              " 'JustGelo Its sad knowing how they found her body. What a crazy world we live in. No child deserves that ya know.',\n",
              " 'I wanna rock a Maxi dress Coachella but lacking funding',\n",
              " 'Of course the baby is screaming in the crib the night before my first day back at work. So I let him cry it out Boooo',\n",
              " 'I am home missing my baby Busy week ahead Fri is a chill day with my guy and kids Egg hunt Sat Spiral and DMB Sat night and Easter',\n",
              " 'FollowSavvy I never found her. everytime I click on her twitter thing through your myspace..... it goes to some dudes page',\n",
              " 'stephenkruiser thats horrible sorryto hear that.',\n",
              " 'DavidHenrie i cant find it it says dtmafiaofficial doesnt existtheres jus 1 dtmafia i found n u dun seem to b followin it wht to do',\n",
              " 'i thought you would support me on this',\n",
              " 'augh eff sarth i stole some of MILs nail polish its pretty o',\n",
              " 'Wishes I was on the Spring Fling Tour with Dawn amp neecee Sigh Gknight',\n",
              " 'The Hoyts Distribution NZ website isnt working for me. Great. And it HAS to be like this when I need it most',\n",
              " 'crummyasshole i dont like that they only had him in the first movie cillian murphy is hot',\n",
              " 'wishes for good old cartoons to come back. ie Captain Planet Wheres Wally Daria sadness',\n",
              " 'not feeling v good abt myself',\n",
              " 'i just got a text from sarah',\n",
              " 'Another day at centre point this time an early start so guessing train will be rammed with commuters',\n",
              " 'Four games this fall Modern Warfare 2 Halo 3 ODST Brutal Legend and Bioshock 2 I need more cash',\n",
              " 'dcastillo ugh thats disconcerting',\n",
              " 'Devinthedude007 actually the whole site has finally been taken over by those damn russians',\n",
              " 'MY TWEETS ARENT GOING THROUGH',\n",
              " 'Just finished a delicious breakfast my last in Paris Ill miss milk europe has the tastiest milk in the world I dont drink it in US.',\n",
              " 'iLearn is down and out great considering finals are this week...',\n",
              " 'Carmonium im stressed outttt',\n",
              " 'Also I shouldve Bern asleep an hour ago',\n",
              " 'trying to figure out this twitter thing...none of my friends are on yet. Soo much to see.',\n",
              " 'Repinging Kyle44 Custom icons I made look cool but the edges could be less square might look bad on a diff background',\n",
              " 'I want to be back in LA',\n",
              " 'enterbelladonna i dunno how to use the forum and i get frustrated with it. ill miss talking to you on here.',\n",
              " 'lilbucknuts41 not an option',\n",
              " 'georgeruiz Unfortunately one of those moments wasnt a giant squid monster.',\n",
              " 'thestarrage I want one too is the branch in MidV open too',\n",
              " 'daNanner aw sorry to hear that.',\n",
              " 'Change of plans we ordered Macs instead. Time to hit the books',\n",
              " 'polhillian YUP.',\n",
              " 'NKDreamer did you see Donnies tweet stats almost 700 replies...and no JRK',\n",
              " 'timgrainger nah I havent received my stimulus yet',\n",
              " 'ikimb0 i just seen ur tweet . plz write bak if u get this. i havnt got one reply bak',\n",
              " 'McCainBlogette Awwwww and you were trying to go to sleep 3 hours ago',\n",
              " 'stephenkruiser Im so sorry to hear that. Its always sad when we lose those close to us as we loved them.',\n",
              " 'cant sleep but im still feelin like a piece of shit',\n",
              " 'stephenkruiser oh my sympathies. its a hard decision. i always hope my old border collie will go in her sleep. shes 14 too. hug',\n",
              " 'josephheustess ...well there was this really cool part where... i wont spoil it',\n",
              " 'limbecky Im doing the time warp without you and am sad.',\n",
              " 'theresawhite this is true lol but its still a slap in the face after such a warm end of march',\n",
              " 'I hate converting movies just to put em on my itouch',\n",
              " 'gross i have a pimple',\n",
              " 'Torturedlady yea manaaaaaaaaaaaaaaaaaaaam marwan accent lol',\n",
              " 'one less follower tonight',\n",
              " 'Unholyknight so did your mom last night. brb while I figure out whether I just burned you or myself the most.',\n",
              " 'I feel like I am the only quottwittererquot not on TweetDeck.',\n",
              " 'Wyldceltic1 He has Karate tournament in 8 weeks',\n",
              " 'going to sleep cant be up until 1am again...and no I didnt finish my statistics homework',\n",
              " 'North carolina baby Aint nothin stopping them but the nba draft',\n",
              " 'Cant sleep again',\n",
              " 'I reaaly miss john mayers twitters. fuck rude people. Remove him if you dont like his tweets bitchfucks.',\n",
              " 'is going to have a late one at MQU today',\n",
              " 'im so tired of work...i need a life....',\n",
              " 'just leavin work my feet hurt',\n",
              " 'PlayRadioPlay that was so sad and unexpected. i totally cried haha.',\n",
              " 'Superneej Morning Bah Car wont start.. waiting for the ANWB....',\n",
              " 'Still doing my homework',\n",
              " 'A little sad',\n",
              " 'deepbluesealove My Mom amp Dad both get up around 6...too early for me',\n",
              " 'shouldve been asleep two hours ago',\n",
              " 'stephenkruiser So sorry to hear about your dog.',\n",
              " 'Funny how the little things make me homesick criminals breakn n2 a brownstone on LawNOrd CI made me misty',\n",
              " 'djsoulsister yeah great vid. I had the 12quot single but sold it a few years ago',\n",
              " 'im on vals mac in iitsc clubroom. still sweating',\n",
              " 'stephenkruiser awww so sad Im so sorry',\n",
              " 'UpdatingFFE. That gives me nothing to do for a good twenty minutes',\n",
              " 'Hot compress not rily workng for pulled muscle gng to bed good night',\n",
              " 'Have an invite for quotHealthy Diningquot session at Ashok Hotel today with Exec Chef R.Chopra but damn workload will have to skip it',\n",
              " 'stephenkruiser Im so sorry to hear about your dog wishing you happier days to come',\n",
              " 'hellivina I miss em too.',\n",
              " 'misstoriblack cool i have no tweet apps for my razr 2',\n",
              " 'Im off too bed. I gotta wake up hella early tomorrow morning.',\n",
              " 'still procrastinating... i hate organizing my clothes theres just so much....',\n",
              " 'Just been playing with the new Mobbler v0.4.0 and it adds some great new features but wont play music on my E71 like v0.3.5 did',\n",
              " 'latroneb oh but theres one bestie missing',\n",
              " 'esuriospiritus time to come back to flawda for double dates no seriously im sorry to hear that.',\n",
              " 'Im up way to late to be working for a client 1210 AM fb',\n",
              " 'blegghhhh i have to go to work',\n",
              " 'Back at work John Muir Dr httploopt.usKoqAbg',\n",
              " 'DAChesterFrench I emailed you yesterday and u never responded',\n",
              " 'Its going to be a long year for As fans.',\n",
              " 'Just got done watching the new House episode.Definitely one of the saddest episodes ever.',\n",
              " 'Man... taxes suck. Im horrified that i did something wrong on them. TurboTax decided to keep around a lot of the stuff I turned off.',\n",
              " 'Sleep time. Tomorrow is gonna suck',\n",
              " 'WTF Not Kutner Oh Im so pissed House. I mean really WTF',\n",
              " 'AmberPacific i know i dont know why i said that',\n",
              " 'CASH RULES EVERYTHING AROUND ME',\n",
              " 'Just picked up some oats from the market to eat for breakfast with my boy adamgoldston now time to do work at USC... late night',\n",
              " 'i feel sick too much icecream',\n",
              " 'faithgg45 this computer doesnt have shockwave...blah no account yet',\n",
              " 'RubyRose1 awww wish i could go but its in sydney',\n",
              " 'luxuryprgal hahaha suriously I feel like everyone knows this show is a joke except lc',\n",
              " 'I just cant spell today i totally suck',\n",
              " 'are you vaccinated against chickenpoxshould I vaccinate my 3 yr oldI am SO confused about this damn thing',\n",
              " 'school at least last day',\n",
              " 'Yawwwn Gotta get up early tomorrow whos ready for the weekend',\n",
              " 'i miss being at home for calving season...sad that black one died this morning wish i had some bottle calves',\n",
              " 'stephenkruiser i so sorry for your loss. my brother dog sam is sick wcushings disease.',\n",
              " 'Back to work. I get to spend the next 8 hours alone in the dispatch room',\n",
              " 'theres a new kind of starburst favereds. sounds fab bc cherry and strawberry are MY fav reds but they added watermellon and its yucky',\n",
              " 'Well so much for being productive on my 2 days off from Angels',\n",
              " 'andreabakes Oh gee insomnia sucks Have you tried melatonin',\n",
              " 'vivekg86 Good to hear that we have support in NetBeans. But then its NetBeans you know',\n",
              " 'heidimontag I wish we would get it the same time at the rest of world. unfornately I am in Cairo and have to wait for mtva to play it',\n",
              " 'heartbroken over little sandra. prayers are with the family.',\n",
              " 'hadtobeyou Im at 900 words its all can do Ill finish tomorrow maybe',\n",
              " 'shortiethug how come ur background covers the screen but mine is in the corner',\n",
              " 'Wrote exactly three lines of dialogue tonight then gave up.',\n",
              " 'ashman01 My only complaint about Facebook is theyve changed it so much its confusing',\n",
              " 'Sliced my finger',\n",
              " 'going thru town and traffic on bypass is stink',\n",
              " 'Ugh cant sleep. Wish i had a good cuddle to Make me ass out',\n",
              " 'I hate when software update downloads updates without telling me. I never know whats downloading when it is',\n",
              " 'went to ihop again. Great for a diet At least there were no kung foo pancakes this time. Rachel',\n",
              " 'The internet is hating me. Night i spose.',\n",
              " 'contactabe Im so jealous. I missed all of opening day this year',\n",
              " 'Its so cruel having to wait for reviews to come in. So far it looks like Star Trek will be the best film of the year',\n",
              " 'Debating if I should just shower now so I can sleep a little later tomorrow',\n",
              " 'My stress always culminates with physical pain.',\n",
              " 'need to sleep. but cant',\n",
              " 'my little pinky finger hurts so much..',\n",
              " 'sarahsaturday Im sorry your gift card ran out.... That suck.',\n",
              " 'home for spring break. no one is home tho',\n",
              " 'I accessed oneindia.mobi from a Google phone Android. It doesnt support Indic languages',\n",
              " 'Why does school take over your life so much you dont get sleep anymore .. I am still doing school work and have more to do as well',\n",
              " 'MTsiaklides aw I wish I could I cant really speak to my yia yia because she doesnt speak english',\n",
              " 'winggato no free comp lol',\n",
              " 'getting ready to clean the house from top to bottom',\n",
              " 'crying my friend is moving away tomorrow',\n",
              " 'swiftkaratechop STOP NOMNOMS ..you didnt share Night peoples',\n",
              " 'thinks she is getting a cold that she caught from Daddy. Oh no my trip would be ruined.',\n",
              " 'moony394 yeah is it bcoz kal penn is a visiting lecturer somewhere n cant be on the show anymore im gonna miss him',\n",
              " 'So the procrastination starts early in the quarter. I dont want to reaaaaad and write a paper',\n",
              " 'What Apparently its 52 degress at 11pm in Washington State. I miss winter already',\n",
              " 'being bored at homee',\n",
              " 'massage was great. i just spilled coke all over my desk lame. watching the simpsons as ive finished all my hills dvds.',\n",
              " 'nickcarter I will wait for you at fanclub chat.. even tough youre not my favorite',\n",
              " 'has a mild left inner ear infection.. and its got this irritating quotblockedquot feeling since sunday..',\n",
              " 'JaneBodehouse Hey there Nope. My cuteness HoytFortenberry is away for awhile',\n",
              " 'Its only tuesday',\n",
              " 'Haha not even yo. I just didnt know how to do that thang to you on twitter Haha',\n",
              " 'Someones alarm clock or a phone woke me up at 3am...Still got my headache from yesterday night',\n",
              " 'kalichosich awww poor puppy is she ok',\n",
              " 'i might break down and eat some buffalo wings tomorrow',\n",
              " 'Sittin here wKayla I really really reallly dislike having a brother. He doesnt know when to stop. Im waiting for the day when one of',\n",
              " 'Fed bokkie too many birthday treatssicko',\n",
              " 's Kinda Bummed. An Agency Just Told Me That I Didnt Have The LOOK That They Were Going For. Ladies Please Help Me Create The LOOK',\n",
              " 'Work laptop is officially dead .. Not happy at all..',\n",
              " 'My goodness its freezing down here',\n",
              " 'heidimontag why do you put up with him You should listen to your mom',\n",
              " 'is stucked in Paris and cant even travel into France for work',\n",
              " 'JoannaFBeckett i dont think theres one close by. Last time I checked the closest one was 70 miles away',\n",
              " 'aaaaand back to my literature review At least i have a friendly cup of coffee to keep me company.',\n",
              " 'I am worried that I wont get my 900 even though I paid a buttload of tax last year',\n",
              " 'Okay so.. STILL NO SCHOOL',\n",
              " 'IS PHONELESS ...NOW OFF TOO SLEEP',\n",
              " 'Reeked of alcohol at the dentist this morning was drunk when i put on my tshirt realised it had a huge stain. Dentist looked disgusted',\n",
              " 'it is really cold',\n",
              " 'cant sleep she wants to dream',\n",
              " 'Cant sleep I hate these nights when I try to go to bed early and stay awake for hours. Poo',\n",
              " 'rootbeersoup Yeah. Too bad people like a certain burritoeating man exist.',\n",
              " 'Cronotriggers thats sad.',\n",
              " 'just found out that my mum and my adopted auntie are on twitter.. check them out my mum is TessM64 and my awesome as auntie is tania38',\n",
              " 'Cannot sleep....WIDE AWAKE...and I gotta go to work later too. Boy am I going to have a crappy day.',\n",
              " 'struggling with eam law',\n",
              " 'Why cant I sleep like her httptwitpic.com2y3ty',\n",
              " 'Bouts to start 24. Poor linny couldnt keep her eyes open. she tired and has to work tomorrow morning. Night ilovefatsusan.',\n",
              " 'My 4 year old 18 year old is driving me around the twist...i dont want to be like this',\n",
              " 'Well that sucked.',\n",
              " 'i have to be out of my place in 3 days...any help i can get packingpaintingcleaning is much appreciated. oh. and RIP my wall',\n",
              " 'Opening a partially frozen beer Beer on the toaster under the oven and all over the counter. The kitchen smells like we run a speakeasy',\n",
              " 'stewiebrittany no i dont even know how to ride it',\n",
              " 'MonaSmith sadly yes. i think i need counciling now',\n",
              " 'making myself ready for school...dont feel like going but have two tests',\n",
              " 'Smugness gone. My train also failed to turn up so Im heading for Victoria followed by an unpleasant tube ride to the City.',\n",
              " 'Anybody know where my iPod went',\n",
              " 'aww this earthquake in Italy has made me sad just saw the pics on the news its terrible.',\n",
              " 'lilibat never been able to afford to get them out. Id much rather go through the extraction than continue with what I deal with now.',\n",
              " 'Why do other pet care people try to run others out of business Or send suspicious emails fishing for info',\n",
              " 'kameelahwrites lt33 codez....theyve been MIA tho',\n",
              " 'reannaremick doesnt work on my cell go to sleep P',\n",
              " 'its going to be nice again tomorrow and then the rain comes back. no nice weather for my birthday',\n",
              " 'taliasunset rock band is hard..on expert',\n",
              " 'beachbassbone roomie was home all day. All he had to do was scratch at the door. Comforter has to goto large laundromat machines.',\n",
              " 'One of the hardest thing with this schedule no one I can chat with at the end of my day usually',\n",
              " 'they killed off a character on one of my favorite shows and now im upset',\n",
              " 'jealous of my friends who got to see the new Star Trek film in Austin tonight.',\n",
              " 'Listening to Q I got a really bad headache and a drivin lesson in ten mins.... Lucky me i just wana sleeep',\n",
              " 'SteveBuscemi SHUT UP you...I am hungry And in Britain home of the most boring suckassy breakfast in the world I MISS DENNYS',\n",
              " 'good morning everybody pkoi y fais pas beau bon coffee time',\n",
              " 'work again',\n",
              " 'Essay time....',\n",
              " 'chauncey78 hey did u talk to mom.. r they home yet i hope they r ok.. wish i was coming home 4 easter..',\n",
              " 'endearingevania I sure wish I could go with you guys tmrw. But as per usual school interferes.',\n",
              " 'widyatarina say what kal penns leaving house noooooo awww... i totally missed it tonight',\n",
              " 'no i lost a loyal',\n",
              " 'played another hand very bad and lost half my stack',\n",
              " 'Its 6.20pm early days in a looooooong night at work',\n",
              " 'yaykimo Its sad its the last season. I wanna see when Spencer calls Lauren',\n",
              " 'So Im done editing quotThe Phipstapequot. Back 2 crack a brew and see what we got............. no trees though',\n",
              " 'rj i dont know what half of that means',\n",
              " 'homework homework homework',\n",
              " 'worked his heart out today doubling my weight at each station. Result Stiff and pulled muscles... OUCH',\n",
              " 'MeLaMachinko now i feel bad for unfollowing.',\n",
              " 'So... Tired... God I hate the new job and only two days in',\n",
              " 'Sad about Kutner being killed off my fav show House',\n",
              " 'happy for Coach Stringer HOF co 2009...now if I can only finish my term paper on her',\n",
              " 'I feel like shit. This is NOT the way I want to spend my birthdays eve',\n",
              " 'so far i have 311 on all my sites put together most of them were me checking out the updates i made......',\n",
              " 'has a lot on her mind. I needs to make some moneys. And ideas. Nowhere is hiring',\n",
              " 'emiliexclarkex miss you',\n",
              " 'stephenkruiser Im so sorry to hear about your dog',\n",
              " 'Too much internet. How it plagues me',\n",
              " 'i miss my ex soo much',\n",
              " 'Sodding M25 httptwitpic.com2y3y0',\n",
              " 'oh I thought the pirate bay trial verdict was today s apparently its in 10 days',\n",
              " 'featherinair call me back.',\n",
              " 'is in study hall now and i will log off immediately. httpplurk.compmzxbg',\n",
              " 'it so tired that im crying for no reason at all. im about to try to get an hour and a half in for tonight. half what i got last night.',\n",
              " 'I totally have like... four girl options and none of them seem viable',\n",
              " 'so far i have 311 veiws on all my sites put together most of them were me checking out the updates i made......',\n",
              " 'Listening to Bjorks All is full of Love... crying',\n",
              " 'Omg the hills Then I love money2 aww I think becky buckwild goes',\n",
              " 'is craving for some tantan nabe httpplurk.compmzxcs',\n",
              " 'Cmon sean man',\n",
              " 'Watching quota league of their ownquot...makes me miss mint chocolate ice cream cones and my grand ma fuck.',\n",
              " 'Lost my phone some where.maybe in the grass its just so long hope it doesnt rain',\n",
              " 'tomatosalsa hope you arent referring to me say hello to your new stalker LOL',\n",
              " 'brettyboo why were u sleeping Ur gonna be up all night now how was ur bike ride',\n",
              " 'SLessard But but but . . I wanna listen to the masters too',\n",
              " 'Sephystryx Ive been looking about for good stuff to write but also been doing loads of uni work',\n",
              " 'i feel unloved. dropped 2 tweeters',\n",
              " 'Off to London for the day on Thursday its gonna be raining oh fab wont see much on the London Eye Still Robbie wont care bless him',\n",
              " 'Completely wiped and reloaded the 8900.... Lost all BBM contacts Going to bed sad.',\n",
              " 'mykele sayyyy whaaaa I wasnt invited...',\n",
              " 'Didnt sleep too bad considering I have a workshop starting today. Beginnings of a sore throat though',\n",
              " 'astewart87 oh my gosh that made me emotional haha idk why i dont want to get old',\n",
              " 'Were going to post some stories manually due to Twitterfeed problem. Thats why weve been so quiet lately ....',\n",
              " 'i dont like lukewarm showers',\n",
              " 'linuxfoundation Who should I contact if I need a 1099 for freelance work I did on Linux.com throughout 2008 Never got one.',\n",
              " 'skylineking722 connor its me febi are you really really mad at me',\n",
              " 'Need some help Want to buy a Macbook pro but still inlove with my old powerbook',\n",
              " 'michaeltao man america is so .. borez anyway lol i just realized you sent me msgs on here. i had no idea. i have no flash',\n",
              " 'hellivina IHOPness gknight lovely ladies. im sleepy now',\n",
              " 'Does anybody know how to get your electrolytes back in order With a vitamin work I feel sick.',\n",
              " 'sudam08 yes someone hasnt been reading my posts properly',\n",
              " 'Almost through with my Italian homework Weeeeee.... now if I only understood what I was doing.',\n",
              " 'Who turned the lights on It will be time to get ip then',\n",
              " 'therealnph Twitter hates us both then.',\n",
              " 'I hate the fact that Im genuinely excited about my last tweet',\n",
              " 'celesteclara i cant sleep either',\n",
              " 'TheBlondeTheory Between that and the Italy earthquake its been a very sad news day.',\n",
              " 'champagnemanoir all rain today . Garden really needs it so not quite so bad.',\n",
              " 'Yawn yawn yawn 10 more minutes in bed',\n",
              " 'MajesticFlame ouch sounds very sucky',\n",
              " 'bengottlieb Great idea with the iTunes promo codes they dont work in the UK iTunes store though',\n",
              " 'mattycus cry',\n",
              " 'RumLover An empty rum barrel is a sad rum barrel. shakes head the horror the horror.',\n",
              " 'It was a sleepless night',\n",
              " 'BillyBush she admitted to being a fake',\n",
              " 'BATMANNN i love chutney......',\n",
              " 'got highly bored today scanning 24 pages for daddy. Had to do it',\n",
              " 'misterphipps you cooked risotto without me',\n",
              " 'Keeeerrrrriiiiii I really have nothing better to do then post on this thing at 224am.....wonderfullll. Says alot.',\n",
              " 'YAY FOR BASEBALL SEASON boo to yankees losing their first game',\n",
              " 'LOST sucks ... because i have to work today',\n",
              " 'omg exams killed me how could i not know the difference between haif a circle and a hemisphere lt3 gt',\n",
              " 'In bed... I suddenly feel I wish MA was here wme... Goodnight TwitterFam',\n",
              " 'RachelCmrn I dont like goodnights',\n",
              " 'Greaaat my lappy wont turn on wtffffff',\n",
              " 'M4RKM You stuck in traffic then My journey was traffic free this morning if its any help.',\n",
              " 'Lookin like an allnighter. I hate it when I do this to myself',\n",
              " 'vernonhamilton Youre a stranger',\n",
              " 'Debbugging old VB6 code the day could have started better . . .',\n",
              " 'toddlington only a quarter of a percent babe darn',\n",
              " 'Coming down with a cold or bad allergies. Either way Im miserable',\n",
              " 'Cant believe Cutner is dead on House Sad day. in Santa Rosa CA httploopt.usORPL3A',\n",
              " 'Damnit I was really digging this season of Reaper httpwww.tv.comstory13720.htmlrefstoryid13720ampreftype1101amprefnamestory',\n",
              " 'Gah Comcast doubled our cableinternet bill wout telling us so we are quitting. Can anyone suggest an affordable company in bay area',\n",
              " 'Another early morn with the Duchess although she hadnt managed to cross her legs this time',\n",
              " 'alexbigman you left without saying hi',\n",
              " 'l want but I cant sleep',\n",
              " 'TracyDowds When an Xbox shows those three red lights it means its dead and needs to be fixed for a month at Microsoft HQ.',\n",
              " 'A king sized bed is nice but sad and lonely with no hubby puppy or kitty I am over this whole conferenceing thing.',\n",
              " 'im boredd. gah i wish i could just sleep and get it over with but ive had toooo muchh cooooffffeeeeeeeee bahaha',\n",
              " 'whyinthehell If I may butt in again. Im done being nosey. sorry your conversation was just so interesting',\n",
              " 'smokeyrobinson you fell off my list...did I offend you',\n",
              " '18 weeks till sisters home.. i missed her call again its the worst feeling in the world.',\n",
              " 'gonna try to get some sleep in this hotel room',\n",
              " 'chimpytwit brilliant idea. just bring as much as you think youll spend amp Ill swap you. bring a brolly',\n",
              " 'SimoneKali Get me an autograph and shout out You have to record it though. My computer is dead so I cant listen.',\n",
              " 'rt KimKardashian Khloe just got fired from the Apprentice',\n",
              " 'in biology class. my lip hurts',\n",
              " 'Sorry for the next tweet...',\n",
              " 'Grr not down to go to school today',\n",
              " 'jennlopez I had to get an HD tivo and just got it set up tonight in order to get channels 40 59.',\n",
              " 'mouthsex i asked everyone how they were doing yesterday and not a single tweet back',\n",
              " 'adame ruddy money i like it still waiting here.... oh what a surprise... by state too hmmm....... seriously',\n",
              " 'KalPenn I am so sad Kutner was my favorite of the new team',\n",
              " 'MarcusMims wow i didnt get an quothelloquot u suck',\n",
              " 'bedtime school tomorrow and i still have no books being broke suckssss',\n",
              " 'Seating here helping my baby with his papers...well he is forcing me too seat with him im sleepy',\n",
              " 'Synching my contacts from my old mobile to iPhone. Import does not work well',\n",
              " 'i cant concentrate',\n",
              " 'Ive just spent 1 hour to enter all the bureaucratic nonsense for March. What a waste of my time',\n",
              " 'Nw more confused then ever',\n",
              " 'Not feeling well again. Stupid migraine making my tummy upset. and my whole body aches shoot me now.',\n",
              " 'Reading Buyology before bedtime... great premise but only turning out to be an quotOKquot book lots of info I already knew',\n",
              " 'Home I really wana sleep but due to wasting my free line in town I have an assignment to finish',\n",
              " 'omg... quotThe Readerquot is making me',\n",
              " 'cloudpimps doh At least youre getting a decent exchange rate at the moment sterling is still getting flogged',\n",
              " 'tried to download tweetdeck but it wont download',\n",
              " 'Theres an inch of snow on the ground and counting. Im worried about the poor flowers',\n",
              " 'dweeman why arent you a happy camper',\n",
              " 'dkoenigs thanks man. Im so very grateful. I feel unworthy of such attention though because Im in this because of myself...',\n",
              " 'twolfe i miss u too. im totally comin back tho Lastnight was sooooooo much fun',\n",
              " 'sniffinglue ohhh. I love it. ps Im sad we didnt get to hang out',\n",
              " 'And somehow I still end up in this place',\n",
              " 'kisluvkis oh that is very sad poor boy.',\n",
              " 'JonathanRKnight silvertulip27 Um that would be a hell no to the fugly poker dog pants on the cruise Hi Jonathan sorry I missed you',\n",
              " 'Im doing my homework. Its gosh darn hard',\n",
              " 'definitely no vacation for me... httpplurk.compmzygb',\n",
              " 'study group extraordinare about to leave campus...some of torts outline finishedgood thingbut sleepy as h...will be back 7am',\n",
              " 'Why cant airfare go down Or why do I have to live in Texas I dont know if Ill be able to make it to Iowa.',\n",
              " 'jstricko I found it pretty frustrating. stupid monkeys',\n",
              " 'lenesha but Im not feeling well mommy...',\n",
              " 'lovebscott Nope Im Right Along Wit You',\n",
              " 'so a murder gotcha. Cant believe it',\n",
              " 'nbensalem im sitting at my house and im sooo not looking forward to my one class tomorrow.',\n",
              " 'Ordered some maternity clothes online which came today. I got something strange I didnt order not in my size. And stuff is missing.',\n",
              " 'citizensheep',\n",
              " 'thesage1014 I wish I could Gotta work though',\n",
              " 'SeaGhostdesigns what happened to you on Saturday You didnt show up...',\n",
              " 'GirlsGoneChild',\n",
              " 'Cant fall asleep',\n",
              " ...]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YCIeqpDMSbWD"
      },
      "source": [
        "final_df['Updated_Text'] = new_list_1\n",
        "final_df['Updated_Text'] = final_df['Updated_Text'].str.lower()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jDozVY2sEcTe",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        },
        "outputId": "682ae60f-8794-460d-fe65-5f90e569bdaf"
      },
      "source": [
        "final_df"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Text</th>\n",
              "      <th>label</th>\n",
              "      <th>Updated_Text</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>@switchfoot http://twitpic.com/2y1zl - Awww, t...</td>\n",
              "      <td>1</td>\n",
              "      <td>switchfoot httptwitpic.com2y1zl awww thats a b...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>is upset that he can't update his Facebook by ...</td>\n",
              "      <td>1</td>\n",
              "      <td>is upset that he cant update his facebook by t...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>@Kenichan I dived many times for the ball. Man...</td>\n",
              "      <td>1</td>\n",
              "      <td>kenichan i dived many times for the ball. mana...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>my whole body feels itchy and like its on fire</td>\n",
              "      <td>0</td>\n",
              "      <td>my whole body feels itchy and like its on fire</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>@nationwideclass no, it's not behaving at all....</td>\n",
              "      <td>1</td>\n",
              "      <td>nationwideclass no its not behaving at all. im...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>278587</th>\n",
              "      <td>`   These sources don't exactly exude a sense ...</td>\n",
              "      <td>0</td>\n",
              "      <td>these sources dont exactly exude a sense of im...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>278588</th>\n",
              "      <td>The Institute for Historical Review is a pee...</td>\n",
              "      <td>0</td>\n",
              "      <td>the institute for historical review is a peerr...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>278589</th>\n",
              "      <td>:The way you're trying to describe it in this...</td>\n",
              "      <td>0</td>\n",
              "      <td>the way youre trying to describe it in this ar...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>278590</th>\n",
              "      <td>== Warning ==  There is clearly a protection...</td>\n",
              "      <td>0</td>\n",
              "      <td>warning there is clearly a protectionist regim...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>278591</th>\n",
              "      <td>Alternate option=== Is there perhaps enough ne...</td>\n",
              "      <td>0</td>\n",
              "      <td>alternate option is there perhaps enough newsw...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>278592 rows × 3 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                     Text  ...                                       Updated_Text\n",
              "0       @switchfoot http://twitpic.com/2y1zl - Awww, t...  ...  switchfoot httptwitpic.com2y1zl awww thats a b...\n",
              "1       is upset that he can't update his Facebook by ...  ...  is upset that he cant update his facebook by t...\n",
              "2       @Kenichan I dived many times for the ball. Man...  ...  kenichan i dived many times for the ball. mana...\n",
              "3         my whole body feels itchy and like its on fire   ...     my whole body feels itchy and like its on fire\n",
              "4       @nationwideclass no, it's not behaving at all....  ...  nationwideclass no its not behaving at all. im...\n",
              "...                                                   ...  ...                                                ...\n",
              "278587  `   These sources don't exactly exude a sense ...  ...  these sources dont exactly exude a sense of im...\n",
              "278588    The Institute for Historical Review is a pee...  ...  the institute for historical review is a peerr...\n",
              "278589   :The way you're trying to describe it in this...  ...  the way youre trying to describe it in this ar...\n",
              "278590    == Warning ==  There is clearly a protection...  ...  warning there is clearly a protectionist regim...\n",
              "278591  Alternate option=== Is there perhaps enough ne...  ...  alternate option is there perhaps enough newsw...\n",
              "\n",
              "[278592 rows x 3 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 13
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1y2utx00b-7I",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "7b7a2a12-fc11-441a-8af2-821c95ff8a17"
      },
      "source": [
        "final_df.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(278592, 3)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "U5RRi0q3o5gW"
      },
      "source": [
        "master_df = pd.DataFrame([final_df.Updated_Text, final_df.label]).transpose()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        },
        "id": "teSsJ5anpOmJ",
        "outputId": "6c2414f2-d2bf-4464-ce8c-685b94dc57d3"
      },
      "source": [
        "master_df = master_df.rename(columns={\"Updated_Text\": \"text\"})\n",
        "master_df"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>text</th>\n",
              "      <th>label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>switchfoot httptwitpic.com2y1zl awww thats a b...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>is upset that he cant update his facebook by t...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>kenichan i dived many times for the ball. mana...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>my whole body feels itchy and like its on fire</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>nationwideclass no its not behaving at all. im...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>278587</th>\n",
              "      <td>these sources dont exactly exude a sense of im...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>278588</th>\n",
              "      <td>the institute for historical review is a peerr...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>278589</th>\n",
              "      <td>the way youre trying to describe it in this ar...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>278590</th>\n",
              "      <td>warning there is clearly a protectionist regim...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>278591</th>\n",
              "      <td>alternate option is there perhaps enough newsw...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>278592 rows × 2 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                     text label\n",
              "0       switchfoot httptwitpic.com2y1zl awww thats a b...     1\n",
              "1       is upset that he cant update his facebook by t...     1\n",
              "2       kenichan i dived many times for the ball. mana...     1\n",
              "3          my whole body feels itchy and like its on fire     0\n",
              "4       nationwideclass no its not behaving at all. im...     1\n",
              "...                                                   ...   ...\n",
              "278587  these sources dont exactly exude a sense of im...     0\n",
              "278588  the institute for historical review is a peerr...     0\n",
              "278589  the way youre trying to describe it in this ar...     0\n",
              "278590  warning there is clearly a protectionist regim...     0\n",
              "278591  alternate option is there perhaps enough newsw...     0\n",
              "\n",
              "[278592 rows x 2 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 16
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0F19OCa_UM90"
      },
      "source": [
        "## Split into Training and Test"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Sux9ocH-TNR3"
      },
      "source": [
        "y = final_df.label\n",
        "x = final_df.Updated_Text"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qX_29m6WTcg7",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "75a39891-cc6f-4085-8e14-c988843458c2"
      },
      "source": [
        "train_data, test_data = train_test_split(master_df, test_size=0.2,\n",
        "                                         random_state=7) # Splits Dataset into Training and Testing set\n",
        "print(\"Train Data size:\", len(train_data))\n",
        "print(\"Test Data size\", len(test_data))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train Data size: 222873\n",
            "Test Data size 55719\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        },
        "id": "0XUkPY3wilnM",
        "outputId": "32222c8e-054a-4b2a-cd99-b3960b0fd526"
      },
      "source": [
        "train_data"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>text</th>\n",
              "      <th>label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>66626</th>\n",
              "      <td>bruciebabe maxblumenthal jews lived on 40 of t...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>114360</th>\n",
              "      <td>armed forces of the philippine commonwealth i ...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>276313</th>\n",
              "      <td>perhaps you should notify wpast. the thing is ...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>148602</th>\n",
              "      <td>excuse me pardon me but i was accused of spamm...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>51638</th>\n",
              "      <td>just woke up from my nap.</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>235075</th>\n",
              "      <td>you are a notorious troll and vandal too hrafn.</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10742</th>\n",
              "      <td>httptwitpic.com3ivnu walker granny w.short sho...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>49689</th>\n",
              "      <td>listen to music guilty pleasure</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>189636</th>\n",
              "      <td>nationality and flags ok people lets calm this...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>61615</th>\n",
              "      <td>axl roses voice is just so hot you have such b...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>222873 rows × 2 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                     text label\n",
              "66626   bruciebabe maxblumenthal jews lived on 40 of t...     1\n",
              "114360  armed forces of the philippine commonwealth i ...     0\n",
              "276313  perhaps you should notify wpast. the thing is ...     0\n",
              "148602  excuse me pardon me but i was accused of spamm...     0\n",
              "51638                           just woke up from my nap.     1\n",
              "...                                                   ...   ...\n",
              "235075    you are a notorious troll and vandal too hrafn.     1\n",
              "10742   httptwitpic.com3ivnu walker granny w.short sho...     1\n",
              "49689                     listen to music guilty pleasure     1\n",
              "189636  nationality and flags ok people lets calm this...     0\n",
              "61615   axl roses voice is just so hot you have such b...     0\n",
              "\n",
              "[222873 rows x 2 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 19
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "r_tlAi50ZnaI"
      },
      "source": [
        "# Vectorization of training dataset"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GLJ8PYICZbCK"
      },
      "source": [
        "from sklearn.feature_extraction.text import CountVectorizer\n",
        "count_vector = CountVectorizer(stop_words = 'english', lowercase = True)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "na3G92vSD4pc",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "f90518dc-bb5f-4072-8639-7b7efe69b27f"
      },
      "source": [
        "CountVectorizer()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "CountVectorizer(analyzer='word', binary=False, decode_error='strict',\n",
              "                dtype=<class 'numpy.int64'>, encoding='utf-8', input='content',\n",
              "                lowercase=True, max_df=1.0, max_features=None, min_df=1,\n",
              "                ngram_range=(1, 1), preprocessor=None, stop_words=None,\n",
              "                strip_accents=None, token_pattern='(?u)\\\\b\\\\w\\\\w+\\\\b',\n",
              "                tokenizer=None, vocabulary=None)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 64
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "v42_XpYMaqPJ"
      },
      "source": [
        "train_x = count_vector.fit_transform(x_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RZkvM5aVhEFN",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "8890585a-fe69-487c-d072-49cf7eaefe2d"
      },
      "source": [
        "train_x.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(158150, 253077)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 66
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eoACx4bx5bTO"
      },
      "source": [
        "val_x = count_vector.transform(x_val)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "yS9gEZoGGkiM",
        "outputId": "d8139a7f-ff54-4572-e62d-a61eccfba751"
      },
      "source": [
        "val_x"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<27909x253077 sparse matrix of type '<class 'numpy.int64'>'\n",
              "\twith 662959 stored elements in Compressed Sparse Row format>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 68
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Lbex0rSw5bEk",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "5abc09c4-6379-4d9a-d54e-7216b67d7750"
      },
      "source": [
        "val_x.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(27909, 253077)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 69
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aVm7Xpr_avQY"
      },
      "source": [
        "test_x = count_vector.transform(x_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "IYme1ECDbwe6",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "8cdd0508-8d6d-449c-f354-586044c5fcb4"
      },
      "source": [
        "test_x.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(32834, 253077)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 71
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "KVbcrbxrNrhv"
      },
      "source": [
        "# Machine Learning Implementation"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BeKpTOD-UdB_"
      },
      "source": [
        "## KNN Classifier"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WNeGvIvKUiwT"
      },
      "source": [
        "from sklearn.neighbors import KNeighborsClassifier"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cnRUJMQDV7r1",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "7c29ae84-371c-42c0-8dd2-64652f4fab1a"
      },
      "source": [
        "knn = KNeighborsClassifier(n_neighbors=10)\n",
        "knn.fit(train_x, y_train)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "KNeighborsClassifier(algorithm='auto', leaf_size=30, metric='minkowski',\n",
              "                     metric_params=None, n_jobs=None, n_neighbors=10, p=2,\n",
              "                     weights='uniform')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 73
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3Bh1F-O2WHiP",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 307
        },
        "outputId": "4149f247-8903-42a4-bba9-82f386f4bf48"
      },
      "source": [
        "y_pred_KNN = knn.predict(test_x)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "error",
          "ename": "KeyboardInterrupt",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-74-a5c0dc1c8ec4>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0my_pred_KNN\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mknn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpredict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtest_x\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/sklearn/neighbors/_classification.py\u001b[0m in \u001b[0;36mpredict\u001b[0;34m(self, X)\u001b[0m\n\u001b[1;32m    171\u001b[0m         \u001b[0mX\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcheck_array\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0maccept_sparse\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'csr'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    172\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 173\u001b[0;31m         \u001b[0mneigh_dist\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mneigh_ind\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mkneighbors\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    174\u001b[0m         \u001b[0mclasses_\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mclasses_\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    175\u001b[0m         \u001b[0m_y\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_y\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/sklearn/neighbors/_base.py\u001b[0m in \u001b[0;36mkneighbors\u001b[0;34m(self, X, n_neighbors, return_distance)\u001b[0m\n\u001b[1;32m    641\u001b[0m                 \u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_fit_X\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mreduce_func\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mreduce_func\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    642\u001b[0m                 \u001b[0mmetric\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0meffective_metric_\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mn_jobs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mn_jobs\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 643\u001b[0;31m                 **kwds))\n\u001b[0m\u001b[1;32m    644\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    645\u001b[0m         \u001b[0;32melif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_fit_method\u001b[0m \u001b[0;32min\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m'ball_tree'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'kd_tree'\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/sklearn/metrics/pairwise.py\u001b[0m in \u001b[0;36mpairwise_distances_chunked\u001b[0;34m(X, Y, reduce_func, metric, n_jobs, working_memory, **kwds)\u001b[0m\n\u001b[1;32m   1593\u001b[0m             \u001b[0mX_chunk\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mX\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0msl\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1594\u001b[0m         D_chunk = pairwise_distances(X_chunk, Y, metric=metric,\n\u001b[0;32m-> 1595\u001b[0;31m                                      n_jobs=n_jobs, **kwds)\n\u001b[0m\u001b[1;32m   1596\u001b[0m         if ((X is Y or Y is None)\n\u001b[1;32m   1597\u001b[0m                 \u001b[0;32mand\u001b[0m \u001b[0mPAIRWISE_DISTANCE_FUNCTIONS\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmetric\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/sklearn/metrics/pairwise.py\u001b[0m in \u001b[0;36mpairwise_distances\u001b[0;34m(X, Y, metric, n_jobs, force_all_finite, **kwds)\u001b[0m\n\u001b[1;32m   1750\u001b[0m         \u001b[0mfunc\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mpartial\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdistance\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcdist\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmetric\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mmetric\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwds\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1751\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1752\u001b[0;31m     \u001b[0;32mreturn\u001b[0m \u001b[0m_parallel_pairwise\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mY\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mn_jobs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwds\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1753\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1754\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/sklearn/metrics/pairwise.py\u001b[0m in \u001b[0;36m_parallel_pairwise\u001b[0;34m(X, Y, func, n_jobs, **kwds)\u001b[0m\n\u001b[1;32m   1346\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1347\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0meffective_n_jobs\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mn_jobs\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1348\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mY\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwds\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1349\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1350\u001b[0m     \u001b[0;31m# enforce a threading backend to prevent data communication overhead\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/sklearn/metrics/pairwise.py\u001b[0m in \u001b[0;36meuclidean_distances\u001b[0;34m(X, Y, Y_norm_squared, squared, X_norm_squared)\u001b[0m\n\u001b[1;32m    301\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    302\u001b[0m         \u001b[0;31m# if dtype is already float64, no need to chunk and upcast\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 303\u001b[0;31m         \u001b[0mdistances\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0;36m2\u001b[0m \u001b[0;34m*\u001b[0m \u001b[0msafe_sparse_dot\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mY\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mT\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdense_output\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    304\u001b[0m         \u001b[0mdistances\u001b[0m \u001b[0;34m+=\u001b[0m \u001b[0mXX\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    305\u001b[0m         \u001b[0mdistances\u001b[0m \u001b[0;34m+=\u001b[0m \u001b[0mYY\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/sklearn/utils/extmath.py\u001b[0m in \u001b[0;36msafe_sparse_dot\u001b[0;34m(a, b, dense_output)\u001b[0m\n\u001b[1;32m    149\u001b[0m             \u001b[0mret\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdot\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0ma\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mb\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    150\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 151\u001b[0;31m         \u001b[0mret\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0ma\u001b[0m \u001b[0;34m@\u001b[0m \u001b[0mb\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    152\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    153\u001b[0m     if (sparse.issparse(a) and sparse.issparse(b)\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/scipy/sparse/base.py\u001b[0m in \u001b[0;36m__matmul__\u001b[0;34m(self, other)\u001b[0m\n\u001b[1;32m    562\u001b[0m             raise ValueError(\"Scalar operands are not allowed, \"\n\u001b[1;32m    563\u001b[0m                              \"use '*' instead\")\n\u001b[0;32m--> 564\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__mul__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mother\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    565\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    566\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m__rmatmul__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mother\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/scipy/sparse/base.py\u001b[0m in \u001b[0;36m__mul__\u001b[0;34m(self, other)\u001b[0m\n\u001b[1;32m    482\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshape\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m!=\u001b[0m \u001b[0mother\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshape\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    483\u001b[0m                 \u001b[0;32mraise\u001b[0m \u001b[0mValueError\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'dimension mismatch'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 484\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_mul_sparse_matrix\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mother\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    485\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    486\u001b[0m         \u001b[0;31m# If it's a list or whatever, treat it like a matrix\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/scipy/sparse/compressed.py\u001b[0m in \u001b[0;36m_mul_sparse_matrix\u001b[0;34m(self, other)\u001b[0m\n\u001b[1;32m    524\u001b[0m            \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0masarray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mother\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mindices\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0midx_dtype\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    525\u001b[0m            \u001b[0mother\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 526\u001b[0;31m            indptr, indices, data)\n\u001b[0m\u001b[1;32m    527\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    528\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__class__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mindices\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mindptr\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mshape\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mM\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mN\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lJIZIrgIiHAl"
      },
      "source": [
        "print('F1 Score:', metrics.f1_score(y_test, y_pred_KNN))\n",
        "print('Accuracy Score:', metrics.accuracy_score(y_test, y_pred_KNN))\n",
        "print('Precision score: ', metrics.precision_score(y_test, y_pred_KNN))\n",
        "print('Recall score: ', metrics.recall_score(y_test, y_pred_KNN))\n",
        "print('Confusion Matrix:', metrics.confusion_matrix(y_test, y_pred_KNN))\n",
        "print('Mean Absolute Error:', metrics.mean_absolute_error(y_test, y_pred_KNN))\n",
        "print('Mean Squared Error:', metrics.mean_squared_error(y_test, y_pred_KNN))\n",
        "print('Root Mean Squared Error:', np.sqrt(metrics.mean_squared_error(y_test, y_pred_KNN)))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JE4VJHfwdxg8"
      },
      "source": [
        "## Adaboost "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "n84n7N12d5PO"
      },
      "source": [
        "from sklearn.ensemble import AdaBoostClassifier"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "R9sLjY7DeDJg"
      },
      "source": [
        "ada = AdaBoostClassifier()\n",
        "ada.fit(train_x, y_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aIXWYOt4eXsl"
      },
      "source": [
        "y_pred_ada = ada.predict(test_x)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qm5LlDRohfMS"
      },
      "source": [
        "print('F1 Score:', metrics.f1_score(y_test, y_pred_ada))\n",
        "print('Accuracy Score:', metrics.accuracy_score(y_test, y_pred_ada))\n",
        "print('Precision score: ', metrics.precision_score(y_test, y_pred_ada))\n",
        "print('Recall score: ', metrics.recall_score(y_test, y_pred_ada))\n",
        "print('Confusion Matrix:', metrics.confusion_matrix(y_test, y_pred_ada))\n",
        "print('Mean Absolute Error:', metrics.mean_absolute_error(y_test, y_pred_ada))\n",
        "print('Mean Squared Error:', metrics.mean_squared_error(y_test, y_pred_ada))\n",
        "print('Root Mean Squared Error:', np.sqrt(metrics.mean_squared_error(y_test, y_pred_ada)))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ga8ZpWGuoRzL"
      },
      "source": [
        "plt.scatter(y_test, y_pred_ada)\n",
        "plt.xlabel('Actual Score')\n",
        "plt.ylabel('Predicted Score')\n",
        "plt.title('Actual vs Predicted Score')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JtdYhRMEjd52"
      },
      "source": [
        "## Support Vector Machine"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "H7lHUb6_jf68"
      },
      "source": [
        "from sklearn import svm"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LhkKjYcog0bN"
      },
      "source": [
        "svm_model = svm.SVC()\n",
        "svm_model.fit(train_x, y_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fE7OeoNvhI0V"
      },
      "source": [
        "y_pred_svm = svm_model.predict(test_x)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "o72vG3BL4LD8"
      },
      "source": [
        "print('F1 Score:', metrics.f1_score(y_test, y_pred_svm))\n",
        "print('Accuracy Score:', metrics.accuracy_score(y_test, y_pred_svm))\n",
        "print('Precision score: ', metrics.precision_score(y_test, y_pred_svm))\n",
        "print('Recall score: ', metrics.recall_score(y_test, y_pred_svm))\n",
        "print('Confusion Matrix:', metrics.confusion_matrix(y_test, y_pred_svm))\n",
        "print('Mean Absolute Error:', metrics.mean_absolute_error(y_test, y_pred_svm))\n",
        "print('Mean Squared Error:', metrics.mean_squared_error(y_test, y_pred_svm))\n",
        "print('Root Mean Squared Error:', np.sqrt(metrics.mean_squared_error(y_test, y_pred_svm)))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9TbFadT1RQQ6"
      },
      "source": [
        "##Bagging Classifier"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VTR7ftUMRxJ9"
      },
      "source": [
        "from sklearn.ensemble import BaggingClassifier\n",
        "\n",
        "BGC = BaggingClassifier()\n",
        "BGC.fit(train_x,y_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cPiyrtoh90EW"
      },
      "source": [
        "y_pred_bgc = BGC.predict(test_x)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RLfQdFND958E"
      },
      "source": [
        "model = 'Bagging Classifier'\n",
        "print('Accuracy for {} is {}'.format(model,round(accuracy_score(y_pred_bgc,y_test),3)))\n",
        "print('F1 Score for {} is {}'.format(model,round(f1_score(y_pred_bgc,y_test),3)))\n",
        "print('Precision Score for {} is {}'.format(model,round(precision_score(y_pred_bgc,y_test),3)))\n",
        "print('Recall Score for {} is {}'.format(model,round(recall_score(y_pred_bgc,y_test),3)))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "foV4LCLe_JkK"
      },
      "source": [
        "##Logistic Regression"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hpAE3NFo_Xs1"
      },
      "source": [
        "from sklearn.linear_model import LogisticRegression\n",
        "#LogisticRegression\n",
        "LR = LogisticRegression()\n",
        "LR.fit(train_x,y_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Hws9z1S__abN"
      },
      "source": [
        "y_pred_lr = LR.predict(test_x)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QEK0QuXG_dzX"
      },
      "source": [
        "model = 'Logistic Regression'\n",
        "print('Accuracy for {} is {}'.format(model,round(accuracy_score(y_pred_lr,y_test),3)))\n",
        "print('F1 Score for {} is {}'.format(model,round(f1_score(y_pred_lr,y_test),3)))\n",
        "print('Precision Score for {} is {}'.format(model,round(precision_score(y_pred_lr,y_test),3)))\n",
        "print('Recall Score for {} is {}'.format(model,round(recall_score(y_pred_lr,y_test),3)))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-8Skq2wt_jfE"
      },
      "source": [
        "##Naive Bayes"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "85TZlJWz_lNm"
      },
      "source": [
        "from sklearn.naive_bayes import MultinomialNB\n",
        "#Multinomial NB\n",
        "MNB = MultinomialNB()\n",
        "MNB.fit(train_x,y_train)\n",
        "y_pred = MNB.predict(test_x)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WgjEKffk_ma7"
      },
      "source": [
        "model = 'Naive Bayes'\n",
        "print('Accuracy for {} is {}'.format(model,round(accuracy_score(y_pred,y_test),3)))\n",
        "print('F1 Score for {} is {}'.format(model,round(f1_score(y_pred,y_test),3)))\n",
        "print('Precision Score for {} is {}'.format(model,round(precision_score(y_pred,y_test),3)))\n",
        "print('Recall Score for {} is {}'.format(model,round(recall_score(y_pred,y_test),3)))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "faIBQ8apR1LG"
      },
      "source": [
        "# PreProcessing for Neural network"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "INM28eJOSwJV"
      },
      "source": [
        "from keras.preprocessing.text import Tokenizer\n",
        "from keras.preprocessing.sequence import pad_sequences\n",
        "from keras.models import Sequential\n",
        "from keras import layers\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.metrics import confusion_matrix\n",
        "import pandas as pd\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.preprocessing import LabelEncoder"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xn7H6lTHi_nY"
      },
      "source": [
        "MAXLENGTH = 30"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        },
        "id": "itpg3h8TTtyc",
        "outputId": "ec2591ff-7612-4890-a492-807eb8e14025"
      },
      "source": [
        "train_data"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>text</th>\n",
              "      <th>label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>66626</th>\n",
              "      <td>bruciebabe maxblumenthal jews lived on 40 of t...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>114360</th>\n",
              "      <td>armed forces of the philippine commonwealth i ...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>276313</th>\n",
              "      <td>perhaps you should notify wpast. the thing is ...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>148602</th>\n",
              "      <td>excuse me pardon me but i was accused of spamm...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>51638</th>\n",
              "      <td>just woke up from my nap.</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>235075</th>\n",
              "      <td>you are a notorious troll and vandal too hrafn.</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10742</th>\n",
              "      <td>httptwitpic.com3ivnu walker granny w.short sho...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>49689</th>\n",
              "      <td>listen to music guilty pleasure</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>189636</th>\n",
              "      <td>nationality and flags ok people lets calm this...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>61615</th>\n",
              "      <td>axl roses voice is just so hot you have such b...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>222873 rows × 2 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                     text label\n",
              "66626   bruciebabe maxblumenthal jews lived on 40 of t...     1\n",
              "114360  armed forces of the philippine commonwealth i ...     0\n",
              "276313  perhaps you should notify wpast. the thing is ...     0\n",
              "148602  excuse me pardon me but i was accused of spamm...     0\n",
              "51638                           just woke up from my nap.     1\n",
              "...                                                   ...   ...\n",
              "235075    you are a notorious troll and vandal too hrafn.     1\n",
              "10742   httptwitpic.com3ivnu walker granny w.short sho...     1\n",
              "49689                     listen to music guilty pleasure     1\n",
              "189636  nationality and flags ok people lets calm this...     0\n",
              "61615   axl roses voice is just so hot you have such b...     0\n",
              "\n",
              "[222873 rows x 2 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 24
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-S7_XHkqrtgG"
      },
      "source": [
        "## Tokenizer\n",
        "\n",
        "Creating Token sequences for train and test data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "8yrnmR2KUQx9",
        "outputId": "f8b42d92-f2d0-4f3b-96b6-c9eb693bcc7c"
      },
      "source": [
        "tokenizer = Tokenizer()\n",
        "tokenizer.fit_on_texts(train_data.text)\n",
        "\n",
        "word_index = tokenizer.word_index\n",
        "vocab_size = len(tokenizer.word_index) + 1\n",
        "print(\"Vocabulary Size :\", vocab_size)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Vocabulary Size : 305130\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "pGLAslp0U05s",
        "outputId": "5ea475bd-cbaf-49b6-df9d-339b398a1f7e"
      },
      "source": [
        "x_train = pad_sequences(tokenizer.texts_to_sequences(train_data.text),\n",
        "                        maxlen = MAXLENGTH)\n",
        "x_test = pad_sequences(tokenizer.texts_to_sequences(test_data.text),\n",
        "                       maxlen = MAXLENGTH)\n",
        "\n",
        "print(\"Training Shape:\",x_train.shape)\n",
        "print(\"Testing Shape:\",x_test.shape)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Training Shape: (222873, 30)\n",
            "Testing Shape: (55719, 30)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "WLPm72eCYJ8U",
        "outputId": "dea2297f-8909-4a3d-efe4-024b4c94a882"
      },
      "source": [
        "labels = train_data.label.unique().tolist()\n",
        "print(labels)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[1, 0]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "gc3SiQz1qpcv",
        "outputId": "23eed34f-f2b5-4078-f57f-459915ea762e"
      },
      "source": [
        "y_train = train_data.label\n",
        "y_train.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(222873,)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 28
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "uQn9-HMTrQLH",
        "outputId": "4da57b02-25f1-4c6c-ef42-a862c4b2a61a"
      },
      "source": [
        "encoder = LabelEncoder()\n",
        "encoder.fit(train_data.label.to_list())\n",
        "\n",
        "y_train = encoder.transform(train_data.label.to_list())\n",
        "y_test = encoder.transform(test_data.label.to_list())\n",
        "\n",
        "y_train = y_train.reshape(-1,1)\n",
        "y_test = y_test.reshape(-1,1)\n",
        "\n",
        "print(\"y_train shape:\", y_train.shape)\n",
        "print(\"y_test shape:\", y_test.shape)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "y_train shape: (222873, 1)\n",
            "y_test shape: (55719, 1)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tL00k8DgsLJg"
      },
      "source": [
        "## Glove Embeddings\n",
        "\n",
        "Glove Embeddings to identify word contexts"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YYJKWE8wsSDV"
      },
      "source": [
        "#!wget http://nlp.stanford.edu/data/glove.6B.zip\n",
        "#!unzip glove.6B.zip"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Urb4QEVutS3-",
        "outputId": "30ea0f8c-a989-4e11-acfd-c69f96c16308"
      },
      "source": [
        "embeddings_index = {}\n",
        "GLOVE_PATH = './glove.6B.300d.txt'\n",
        "f = open(GLOVE_PATH)\n",
        "for line in f:\n",
        "  values = line.split()\n",
        "  word = value = values[0]\n",
        "  coefs = np.asarray(values[1:], dtype='float32')\n",
        "  embeddings_index[word] = coefs\n",
        "f.close()\n",
        "\n",
        "print('Found %s word vectors.' %len(embeddings_index))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 400000 word vectors.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wq8Pz1TKt013"
      },
      "source": [
        "EMBEDDING_DIM = 300\n",
        "LR = 1e-3\n",
        "BATCH_SIZE = 1024\n",
        "EPOCHS = 15\n",
        "\n",
        "embedding_matrix = np.zeros((vocab_size, EMBEDDING_DIM))\n",
        "for word, i in word_index.items():\n",
        "  embedding_vector = embeddings_index.get(word)\n",
        "  if embedding_vector is not None:\n",
        "    embedding_matrix[i] = embedding_vector\n",
        "embedding_layer = layers.Embedding(vocab_size,\n",
        "                                          EMBEDDING_DIM,\n",
        "                                          weights=[embedding_matrix],\n",
        "                                          input_length=MAXLENGTH,\n",
        "                                          trainable=False)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "TScsqhIQyxiP"
      },
      "source": [
        "#Helper Functions for NN"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "V5hjtcHdy060"
      },
      "source": [
        "## Function to plot the loss and accuracy\n",
        "def plot_history(history):\n",
        "\t  # plot loss\n",
        "    plt.title('Loss')\n",
        "    plt.plot(history.history['loss'], color='blue', label='train')\n",
        "    plt.plot(history.history['val_loss'], color='red', label='test')\n",
        "    plt.ylabel('Loss')\n",
        "    plt.xlabel('Epoch')\n",
        "    plt.legend(['Train', 'Test'])\n",
        "    plt.show()\n",
        "    \n",
        "    # plot accuracy\n",
        "    plt.title('Accuracy')\n",
        "    plt.plot(history.history['accuracy'], color='blue', label='train')\n",
        "    plt.plot(history.history['val_accuracy'], color='red', label='test')\n",
        "    plt.ylabel('Accuracy')\n",
        "    plt.xlabel('Epoch')\n",
        "    plt.legend(['Train', 'Test'])\n",
        "    plt.show()\n",
        "\n",
        "    #plot F1\n",
        "    plt.title('F1 score')\n",
        "    plt.plot(history.history['f1_m'], color='blue', label='train')\n",
        "    plt.plot(history.history['val_f1_m'], color='red', label='test')\n",
        "    plt.ylabel('F1 score')\n",
        "    plt.xlabel('Epoch')\n",
        "    plt.legend(['Train', 'Test'])\n",
        "    plt.show()\n",
        "\n",
        "    #plot Recall\n",
        "    plt.title('Recall')\n",
        "    plt.plot(history.history['recall'], color='blue', label='train')\n",
        "    plt.plot(history.history['val_recall'], color='red', label='test')\n",
        "    plt.ylabel('Recall')\n",
        "    plt.xlabel('Epoch')\n",
        "    plt.legend(['Train', 'Test'])\n",
        "    plt.show()\n",
        "\n",
        "\n",
        "    "
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uEFR-Nzy5IDt"
      },
      "source": [
        "# LSTM"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dNN4QGIjRvwh"
      },
      "source": [
        "import keras\n",
        "from sklearn.metrics import log_loss, roc_auc_score, accuracy_score\n",
        "from keras.losses import binary_crossentropy\n",
        "from keras.metrics import binary_accuracy\n",
        "from keras import backend as K\n",
        "from keras.callbacks import *\n",
        "\n",
        "def recall_m(y_true, y_pred):\n",
        "    true_positives = K.sum(K.round(K.clip(y_true * y_pred, 0, 1)))\n",
        "    possible_positives = K.sum(K.round(K.clip(y_true, 0, 1)))\n",
        "    recall = true_positives / (possible_positives + K.epsilon())\n",
        "    return recall\n",
        "\n",
        "def precision_m(y_true, y_pred):\n",
        "    true_positives = K.sum(K.round(K.clip(y_true * y_pred, 0, 1)))\n",
        "    predicted_positives = K.sum(K.round(K.clip(y_pred, 0, 1)))\n",
        "    precision = true_positives / (predicted_positives + K.epsilon())\n",
        "    return precision\n",
        "    \n",
        "def f1_m(y_true, y_pred):\n",
        "    precision = precision_m(y_true, y_pred)\n",
        "    recall = recall_m(y_true, y_pred)\n",
        "    return 2*((precision*recall)/(precision+recall+K.epsilon()))\n",
        "\n",
        "\n",
        "METRICS = [\n",
        "      keras.metrics.TruePositives(name='tp'),\n",
        "      keras.metrics.FalsePositives(name='fp'),\n",
        "      keras.metrics.TrueNegatives(name='tn'),\n",
        "      keras.metrics.FalseNegatives(name='fn'), \n",
        "      keras.metrics.BinaryAccuracy(name='accuracy'),\n",
        "      keras.metrics.Precision(name='precision'),\n",
        "      keras.metrics.Recall(name='recall'),\n",
        "      keras.metrics.AUC(name='auc'),\n",
        "      keras.metrics.AUC(name='prc', curve='PR'), # precision-recall curve\n",
        "      f1_m\n",
        "]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_weW3nFC5K6n",
        "outputId": "b675d5d0-d92c-4477-9f53-08cd20f4961f"
      },
      "source": [
        "sequence_input = Input(shape=(MAXLENGTH,), dtype='int32')\n",
        "embedding_sequences = embedding_layer(sequence_input)\n",
        "x = SpatialDropout1D(0.2)(embedding_sequences)\n",
        "x = LSTM(64, dropout=0.2, recurrent_dropout=0.2)(x)\n",
        "x = Dense(512, activation='relu')(x)\n",
        "x = Dropout(0.2)(x)\n",
        "x = Dense(512, activation='relu')(x)\n",
        "outputs = Dense(1, activation='sigmoid')(x)\n",
        "lstm_model = tf.keras.Model(sequence_input, outputs)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Layer lstm_9 will not use cuDNN kernel since it doesn't meet the cuDNN kernel criteria. It will use generic GPU kernel as fallback when running on GPU\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uoag0BCu5o5w"
      },
      "source": [
        "lstm_model.compile(optimizer=Adam(learning_rate=0.001), loss='binary_crossentropy',\n",
        "              metrics=[METRICS])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "HY5P4U-V1hxo",
        "outputId": "98087c24-4cd2-4570-c224-d408fd9eae6a"
      },
      "source": [
        "lstm_model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"model_7\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_8 (InputLayer)         [(None, 30)]              0         \n",
            "_________________________________________________________________\n",
            "embedding (Embedding)        (None, 30, 300)           91539000  \n",
            "_________________________________________________________________\n",
            "spatial_dropout1d_7 (Spatial (None, 30, 300)           0         \n",
            "_________________________________________________________________\n",
            "lstm_9 (LSTM)                (None, 64)                93440     \n",
            "_________________________________________________________________\n",
            "dense_21 (Dense)             (None, 512)               33280     \n",
            "_________________________________________________________________\n",
            "dropout_7 (Dropout)          (None, 512)               0         \n",
            "_________________________________________________________________\n",
            "dense_22 (Dense)             (None, 512)               262656    \n",
            "_________________________________________________________________\n",
            "dense_23 (Dense)             (None, 1)                 513       \n",
            "=================================================================\n",
            "Total params: 91,928,889\n",
            "Trainable params: 389,889\n",
            "Non-trainable params: 91,539,000\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "VAB0R3UX5uJA",
        "outputId": "8a876097-bde3-458e-84d8-b53bfba8517e"
      },
      "source": [
        "lstm_history = lstm_model.fit(x_train, y_train, batch_size=512, epochs=15,\n",
        "                    validation_data=(x_test, y_test))\n",
        "                  "
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/15\n",
            "436/436 [==============================] - 57s 131ms/step - loss: 0.3405 - tp: 42151.0000 - fp: 17952.0000 - tn: 145001.0000 - fn: 17769.0000 - accuracy: 0.8397 - precision: 0.7013 - recall: 0.7035 - auc: 0.9039 - prc: 0.7475 - f1_m: 0.7013 - val_loss: 0.2999 - val_tp: 12590.0000 - val_fp: 5588.0000 - val_tn: 35228.0000 - val_fn: 2313.0000 - val_accuracy: 0.8582 - val_precision: 0.6926 - val_recall: 0.8448 - val_auc: 0.9315 - val_prc: 0.8157 - val_f1_m: 0.7604\n",
            "Epoch 2/15\n",
            "436/436 [==============================] - 56s 128ms/step - loss: 0.2971 - tp: 44555.0000 - fp: 14923.0000 - tn: 148030.0000 - fn: 15365.0000 - accuracy: 0.8641 - precision: 0.7491 - recall: 0.7436 - auc: 0.9285 - prc: 0.8152 - f1_m: 0.7453 - val_loss: 0.2751 - val_tp: 12530.0000 - val_fp: 4642.0000 - val_tn: 36174.0000 - val_fn: 2373.0000 - val_accuracy: 0.8741 - val_precision: 0.7297 - val_recall: 0.8408 - val_auc: 0.9424 - val_prc: 0.8487 - val_f1_m: 0.7805\n",
            "Epoch 3/15\n",
            "436/436 [==============================] - 56s 128ms/step - loss: 0.2792 - tp: 45478.0000 - fp: 13540.0000 - tn: 149413.0000 - fn: 14442.0000 - accuracy: 0.8744 - precision: 0.7706 - recall: 0.7590 - auc: 0.9374 - prc: 0.8390 - f1_m: 0.7637 - val_loss: 0.2582 - val_tp: 12066.0000 - val_fp: 3592.0000 - val_tn: 37224.0000 - val_fn: 2837.0000 - val_accuracy: 0.8846 - val_precision: 0.7706 - val_recall: 0.8096 - val_auc: 0.9472 - val_prc: 0.8647 - val_f1_m: 0.7890\n",
            "Epoch 4/15\n",
            "436/436 [==============================] - 56s 129ms/step - loss: 0.2666 - tp: 46088.0000 - fp: 12598.0000 - tn: 150355.0000 - fn: 13832.0000 - accuracy: 0.8814 - precision: 0.7853 - recall: 0.7692 - auc: 0.9432 - prc: 0.8544 - f1_m: 0.7764 - val_loss: 0.2576 - val_tp: 12289.0000 - val_fp: 3780.0000 - val_tn: 37036.0000 - val_fn: 2614.0000 - val_accuracy: 0.8852 - val_precision: 0.7648 - val_recall: 0.8246 - val_auc: 0.9490 - val_prc: 0.8709 - val_f1_m: 0.7928\n",
            "Epoch 5/15\n",
            "436/436 [==============================] - 56s 128ms/step - loss: 0.2588 - tp: 46619.0000 - fp: 12232.0000 - tn: 150721.0000 - fn: 13301.0000 - accuracy: 0.8854 - precision: 0.7922 - recall: 0.7780 - auc: 0.9466 - prc: 0.8641 - f1_m: 0.7840 - val_loss: 0.2493 - val_tp: 12463.0000 - val_fp: 3814.0000 - val_tn: 37002.0000 - val_fn: 2440.0000 - val_accuracy: 0.8878 - val_precision: 0.7657 - val_recall: 0.8363 - val_auc: 0.9516 - val_prc: 0.8769 - val_f1_m: 0.7988\n",
            "Epoch 6/15\n",
            "436/436 [==============================] - 55s 127ms/step - loss: 0.2507 - tp: 47191.0000 - fp: 11794.0000 - tn: 151159.0000 - fn: 12729.0000 - accuracy: 0.8900 - precision: 0.8001 - recall: 0.7876 - auc: 0.9500 - prc: 0.8730 - f1_m: 0.7932 - val_loss: 0.2413 - val_tp: 12291.0000 - val_fp: 3324.0000 - val_tn: 37492.0000 - val_fn: 2612.0000 - val_accuracy: 0.8935 - val_precision: 0.7871 - val_recall: 0.8247 - val_auc: 0.9539 - val_prc: 0.8831 - val_f1_m: 0.8049\n",
            "Epoch 7/15\n",
            "436/436 [==============================] - 56s 128ms/step - loss: 0.2454 - tp: 47657.0000 - fp: 11687.0000 - tn: 151266.0000 - fn: 12263.0000 - accuracy: 0.8925 - precision: 0.8031 - recall: 0.7953 - auc: 0.9521 - prc: 0.8780 - f1_m: 0.7986 - val_loss: 0.2503 - val_tp: 12769.0000 - val_fp: 3979.0000 - val_tn: 36837.0000 - val_fn: 2134.0000 - val_accuracy: 0.8903 - val_precision: 0.7624 - val_recall: 0.8568 - val_auc: 0.9538 - val_prc: 0.8833 - val_f1_m: 0.8063\n",
            "Epoch 8/15\n",
            "436/436 [==============================] - 55s 127ms/step - loss: 0.2402 - tp: 47832.0000 - fp: 11315.0000 - tn: 151638.0000 - fn: 12088.0000 - accuracy: 0.8950 - precision: 0.8087 - recall: 0.7983 - auc: 0.9541 - prc: 0.8840 - f1_m: 0.8028 - val_loss: 0.2378 - val_tp: 12371.0000 - val_fp: 3323.0000 - val_tn: 37493.0000 - val_fn: 2532.0000 - val_accuracy: 0.8949 - val_precision: 0.7883 - val_recall: 0.8301 - val_auc: 0.9552 - val_prc: 0.8882 - val_f1_m: 0.8082\n",
            "Epoch 9/15\n",
            "436/436 [==============================] - 56s 128ms/step - loss: 0.2346 - tp: 48157.0000 - fp: 10974.0000 - tn: 151979.0000 - fn: 11763.0000 - accuracy: 0.8980 - precision: 0.8144 - recall: 0.8037 - auc: 0.9563 - prc: 0.8898 - f1_m: 0.8085 - val_loss: 0.2386 - val_tp: 12426.0000 - val_fp: 3252.0000 - val_tn: 37564.0000 - val_fn: 2477.0000 - val_accuracy: 0.8972 - val_precision: 0.7926 - val_recall: 0.8338 - val_auc: 0.9555 - val_prc: 0.8890 - val_f1_m: 0.8121\n",
            "Epoch 10/15\n",
            "436/436 [==============================] - 56s 128ms/step - loss: 0.2312 - tp: 48423.0000 - fp: 10812.0000 - tn: 152141.0000 - fn: 11497.0000 - accuracy: 0.8999 - precision: 0.8175 - recall: 0.8081 - auc: 0.9575 - prc: 0.8937 - f1_m: 0.8121 - val_loss: 0.2337 - val_tp: 12233.0000 - val_fp: 2967.0000 - val_tn: 37849.0000 - val_fn: 2670.0000 - val_accuracy: 0.8988 - val_precision: 0.8048 - val_recall: 0.8208 - val_auc: 0.9567 - val_prc: 0.8926 - val_f1_m: 0.8122\n",
            "Epoch 11/15\n",
            "436/436 [==============================] - 55s 127ms/step - loss: 0.2283 - tp: 48565.0000 - fp: 10661.0000 - tn: 152292.0000 - fn: 11355.0000 - accuracy: 0.9012 - precision: 0.8200 - recall: 0.8105 - auc: 0.9586 - prc: 0.8966 - f1_m: 0.8143 - val_loss: 0.2368 - val_tp: 12444.0000 - val_fp: 3210.0000 - val_tn: 37606.0000 - val_fn: 2459.0000 - val_accuracy: 0.8983 - val_precision: 0.7949 - val_recall: 0.8350 - val_auc: 0.9567 - val_prc: 0.8928 - val_f1_m: 0.8139\n",
            "Epoch 12/15\n",
            "436/436 [==============================] - 56s 128ms/step - loss: 0.2258 - tp: 48404.0000 - fp: 10392.0000 - tn: 152561.0000 - fn: 11516.0000 - accuracy: 0.9017 - precision: 0.8233 - recall: 0.8078 - auc: 0.9595 - prc: 0.8988 - f1_m: 0.8148 - val_loss: 0.2373 - val_tp: 12253.0000 - val_fp: 2998.0000 - val_tn: 37818.0000 - val_fn: 2650.0000 - val_accuracy: 0.8986 - val_precision: 0.8034 - val_recall: 0.8222 - val_auc: 0.9563 - val_prc: 0.8921 - val_f1_m: 0.8122\n",
            "Epoch 13/15\n",
            "436/436 [==============================] - 55s 127ms/step - loss: 0.2223 - tp: 48803.0000 - fp: 10254.0000 - tn: 152699.0000 - fn: 11117.0000 - accuracy: 0.9041 - precision: 0.8264 - recall: 0.8145 - auc: 0.9608 - prc: 0.9019 - f1_m: 0.8197 - val_loss: 0.2361 - val_tp: 12138.0000 - val_fp: 2826.0000 - val_tn: 37990.0000 - val_fn: 2765.0000 - val_accuracy: 0.8997 - val_precision: 0.8111 - val_recall: 0.8145 - val_auc: 0.9567 - val_prc: 0.8928 - val_f1_m: 0.8123\n",
            "Epoch 14/15\n",
            "436/436 [==============================] - 55s 126ms/step - loss: 0.2205 - tp: 48779.0000 - fp: 10045.0000 - tn: 152908.0000 - fn: 11141.0000 - accuracy: 0.9049 - precision: 0.8292 - recall: 0.8141 - auc: 0.9614 - prc: 0.9041 - f1_m: 0.8209 - val_loss: 0.2361 - val_tp: 12581.0000 - val_fp: 3307.0000 - val_tn: 37509.0000 - val_fn: 2322.0000 - val_accuracy: 0.8990 - val_precision: 0.7919 - val_recall: 0.8442 - val_auc: 0.9577 - val_prc: 0.8948 - val_f1_m: 0.8166\n",
            "Epoch 15/15\n",
            "436/436 [==============================] - 55s 126ms/step - loss: 0.2175 - tp: 48967.0000 - fp: 9937.0000 - tn: 153016.0000 - fn: 10953.0000 - accuracy: 0.9063 - precision: 0.8313 - recall: 0.8172 - auc: 0.9624 - prc: 0.9068 - f1_m: 0.8239 - val_loss: 0.2335 - val_tp: 12518.0000 - val_fp: 3206.0000 - val_tn: 37610.0000 - val_fn: 2385.0000 - val_accuracy: 0.8997 - val_precision: 0.7961 - val_recall: 0.8400 - val_auc: 0.9577 - val_prc: 0.8946 - val_f1_m: 0.8169\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dhY5cQAXA3MP"
      },
      "source": [
        "Plot for first 15 epochs"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "mZCl4HmF6Gtr",
        "outputId": "ddd5fcce-e2a0-454a-8b90-465db340db4b"
      },
      "source": [
        "plot_history(lstm_history)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEWCAYAAAB8LwAVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3dd5iU5dXH8e9hF+mIAqKyShFEVmmyUgwKKEasbIotNpL4Go01GLEniCW2RGPE2GLsYo+9i4oi6qJIR1CJLCJFpUXp5/3jnnGHZXbZMs8+s8vvc11z7cxTzyY4Z5+7nNvcHRERkdLqxR2AiIhkJyUIERFJSwlCRETSUoIQEZG0lCBERCQtJQgREUlLCUJERNJSghCpAjObZ2ZD4o5DJEpKECIikpYShEiGmFkDM7vJzL5KvG4yswaJfa3M7DkzW2Zm35rZeDOrl9h3gZktMLOVZjbbzA6M9zcRCXLjDkCkDrkE6Af0BBx4GrgUuAw4DygGWieO7Qe4mXUBzgT2cfevzKw9kFOzYYukpycIkcw5Hhjt7ovdfQlwOXBiYt86YCegnbuvc/fxHgqhbQAaAPlmVt/d57n7Z7FEL1KKEoRI5uwM/Dfl838T2wCuB+YCr5jZ52Z2IYC7zwXOBUYBi81srJntjEgWUIIQyZyvgHYpn3dNbMPdV7r7ee7eETgSGJHsa3D3h9x9QOJcB66t2bBF0lOCEKm6+mbWMPkCHgYuNbPWZtYK+BPwAICZHW5mnczMgOWEpqWNZtbFzA5IdGavBn4ANsbz64hsSglCpOpeIHyhJ18NgSJgCjAV+Ai4MnFsZ+A1YBXwHnCru48j9D9cAywFvgZ2AC6quV9BpGymBYNERCQdPUGIiEhaShAiIpKWEoSIiKSlBCEiImnVmVIbrVq18vbt28cdhohIrTJp0qSl7t463b46kyDat29PUVFR3GGIiNQqZvbfsvapiUlERNJSghARkbSUIEREJK060wchIlJZ69ato7i4mNWrV8cdSuQaNmxIXl4e9evXr/A5ShAistUqLi6mWbNmtG/fnlBHsW5yd7755huKi4vp0KFDhc9TE5OIbLVWr15Ny5Yt63RyADAzWrZsWeknpUgThJkNTayxOze5QEqp/aeZ2VQzm2xm75hZfqn9u5rZKjP7Y5RxisjWq64nh6Sq/J6RJQgzywHGAIcA+cBxpRMA8JC7d3P3nsB1wN9K7f8b8GJUMQJ8+y2MHg2ffBLlXUREap8onyD6AHPd/XN3XwuMBYalHuDuK1I+NiGspgWAmRUCXwDTI4yRevXgiivg4YejvIuIyOa++eYbevbsSc+ePdlxxx1p27btj5/Xrl1b7rlFRUWcffbZkcYXZSd1W2B+yudioG/pg8zsDGAEsA1wQGJbU+AC4CCgzOYlMzsVOBVg1113rVKQLVrA4MHwn//ANddU6RIiIlXSsmVLJk+eDMCoUaNo2rQpf/xjyVfe+vXryc1N/zVdUFBAQUFBpPHF3knt7mPcfTdCQrg0sXkUcKO7r9rCuXe4e4G7F7RunbaUSIUUFsLs2TBrVpUvISKSEcOHD+e0006jb9++jBw5kg8++ID+/fvTq1cv9t13X2bPng3Am2++yeGHHw6E5PKb3/yGQYMG0bFjR26++eaMxBLlE8QCYJeUz3mJbWUZC/wz8b4v8Eszuw5oQVi7d7W73xJFoEceCWecEZ4iLtysK11EtgbnnguJP+YzpmdPuOmmyp9XXFzMhAkTyMnJYcWKFYwfP57c3Fxee+01Lr74Yp544onNzpk1axbjxo1j5cqVdOnShdNPP71Scx7SiTJBfAh0NrMOhMRwLPCr1APMrLO7z0l8PAyYA+Du+6UcMwpYFVVyAMjLg332UYIQkexw1FFHkZOTA8Dy5cs5+eSTmTNnDmbGunXr0p5z2GGH0aBBAxo0aMAOO+zAokWLyMvLq1YckSUId19vZmcCLwM5wN3uPt3MRgNF7v4McKaZDQHWAd8BJ0cVz5YUFsIll8BXX8HOO8cVhYjEpSp/6UelSZMmP76/7LLLGDx4ME899RTz5s1j0KBBac9p0KDBj+9zcnJYv359teOIdCa1u78AvFBq259S3p9TgWuMynxkm0smiGefhd/9ribuKCKyZcuXL6dt27YA3HPPPTV679g7qbNF167QuXNoZhIRyRYjR47koosuolevXhl5KqgMc/ctH1ULFBQUeHUXDDr/fPj732HpUmjePEOBiUjWmjlzJl27do07jBqT7vc1s0nunna8rJ4gUhQWwrp18GKkc7dFRGoHJYgU/frBDjuomUlEBJQgNpGTE+ZEPP88rFkTdzQiIvFSgiilsBBWroQ334w7EhGReClBlHLggdCkiZqZRESUIEpp2BAOOQSefho2bow7GhGR+ChBpFFYCAsXwocfxh2JiNRl1Sn3DaFg34QJEyKLT2tSp3HooZCbG5qZ+m5WoFxEJDO2VO57S958802aNm3KvvvuG0l8eoJIY7vtYNCg0MwkIlKTJk2axMCBA+nduzcHH3wwCxcuBODmm28mPz+f7t27c+yxxzJv3jxuu+02brzxRnr27Mn48eMzHoueIMowbBicdVZYJ6JLl7ijEZHIZUG9b3fnrLPO4umnn6Z169Y88sgjXHLJJdx9991cc801fPHFFzRo0IBly5bRokULTjvttEo/dVSGniDKMCyxOKqeIkSkpqxZs4Zp06Zx0EEH0bNnT6688kqKi4sB6N69O8cffzwPPPBAmavMZZqeIMqwyy7Qu3fohxg5Mu5oRCRyWVDv293Zc889ee+99zbb9/zzz/P222/z7LPPctVVVzF16tTI49ETRDkKC2HixDCiSUQkag0aNGDJkiU/Joh169Yxffp0Nm7cyPz58xk8eDDXXnsty5cvZ9WqVTRr1oyVK1dGFo8SRDkKC8E9rBEhIhK1evXq8fjjj3PBBRfQo0cPevbsyYQJE9iwYQMnnHAC3bp1o1evXpx99tm0aNGCI444gqeeeiqyTmqV+y6He1gjYvfd4YUXtny8iNQuKvetct9VZhaeIl5/HVasiDsaEZGapQSxBYWFsHYtvPRS3JGIiNQsJYgt6N8fWrdW8T6RuqquNLNvSVV+TyWILcjJgSOOCH0QFSiNIiK1SMOGDfnmm2/qfJJwd7755hsaNmxYqfM0D6ICCgvh7rvhrbfgoIPijkZEMiUvL4/i4mKWLFkSdyiRa9iwIXl5eZU6RwmiAoYMgcaNQzOTEoRI3VG/fn06dOgQdxhZK9ImJjMbamazzWyumV2YZv9pZjbVzCab2Ttmlp/YfpCZTUrsm2RmB0QZ55Y0agRDh2qNCBHZukSWIMwsBxgDHALkA8clE0CKh9y9m7v3BK4D/pbYvhQ4wt27AScD90cVZ0UVFsKCBTBpUtyRiIjUjCifIPoAc939c3dfC4wFhqUe4O6pswuaAJ7Y/rG7f5XYPh1oZGYNIox1iw47LHRYazSTiGwtokwQbYH5KZ+LE9s2YWZnmNlnhCeIs9Nc5xfAR+6+Js25p5pZkZkVRd3JtP32MHCgEoSIbD1iH+bq7mPcfTfgAuDS1H1mtidwLfC7Ms69w90L3L2gdevWkcdaWAgzZsCnn0Z+KxGR2EWZIBYAu6R8zktsK8tYoDD5wczygKeAk9z9s0girCStESEiW5MoE8SHQGcz62Bm2wDHAs+kHmBmnVM+HgbMSWxvATwPXOju70YYY6Xsuiv06qVmJhHZOkSWINx9PXAm8DIwE3jU3aeb2WgzOzJx2JlmNt3MJgMjCCOWSJzXCfhTYgjsZDPbIapYK6OwEN57DxYtijsSEZFoqdx3JU2ZAj16wJ13wimnRH47EZFIqdx3BnXrBh06qJlJROo+JYhKSq4R8dprEOFKfyIisVOCqILCQlizBl5+Oe5IRESiowRRBfvuC61aqZlJROo2JYgqyM0Na0Q89xysWxd3NCIi0VCCqKLCQli+PKwRISJSFylBVNGQIaEMuJqZRKSuUoKoosaN4eCDQ4KoI1NJREQ2oQRRDVojQkTqMiWIajj8cKhXT8X7RKRuUoKohpYtYf/91Q8hInWTEkQ1FRbCtGkwd27ckYiIZJYSRDVpjQgRqauUIKqpfXvo2VPNTCJS9yhBZEBhIbz7LixeHHckIiKZowSRAcOGhbkQzz4bdyQiIpmjBJEBPXpAu3ZqZhKRukUJIgOSa0S8+iqsWhV3NCIimaEEkSFaI0JE6holiAwZMAC2317NTCJSdyhBZEhyjYjnn9caESJSNyhBZFBhIXz3HYwfH3ckIiLVpwSRQT/9qdaIEJG6I9IEYWZDzWy2mc01swvT7D/NzKaa2WQze8fM8lP2XZQ4b7aZHRxlnJnSuHFIElojQkTqgsgShJnlAGOAQ4B84LjUBJDwkLt3c/eewHXA3xLn5gPHAnsCQ4FbE9fLesOGwfz58PHHcUciIlI9UT5B9AHmuvvn7r4WGAsMSz3A3VekfGwCJP/uHgaMdfc17v4FMDdxvWi88w6sWLHl4yoguUaEmplEpLaLMkG0BeanfC5ObNuEmZ1hZp8RniDOruS5p5pZkZkVLVmypGpRzpkD++0HN9xQtfNLad06DHlVghCR2i72Tmp3H+PuuwEXAJdW8tw73L3A3Qtat25dtQA6d4ajj4a//hUWLqzaNUopLISpU+GzzzJyORGRWESZIBYAu6R8zktsK8tYoLCK51bP1VeHyQujRmXkclojQkTqgigTxIdAZzPrYGbbEDqdn0k9wMw6p3w8DJiTeP8McKyZNTCzDkBn4IPIIt1tNzj9dPjXv2DmzGpfrmNH6N5dzUwiUrtFliDcfT1wJvAyMBN41N2nm9loMzsycdiZZjbdzCYDI4CTE+dOBx4FZgAvAWe4+4aoYgXg0kvDONWLLsrI5ZJrRHzxRUYuJyJS48zryID9goICLyoqqt5Frr4aLrkkTIUeMKBal5o3LzxF5OfD22/DNttULzQRkSiY2SR3L0i3L/ZO6qxy7rmw885w/vnVnunWvj3cfTe8/z5ccEFmwhMRqUlKEKkaN4bRo2HiRHjqqWpf7pe/hLPOgptuysjlRERqlJqYSlu/PiwRt24dTJ8O9etX63Jr1oRpFp9+Ch99FDqwRUSyhZqYKiM3F669Nkygu/POal+uQQN45JGw6tzRR4eEISJSGyhBpHPYYbD//nD55bByZbUv16ED3HMPTJoE551X/fBERGqCEkQ6ZnD99bB4ccZKcAwbBiNGwJgx8OijGbmkiEiklCDK0qcPHHVURktwXHMN9O8Pp5wSWrBERLKZEkR5rr46dBpcfnlGLle/fuiPqF8/5J4ffsjIZUVEIqEEUZ5OnUIJjrvuglmzMnLJXXaB+++HTz4J0y5ERLKVEsSWXHZZRktwABx6KFx4IdxxBzz4YMYuKyKSUUoQW9K6dZgK/Z//hOJKGXLFFWF+xO9+l7GHExGRjFKCqIhzz4WddspICY6k3Fx4+OHwcPLLX8L332fksiIiGaMEURFNmoQSHO+9l9GaGW3bhiamGTPgjDMydlkRkYxQgqio4cOha9fQF7FuXcYue9BBodL4PfeEl4hItlCCqKhkCY5PPw2jmjLoz3+GwYPh97+HadMyemkRkSpTgqiMww8PPcujRmWkBEdSTg489BA0bx7mR6xalbFLi4hUmRJEZaSW4PjrXzN66R13DJ3Wn34Kp52Wsb5wEZEqU4KorL59w7CjG26Ar7/O6KUHDw6Tth98MOOtWCIilaYEURUZLsGR6uKL4ac/DQsNTZ6c8cuLiFSYEkRVdO4cZrjdeSfMnp3RS9erBw88AK1ahf6IFSsyenkRkQqrUIIwsyZmVi/xfnczO9LMqrfUWm33pz9Bo0YZLcGR1Lo1jB0LX3wRKr+qP0JE4lDRJ4i3gYZm1hZ4BTgRuCeqoGqFHXYIJTieegomTMj45QcMgKuugsceg1tvzfjlRUS2qKIJwtz9e+DnwK3ufhSwZ3Rh1RJ/+EPGS3CkOv/8sLjdiBFhNToRkZpU4QRhZv2B44HnE9tyKnDSUDObbWZzzezCNPtHmNkMM5tiZq+bWbuUfdeZ2XQzm2lmN5uZVTDWmtOkSeionjAhFPPLsHr14N57oU2b0B+xbFnGbyEiUqaKJohzgYuAp9x9upl1BMaVd4KZ5QBjgEOAfOA4M8svddjHQIG7dwceB65LnLsv8BOgO7AXsA8wsIKx1qxf/xr22CPU785gCY6kli3DEqXz58NvfqP+CBGpORVKEO7+lrsf6e7XJjqrl7r72Vs4rQ8w190/d/e1wFhgWKnrjks0XQFMBPKSu4CGwDZAA6A+sKhCv1FNSy3B8a9/RXKLfv3guutCd8ff/x7JLURENlPRUUwPmVlzM2sCTANmmNn5WzitLTA/5XNxYltZfgu8CODu7xGeUBYmXi+7+8w0cZ1qZkVmVrRkyZKK/CrROOKI0Ks8alRkdTLOPRcKC0O/xMSJkdxCRGQTFW1iynf3FUAh4Uu8A2EkU0aY2QlAAXB94nMnoCvhiaItcICZ7Vf6PHe/w90L3L2gdevWmQqn8pIlOBYtyngJjtRb3H13WLL05z+HmZulSxGRzKpogqifmPdQCDzj7usIzUDlWQDskvI5L7FtE2Y2BLgEONLd1yQ2/wyY6O6r3H0VISn1r2Cs8ejXD37xi5JEEYHttoNnnw39EPvtB0VFkdxGRASoeIK4HZgHNAHeTow22tIc3w+BzmbWwcy2AY4Fnkk9wMx6Ja59pLsvTtn1JTDQzHITiWkgkP1/M0dYgiNpzz1h/Hho1gwOOADeeiuyW4nIVq6indQ3u3tbdz/Ug/8Cg7dwznrgTOBlwpf7o4kRUKPN7MjEYdcDTYHHzGyymSUTyOPAZ8BU4BPgE3d/ttK/XU3bfXc49VS4446Ml+BI1akTvPMO5OXB0KHw/PNbPkdEpLLMKzBu0sy2Bf4M7J/Y9BYw2t2XRxhbpRQUFHhRNrS5LFoUvsF/+lN44olIb7V0aUgQn3wC990Hxx0X6e1EpA4ys0nuXpBuX0WbmO4GVgJHJ14rgH9nJrw6pk0bGDkSnnwyrGEdoVat4I034Cc/geOPh9tui/R2IrKVqWiC2M3d/5yY0/C5u18OdIwysFptxIiwAlBEJThSNW8OL74YSnKcfjpcc02ktxORrUhFE8QPZjYg+cHMfgL8EE1IdUCTJjB6NLz7Ltx0U+S3a9QoPLD86lehuOyFF2rGtYhUX24FjzsNuC/RFwHwHXByNCHVEb/9Lbz0Evzxj9CtGwwZEunt6teH+++HbbcNE7uXL4dbbgnrXYuIVEWFEoS7fwL0MLPmic8rzOxcYEqUwdVq9erBPfdA//5wzDFh0kKHDpHfcswYaNEC/vKXUNzvvvtC8hARqaxKrSjn7isSM6oBRkQQT93SrFmo8rpxI/zsZ/C//0V+S7MwHePaa8OiQz/7GfygxkARqYLqLDmafeW3s1GnTvDwwzBlSmh2qqHOgZEj4fbb4YUXwlBYLV0qIpVVnQShbtCKGjo0tPk88kgoxVFDTj015KYJE2DwYIiznqGI1D7l9kGY2UrSJwIDGkUSUV01ciR89FEYYtSjBxx8cI3c9phjQkvXL34B++8Pr74aZmCLiGxJuU8Q7t7M3ZuneTVz94qOgBIoKcfarRsceyzMnVtjtz70UHjlFfjqq1CVvAZvLSK1WHWamKSymjQJq/7UqxcWd4ho7Yh09tsPxo0L/eQDBoQuERGR8ihB1LSOHUNfxMyZcPLJNTqjbe+9QyXY+vVh4MDIK4GISC2nBBGHIUNCZ/WTT4YxqTVojz1CJdhWrUIYr75ao7cXkVpECSIuf/hDqLB32WU1Xq+7XbuQJDp1gsMPD3lKRKQ0JYi4mIV1I3r2DEWUIlw/Ip02beDNN6F3bzjqqDDpW0QklRJEnBo3Dp3W22wTOq1reDbbdtuFJqYhQ+DXv4ZLL4Xvv6/REEQkiylBxK1dO3jsMZgzB048MZTlqEFNmsAzz8BJJ8FVV4U+iocfVjVYEVGCyA6DBsHf/ha+qa+4osZv36AB3HtvWN+6VavQ4jVgAHz4YY2HIiJZRAkiW5x1Vhj2OmoUPP10LCHsv39ICv/6F3z2GfTpE0L66qtYwhGRmClBZAuzsGboPvvACSfAjBmxhJGTA7/5DXz6KVxwQagIu/vuoflJVWFFti5KENmkYcMw5rRx49BpvWxZbKE0bx6WL505M5SNuvRS6NoVHn1U/RMiWwsliGyTlwePPw5ffBHmSWzYEGs4HTvCE0/AG2+EhYiOOSbMwv7oo1jDEpEaoASRjfbbD26+OSzm8Oc/xx0NEMqFT5oU1piYNQsKCsLyFl9/HXdkIhKVSBOEmQ01s9lmNtfMLkyzf4SZzTCzKWb2upm1S9m3q5m9YmYzE8e0jzLWrHPaaXDKKaHx//HH444GCP0Tp54aRuSed15YA7tz59AUtXp13NGJSKZFliDMLAcYAxwC5APHmVl+qcM+BgrcvTvwOHBdyr77gOvdvSvQB1gcVaxZyQxuuQX69YPhw2Hq1Lgj+tG224ZSUtOnwwEHwEUXQX5+6D5R/4RI3RHlE0QfYK67f+7ua4GxwLDUA9x9nLsn5+5OBPIAEokk191fTRy3KuW4rUeDBqEDoFmz0Gn97bdxR7SJzp3DiNxXXw0T7n7xi5AwJk+OOzIRyYQoE0RbYH7K5+LEtrL8Fngx8X53YJmZPWlmH5vZ9Yknkq3PzjuHP83nz4fjjou90zqdIUPg44/h1lvDg87ee4emqMVb1zOfSJ2TFZ3UZnYCUAAkF2zOBfYD/gjsA3QEhqc571QzKzKzoiV1ecHl/v3Dt+8rr8DFF8cdTVq5uXD66aF/4txz4d//Dk8YN9wAa9fGHZ2IVEWUCWIBsEvK57zEtk2Y2RDgEuBId1+T2FwMTE40T60H/gPsXfpcd7/D3QvcvaB169YZ/wWyyimnhI7r666DG28MK/989FGoArtgQZgzsW5d3FGy3Xahasi0aWEw1vnnQ/fu8PrrcUcmIpUV5brSHwKdzawDITEcC/wq9QAz6wXcDgx198Wlzm1hZq3dfQlwAFAUYay1w9//Hr55R4wo+5j69UOHQJMm0LRpyfvUV7rtTZuGb/T27TMSapcu8Nxz8OKLoYrIkCFw9NEhebQtr6FRRLKGeYTDTszsUOAmIAe4292vMrPRQJG7P2NmrwHdgIWJU7509yMT5x4E/BUwYBJwaqKzO62CggIvKtoKcsjatWFCwv/+F16rVpW8L/3a0r7SY1MbNoRLLgl/9jdokLGQV68Oo56uvjoMlR01Cs45J+QyEYmXmU1y94K0+6JMEDVpq0kQmbRhQ1gA4n//g6VLQyXZRx8Nf/7femsYkpRBX3wREsOzz4ZhsbfcEibgiUh8yksQWdFJLTHJyQlDaHfcEfbaCx55JLQJrVsHBx4YigYuWpSx23XoECqaP/NMKPx3wAGhtLiqxYpkJyUI2dTQoaGf47LLSp4mbrstowsZHXFEmGT3pz+FEbx77BH63bOgj11EUihByOYaNYLRo8Okht69w/jV/v3DZIcM3uLyy0MuGjAg9LvvvTe8/XbGbiEi1aQEIWXr0gVeew0eeADmzQsV+v7wB1i5MmO36NQJnn8+LM29YkWoFHviiSoCKJINlCCkfGah7PisWfC734WhtnvsEQoIZmiAg1moJDJzZhhElWzZuvlmWL8+I7cQkSpQgpCK2W67MLJp4kRo0waOOgoOPTSsTZohjRvDlVeGlq1+/cKIp9694d13M3YLEakEJQipnD594IMP4Kabwjf3XnuFb/U1a7Z8bgXtvju89FJ4SPn229BHMXy4ajuJ1DQlCKm83Nzw5/3MmWFI0mWXQY8eMG5cxm5hFqrDzpwZ1sZ+8MGQOMaMycp6hSJ1khKEVF3btqHDIDl34oADQg9zBudONG0aFiSaMiU0N515JuyzD0yYkLFbiEgZlCCk+pJzJy69NEy222OPjM+d6No1DKh65JGQf37yk9BPcdddGRxU9c47YYELEQGUICRTGjUKpTqmTIFevcLciX33zejqQWah4N+sWfDXv4bE8H//BzvtFIrdTpxYxYFVK1eGePfbLwynuvJKLY0nghKEZNoee4Ta3g88EIov9e4NZ58N332XsVs0axYm1k2bFpqajj0Wxo4Nc/n22ivMyl66tIIXe/XVcNLtt4c5HiecEPpURozI6BOQSK3k7nXi1bt3b5cs8+237r//vXu9eu6tWrnfeaf7hg2R3GrFinD5vn3dwb1+ffejj3Z/5ZUybrlsmfspp4SDu3Rxf/fdsH3DBvdzzgnbTzzRfe3aSOIVyRaE6tppv1dj/2LP1EsJIot9/LH7gAHhn1tBgfvEiZHebupU93PPdd9++3DLdu3cL7/c/csvEwe88IJ7Xl5IXCNHun///aYX2LjR/YorwslHHLH5fpE6RAlC4rdxo/sDD7jvtFP4Z/frX7t//XWkt1y92n3sWPchQ8Itt+Nbf6XtcHfwDV3z3d9/v/wL3Hqru5n7fvuFJw6ROqi8BKE+CKkZyZIds2fDyJGhj2L33cOEu4jKuDZoAMccE7oZFt75LPOa7sngBfdzFRfTbulHnP9YH2bNKucCp58ODz8cer8HDszo8F2R2kAJQmpWs2Zw7bWhnkb//qFjuFcveOONaO737bdw4ons+H9H0rxDK+z99+n1/FXsM6ABN90Uhs/utx/ce29YN2kzxxwTVjiaMydM6Z43L5o4RbKQEoTEo0uXMMHuP/8Jq9odeGCo7/Tll5m7x1NPhaXrxo4Ni08UFZHTpzeHHhrWoSguhuuuCyU8hg+HnXeG004LlUQ8dZTrwQeHSRjffBMmYEyblrkYRbKYEoTExwyGDQurB40eDc89F4bJXnnl5utlV8aSJWHs689/HiZJfPhhWHxim202OaxNm7D89qxZYR2KwkK47z7o2xe6dw/DZZcsSRzcv384yB323x/ee6/q8YnUEkoQEr9GjcLcg1mzQoXYyy6DPfcMa5N6JSesPfZYOPfJJ8PEvQ8+gJ49yz3FrKSZaeFCuOMOaNIkTIXYeedQE+r552H9HnuFAoUtW8KQIfDyy9X4pehLjI0AABIFSURBVEWynxKEZI927UIJ19deCz3Mw4aFhPHpp1s+d/Hi0ER19NHhOpMmhdIf9etXKoRttw2zsydODC1J55wD48fD4YfDrrvCRXd04LN734HOnUOhwkceqeIvK5L9lCAk+xx4IHzyCfztb2Gq9F57wYUXwqpVmx/rHkYa5eeHJ46//CU0/3TrVu0w9twTbrgBFiwIXSX77APXXw+dftKGQxq9ydft++LHHQf//Ge17yWSjZQgJDvVrx9GOM2eHYbHXntt6Nh+6KGSZqevvw79DL/6VVi7dPLkkEhyczMeyrBhoY5fcXEI5YvvWtBxzsu8WO8w+P3v+e//XYlvVP0mqVuUICS77bgj/Pvf4Ulip51Cshg4MMyfyM8PKwtdf33oG+jatUbCGTkyrFPx2ruNefqkJ3k49wTa3XUZ97X8A9f+ZSMLF0YehkiNiDRBmNlQM5ttZnPN7MI0+0eY2Qwzm2Jmr5tZu1L7m5tZsZndEmWcUgv07w/vvx96kGfMCE8X+fmhKeqPf4ScnBoNxywUq7397voc8c29zDjoHE5e9nd2ung4HfLWccQRYZTt2rU1GpZIRkWWIMwsBxgDHALkA8eZWX6pwz4GCty9O/A4cF2p/VcAb0cVo9QyOTmhB/nTT8OworfeCrOxY9a0eT3yX74RrriCk7ifjzv+nBmTfuDnP4e8PDjvPMqfsS2SpaJ8gugDzHX3z919LTAWGJZ6gLuPc/fvEx8nAnnJfWbWG2gDvBJhjFIbbb99GN1Uw08N5TILo6ZuvZWunz3P3N0O5uVHlrH//vCPf4TWryFDQmf3+vVxBytSMVEmiLbA/JTPxYltZfkt8CKAmdUD/gr8sbwbmNmpZlZkZkVLfpzRJBKj00+Hhx7CJr7HT68exONjFlFcDFdfHR58fvYz2G23sIyq/slKtsvscI8qMrMTgAJgYGLT74EX3L3YzMo8z93vAO4AKCgo0BASyQ7HHgstWoQZdgMGsMMrr3DRRR04//wwWfyWW+Cii2DUqFDqKbnONuvWwfLl4bVs2ZZ//vBDWAf8+OPD+uAiGWZe2ZmqFb2wWX9glLsfnPh8EYC7/6XUcUOAfwAD3X1xYtuDwH7ARqApsA1wq7tv1tGdVFBQ4EVFRVH8KiJV8957cNhh0LBhyAIrVvz4Bb+qeBmL5y5n7aJlNPPlbF9vGY02fr/lazZrFpLPttuGz9OmheatIUPgpJPCI0qTJtH+XlKnmNkkdy9Iuy/CBJELfAocCCwAPgR+5e7TU47pReicHuruc8q4znBCR/aZ5d1PCUKy0rRpob9k/vxQCyr55Z74ua7xtsxZ0oL3Z23L59+1YEOTbek1uAWDhm1L686bHkvz5pv3u8yZA/ffH17z5oXk8MtfhmQxaBDUy8KR7O5heNcPP5S8vv9+08+VeTVpEgpr7bBD+Jn6vlWr7OqrykKxJIjEjQ8FbgJygLvd/SozG01YoOIZM3sN6AYkR45/6e5HlrrGcJQgpDZbvz68GjYs8xD3UPH8llvChHAIk/POPBMGDw4PCeXauBHeeSckikcfDU8ru+wS1tg+8cQamSOymcWLQ0zJ1+efl3ypV/V7Jzc31O5Kvho2DDPsFy9Ov66IWUgSZSWQ1Pc77FDu/0d1VWwJoiYpQUhd8eWXcNttcOedsHRp+G4/44zwUNCsWQUu8MMPIcvcd18oKLhhAxQUwMknh/6RVq0yH7R7SADjx4dkMH58SQ2thg1Didz8fGjceNMv+NRXefuSr7JmybuH5rtFi0KySP2Zblu6si0QntKSTx4ZnpEPhN+he3fYe+/w6tw59qc8JQiRWmj16vAwcMstoWJ5s2bhO/6MM0JV9Ar5+utQq+q++0Ipktzc0OR10kmhAmGDBlULbsMGmDKlJCG88w4/TiHfbruwuNKAAaFM7t57V/0+Ufn++/ITydKl4aks05YvD82OyRmUTZuGasO9epUkja5dK11ksjqUIERquQ8+gDFjwtpHa9eGeoZnnAFDh4Y/SitkypTQBPXAAyFxbLddGEZ10knQr1/57Vg//BCCSCaECRNg5cqwb9ddQyJIJoSuXWP/qzirrV0barV89FHJ65NPSpY0bNAgPGWkJo1u3SJr/lKCEKkjliyBu+4KBWST/d79+oX+6EGDwvstJoz16+H118NTxVNPhS//Tp1CojjhBOjQISzV+u67JQmhqKikjX+vvUoSwoABIUFI9WzYEAYcJBPGxx+Hn8uWhf05OaG8cGrS6NGjgm2O5VOCEKlj1q8Py2a88Qa8+WZY/mLjxpAw+vYtSRj9+28hYaxYAU88EZLFm2+Gbe3bl6y9Xb9+mKSRTAj77htmskv03MP/D6kJY9Kk0AwG4Ymvc+eQLPbfP0zSrAIlCJE6bvny8Af/m29WI2HMmwcPPhhOLigICWGffSrRhiWRcw99PcmEkXx16hSeCqtACUJkK7NiRWgZqlbCkNpjzZoqDwRQghDZyilhSFmUIERkE+UljH32CZ3dfftCnz6hD3qLE/Wk1lKCEJFyrVhR0ofx9tuhiXvNmrCvTZuQLJIJY599SkpBSe1XXoLIimquIhKv5s3hkEPCC8JQ/SlTwiJ+yVeyBIhZmKiXTBh9+4Zh+jU4t0tqiJ4gRKRCvvsuzOhOTRpLl4Z9DRtC794lCaNvX2jXTk1TtYGamEQk45LD9JPJ4oMPwojL1avD/h122DRhFBSEyduSXdTEJCIZZxYmXXfoEGoAQphsnWya+uCD8PO550rOadculB7q0SP87NkzzMvTk0Z20hOEiERq2bLQNPXRR6Fe4OTJMHt2ScXv5s03Txp77pl99f3qKj1BiEhsWrSAgw4Kr6Tvvw9FTZMJY/JkuPvuknp1ubmh5l9q4ujRI5pK5VI2JQgRqXGNG4f+iT59SrZt3AiffbZp0njjjVCANikvb9MnjZ49oWNHFY+NihKEiGSFevVC7bnOneGoo0q2L1kSqmFPnlzy86WXQgFUCE8VgwaFlfcGDw5DcNWnkRlKECKS1Vq3hiFDwitp9WqYPj1M6Bs/HsaNg8cfD/vatNk0YXTurIRRVeqkFpFaL7ni6bhxYTb4uHHw1Vdh3847b5owOnZUwkileRAislVxD+vvpCaMRYvCvl122TRhtG8fY6BZQAlCRLZq7jBrVkgUyaSRnAXevn1IFMmkscsuMQYaAyUIEZEUGzfCjBklCeOtt8IqqwC77QYDB0J+fni/226hWapJk3hjjooShIhIOTZuhKlTSxLGO++UJIyknXYqSRilXy1b1t5+jdgShJkNBf4O5AB3ufs1pfaPAE4B1gNLgN+4+3/NrCfwT6A5sAG4yt0fKe9eShAikknffhvmZaR7LViw6bHNm6dPHJ06Qdu2kJMTz+9QEbEkCDPLAT4FDgKKgQ+B49x9Rsoxg4H33f17MzsdGOTux5jZ7oC7+xwz2xmYBHR192Vl3U8JQkRqyg8/wBdfbJ445s4NBQzXrSs5dpttQr2q3XaDLl3C5MB+/bKn2m1cpTb6AHPd/fNEEGOBYcCPCcLdx6UcPxE4IbH905RjvjKzxUBroMwEISJSUxo1Cn0U+fmb79uwAebPT//kMW4c3HhjOK5Nm5Aokq+CAmjatGZ/jy2JMkG0BeanfC4G+pZz/G+BF0tvNLM+wDbAZ2n2nQqcCrDrrrtWJ1YRkYzIyQkjo9q3hwMP3HTf+vWhr2PixJLX00+HffXqhYWXUpPG7rvHW0YkK2ZSm9kJQAEwsNT2nYD7gZPdfWPp89z9DuAOCE1MNRCqiEiV5eZCr17hdfrpYds334TS6MmEMXYs3H572NeiRVhLI5kw+vSB7bevwXgjvPYCIHVEcV5i2ybMbAhwCTDQ3dekbG8OPA9c4u4TI4xTRCQ2LVtuutzrxo2hHHrqU8YVV4TtEPoxUp8y9torJJ4oRNlJnUvopD6QkBg+BH7l7tNTjukFPA4Mdfc5Kdu3ITQ3PevuN1XkfuqkFpG6auVKKCraNGksXhz2NW4MRxwRnjyqIpZOandfb2ZnAi8Thrne7e7TzWw0UOTuzwDXA02Bxyx053/p7kcCRwP7Ay3NbHjiksPdfXJU8YqIZKtmzUpKg0DJcq/JZBHVJD5NlBMR2YqV9wShZTZERCQtJQgREUlLCUJERNJSghARkbSUIEREJC0lCBERSUsJQkRE0lKCEBGRtOrMRDkzWwL8txqXaAUszVA4UatNsULtirc2xQq1K97aFCvUrnirE2s7d2+dbkedSRDVZWZFZc0mzDa1KVaoXfHWplihdsVbm2KF2hVvVLGqiUlERNJSghARkbSUIErcEXcAlVCbYoXaFW9tihVqV7y1KVaoXfFGEqv6IEREJC09QYiISFpKECIiktZWnyDMbKiZzTazuWZ2YdzxlMfMdjGzcWY2w8ymm9k5cce0JWaWY2Yfm9lzcceyJWbWwsweN7NZZjbTzPrHHVNZzOwPiX8D08zsYTNrGHdMqczsbjNbbGbTUrZtb2avmtmcxM/t4owxqYxYr0/8O5hiZk+ZWYs4Y0yVLt6UfeeZmZtZq0zca6tOEGaWA4wBDgHygePMLD/eqMq1HjjP3fOBfsAZWR4vwDnAzLiDqKC/Ay+5+x5AD7I0bjNrC5wNFLj7XoQlfY+NN6rN3AMMLbXtQuB1d+8MvJ74nA3uYfNYXwX2cvfuwKfARTUdVDnuYfN4MbNdgJ8CX2bqRlt1ggD6AHPd/XN3XwuMBYbFHFOZ3H2hu3+UeL+S8AXWNt6oymZmecBhwF1xx7IlZrYtYR30fwG4+1p3XxZvVOXKBRqZWS7QGPgq5ng24e5vA9+W2jwMuDfx/l6gsEaDKkO6WN39FXdfn/g4Ecir8cDKUMb/tgA3AiOBjI082toTRFtgfsrnYrL4CzeVmbUHegHvxxtJuW4i/IPdGHcgFdABWAL8O9EkdpeZRbQUfPW4+wLgBsJfiguB5e7+SrxRVUgbd1+YeP810CbOYCrhN8CLcQdRHjMbBixw908yed2tPUHUSmbWFHgCONfdV8QdTzpmdjiw2N0nxR1LBeUCewP/dPdewP/IniaQTSTa7ocRktrOQBMzOyHeqCrHw/j6rB9jb2aXEJp2H4w7lrKYWWPgYuBPmb721p4gFgC7pHzOS2zLWmZWn5AcHnT3J+OOpxw/AY40s3mEprsDzOyBeEMqVzFQ7O7JJ7LHCQkjGw0BvnD3Je6+DngS2DfmmCpikZntBJD4uTjmeMplZsOBw4HjPbsnjO1G+GPhk8R/b3nAR2a2Y3UvvLUniA+BzmbWwcy2IXT0PRNzTGUyMyO0kc9097/FHU953P0id89z9/aE/13fcPes/SvX3b8G5ptZl8SmA4EZMYZUni+BfmbWOPFv4kCytEO9lGeAkxPvTwaejjGWcpnZUELz6JHu/n3c8ZTH3ae6+w7u3j7x31sxsHfi33S1bNUJItEJdSbwMuE/sEfdfXq8UZXrJ8CJhL/GJydeh8YdVB1yFvCgmU0BegJXxxxPWomnnMeBj4CphP+Os6oshJk9DLwHdDGzYjP7LXANcJCZzSE8BV0TZ4xJZcR6C9AMeDXx39ltsQaZoox4o7lXdj85iYhIXLbqJwgRESmbEoSIiKSlBCEiImkpQYiISFpKECIikpYShEglmNmGlCHGkzNZAdjM2qer0CkSl9y4AxCpZX5w955xByFSE/QEIZIBZjbPzK4zs6lm9oGZdUpsb29mbyTWFXjdzHZNbG+TWGfgk8QrWSojx8zuTKz18IqZNYrtl5KtnhKESOU0KtXEdEzKvuXu3o0wC/emxLZ/APcm1hV4ELg5sf1m4C1370Go+ZScwd8ZGOPuewLLgF9E/PuIlEkzqUUqwcxWuXvTNNvnAQe4++eJgopfu3tLM1sK7OTu6xLbF7p7KzNbAuS5+5qUa7QHXk0sqIOZXQDUd/cro//NRDanJwiRzPEy3lfGmpT3G1A/ocRICUIkc45J+fle4v0ESpYDPR4Yn3j/OnA6/Lhu97Y1FaRIRemvE5HKaWRmk1M+v+TuyaGu2yUqwa4BjktsO4uwSt35hBXrfp3Yfg5wR6IS5wZCsliISBZRH4RIBiT6IArcfWncsYhkipqYREQkLT1BiIhIWnqCEBGRtJQgREQkLSUIERFJSwlCRETSUoIQEZG0/h9sRM7N1rTutQAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEWCAYAAAB8LwAVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3deZzV8/7A8de7aRkpWpGmFFJKFCPZJblZs/xQ16VcdC3Zrq1w6WbLcq1lqRBySzdchZCUdEuaVNqUEWlKmUq7lpl5//54n9FpOs3SnO98z5l5Px+P85hzvss57zNz5ry/n11UFeecc66gSmEH4JxzLjF5gnDOOReTJwjnnHMxeYJwzjkXkycI55xzMXmCcM45F5MnCOcSjIhMFJFrIvd7iMjksGNyFZMnCOeKICI/icjvIrJRRFaIyFARqRF2XM4FzROEc8VznqrWANoAbYE+IcfjXOA8QThXAqq6AvgESxSISHsRmSIia0Vktoicln+siNQRkddEZLmI/CYi/41sry0iH4hIdmT7ByKSFsobcq4QniCcK4HIF/lZQKaINAQ+BB4C6gB3AO+ISP3I4W8C1YFWwH7A05HtlYDXgIOAxsDvwICyeg/OFVflsANwLkn8V0QUqAF8DjwA/A34SFU/ihwzTkQygLNF5FMskdRV1d8i+78AUNXVwDv5TywiDwMTyuZtOFd8XoJwrnguUNWawGlAC6AeVgK4JFK9tFZE1gInAQ2ARsCaqOTwBxGpLiIvi8gSEVkPTAJqiUhKWb0Z54rDE4RzJaCqXwBDgSeBpcCbqlor6ra3qvaP7KsjIrViPM3tQHPgOFXdBzglsl2CfwfOFZ8nCOdK7hmgEzAFOE9E/iQiKSKSKiKniUiaqv4CjAVeiDRKVxGR/ERQE2t3WCsidbDqKucSjicI50pIVbOBN4CbgS7APUA2Vmq4kx3/V1cA24HvgF+BWyPbnwH2AlYBXwEfl1XszpWE+IJBzjnnYvEShHPOuZg8QTjnnIvJE4RzzrmYPEE455yLqdyMpK5Xr542adIk7DCccy6pzJgxY5Wq1o+1r9wkiCZNmpCRkRF2GM45l1REZMnu9nkVk3POuZgCTRAi0llEFopIpoj0jrG/sYhMEJGZIvKtiJwdte9IEZkqIvNEZI6IpAYZq3POuZ0FVsUUmXhsIDYlQRYwXURGq+r8qMPuA0aq6osi0hL4CGgiIpWBYcAVqjpbROpiI1Kdc86VkSBLEO2ATFVdrKrbgBHYtATRFNgncn9fYHnk/pnAt6o6G2x6ZFXNDTBW55xzBQSZIBpic9Pky4psi9YX+IuIZGGlh5si2w8DVEQ+EZFvROSuWC8gIj1FJENEMrKzs+MbvXPOVXBhN1J3A4aqahpwNvCmiFTCqr5OAi6P/LxQRDoWPFlVB6lquqqm168fs5eWc865PRRkgliGLZqSLy2yLdrVwEgAVZ0KpGILsWQBk1R1lapuxkoXRwcYq3POuQKCTBDTgWYi0lREqgJdgdEFjvkZ6AggIodjCSIbWxS+dWTlrcrAqcB8nAvSggXw2WdhR+FcwggsQahqDtAL+7JfgPVWmici/UTk/MhhtwPXishsYDjQQ81vwFNYkpkFfKOqHwYVq3Ns3w7nnQdnnQVz5oQdjXMJodysB5Genq4+ktrtsZdfhuuug9RUOOoo+N//IMWXiHbln4jMUNX0WPvCbqR2Lny//w79+sEJJ8DgwTBtGrz0UthRORc6TxDODRwIy5fDo4/C5ZfDmWdCnz6wrGCfCucqFk8QrmJbt84Sw5/+BKecAiLw4ouQkwM33VT0+c6VY54gXMX21FOwZg088siObQcfDH37wnvv2c25CsoThKu4srMtQVxyCRxdYJjNbbdZY3WvXrB+fTjxORcyTxCu4nrkEdi82RqoC6pSBQYNgl9+gXvuKfvYnEsAniBcxfTzz/DCC9CjB7RoEfuYdu2sHeKFF+Crr8o0POcSgScIVzHllxoeeKDw4x56CBo2hGuvtcF0zlUgniBcxbNwIQwdCtdfD40bF35szZrWDXbuXHjyyTIJz7lE4QnCVTz3328jpovbtnD++XDxxVbqyMwMNjbnEognCFexfPMNjBxpvZT226/45z33HFStatNxlJPpaZwriicIV7Hcdx/Urg133FGy8w48EPr3h/Hj4c03g4nNuQTjCQJg9eqwI3Bl4csvYexY6N0b9t235Of/7W82X9Pf/w6rVsU/PucSjCeItWutquGww+Dqq63x8ocfvBqhvFG1+ZUaNLDBb3uiUiWb9XXdOrj99vjG58q/bdtg9uyk6g3nCULVqg5atLBpFa66Cg49FNLSoGtX68Hy7beQlxd2pK40xo61Kbz/8Q+oXn3Pn+eII+Duu+GNN3xxIVcyd94JbdpAnTq27shjj9nMwTk5YUe2W74eRLS8PJg/36oivvwSJk3aMaNnrVpw4ok2odvJJ8Mxx1ijpUt8eXk2lcaGDbZqXGn/blu2wJFH2vPOmQN77RWfOF35tWABtG4N55wDjRrBxIkwb57tq1EDTjoJOnSA006zz2rlymUWWmHrQXiCKIwq/PTTjoTx5ZfWhx7sS+G44yxZnHwyHH+8/aFd4nn7bSsNDhtm03nHw+efQ8eOVm0VPdGfc7Gccw5MnmzdpOvXt22//gpffGHJYuJEuzgF+x45+WRLFh06QNu2gSaM0BKEiHQGngVSgCGq2r/A/sbA60CtyDG9VfWjAvvnA31VtdBRSmW2otyvv9ofOj9hzJxpV5IpKZb58xPGSSdBvXrBx+MKt307tGoF1arBrFnxXSXuqqss6XzzjV0dOhfLp5/adPKPP27VTLuzcqXVWkyYYAljwQLbXrPmjoRx2mlxTxihJAgRSQEWAZ2ALGx96W6qOj/qmEHATFV9UURaAh+papOo/aMABaYlTIIoaP16mDp1R8KYNg22brV9xx4LH3xQsv72Lr4GD4aePeH9923AWzytXm1tV4cc4kuUuthycqzd4fffrYRQrVrxz125cucSRsGEkV8l1aZNqRJGWEuOtgMyVXWxqm4DRgBdChyjwD6R+/sCy/N3iMgFwI/AvABjLL199rGrg4cesj/munVWwnj4YbuyfPDBsCOsuLZsgX/+E9q3h/POi//z160LTz/tS5S63Rs82NoanniiZMkBYP/94dJLbbLI+fNtZuERI6yadPFiK40ce6x9Dnv0CCT8IEsQ/wd0VtVrIo+vAI5T1V5RxzQAPgVqA3sDZ6jqDBGpAYzDSh93ABtjlSBEpCfQE6Bx48bHLFmyJJD3ssduuME+IAsWWM8oV7aeesq6o37+uV1tBUHVLhC++sr+zg0bBvM6LvmsXQvNmlkV54QJtlphPK1YsaOEUb06/Otfe/Q0YZUgiqMbMFRV04CzgTdFpBLQF3haVTcWdrKqDlLVdFVNr5/f8JNI7r/frhruvTfsSCqe9eut8bhTp+CSA9g//Usv+RKlblcPP2zVkE89Ff/kAHDAAXDZZbZE7h4mh6IEmSCWAY2iHqdFtkW7GhgJoKpTgVSgHnAc8LiI/ATcCtwjIns4uilEBxxgV7AjR8L06WFHU7E8/bT9cz78cPCvdfDBNm24L1Hq8mVmwrPPWtVPwdUKk0iQVUyVsUbqjlhimA78WVXnRR0zFnhbVYeKyOHAeKChRgUlIn3ZTRVTtNAaqYuyYYM1YrZqZVUdQVxJJLK8PBuBXJZWrbIv7U6d4J13yuY1t2+H9HRLSvPnW9uUq7guush6L33/vY3eT2ChVDGpag7QC/gEWACMVNV5ItJPRPK7k9wOXCsis4HhQA8NKmOFpWZNq2qaOBE+/jjsaMpOXh48/7wNMLzmmrJd17l/f9i0qWw7CFSpYu1Ny5d7lWJFN3GilSR790745FAUHyhXFrZtg5YtrSFp5szy3x1y2TIbIzBunPXZnj3bFuYZOhROPTXY187Ksg4BXbva65W1m2+GAQNgyhTrPeXMwoXWw69du7AjCVZurvUsWrXK3nMSjLJP5EbqiqFqVWswnTMH3nor7GiCNXy4zVf0v/9Z4+2MGTb4JyXFGov//nfrEx6UBx+00kvfvsG9RmEeftiXKC3ovfesHr59e/u75OaGHVFw3njDLgIfeywpkkORVLVc3I455hhNaLm5qunpqo0bq/7+e9jRxN/q1aqXXaYKqu3bqy5atPP+DRtUr7/e9h9+uOr06fGPYdEi1ZQU1V694v/cJfH++/Y+H3kkvs+bk6M6e7bqyy+rXnWV6oknqr79dnxfI57y8lQffdR+F8cdp3rFFXb/jDNUV64MO7r4W79e9YAD7POflxd2NMUGZOhuvldD/2KP1y3hE4Sq6uef26/8iSfCjiS+Pv5Y9cADVStXVn3oIdXt23d/7CefqDZsaF/kDzygum1b/OLo2lW1enXVX36J33PuqYsuUk1NVf3++z1/jhUrLNn06aPaoYPq3nvb5wdU69VTbdbM7t98s+rWrfGLPR62bFG98kqLr1u3HRdFr7xiv5cGDVQnTQo3xni79157v1Onhh1JiXiCSCRnnaVau7bqmjVhR1J6mzap3nijfYxatlSdMaN4561Zo/qXv9h5xxyjOm9e6WOZNcuer0+f0j9XPGRlqe6zj2rHjsW7mty6VXXaNNVnn7Uv1KZNdySDypWt9Nmrl+qwYaqZmfac27ap3nbbjiv0n38O/n0Vx6+/WukGVPv12/X9z55tyS0lRfWxx6x0neyWLLHE9+c/hx1JiXmCSCSzZ6uKqN51V9iRlM60aaqHHWYfodtuU928ueTPMWqUXQlXq6b65JNWhbKnzjlHtVatxEq8L7xgv5/XX995e16efaG8/bb97o4/3n4H+QkhLU31//7PfieTJxf9u/3Pf1Rr1lStW9dKc2GaO1e1SRP7siys+mvdOtVLLrH3e+65VkWZzLp1s/e8ZEnYkZSYJ4hEc+WV9oWQKFd8JbFtm1UNpaSoNmqkOn586Z5vxQrVLl3so3jKKaqLF5f8OSZPtvMffbR0scRbbq59+detqzp2rF0tX3ihVa/kJ4PUVNWTTlK94w5LmEuX7tlrLVyo2rq1XXw88EDpku2e+ugjS1QNGqh+/XXRx+flqT7/vGqVKqoHHWQXHcloyhT7W953X9iR7BFPEInmp59Uq1ZV7dEj7EhKZsECq+oAa3D87bf4PG9enuprr1mVTI0aqoMGFb+RLy9P9eSTVfffX3XjxvjEE09z5lgVUX5COPRQq14bMEA1IyO+bTCbNu2o9z/zTNXs7Pg9d2Hy8lSfflq1UiXVtm1LnuSmTbMEUaWK6nPPJVUDr+bmWvVegwbWESMJeYJIRLffbld7334bdiRFy821f9zUVNU6daxKIwhLlqiefrp9LM8+W3X58qLPGTvWjh8wIJiY4uGLL1THjCmbL+y8PEuw1apZVVXQDabbtqn27Gl/gwsv3PMkvXq1VTWBVT2tWxffOIPy1lsW82uvhR3JHvMEkYhWr1bdd1+rO09kS5dat0SwBvbifGmXRsFkNGJE4ce2bWt13onWiydsGRnW0B3kVfnq1TsSep8+pW9szs21ariUFGvEnjUrPnEGZdMmq2Y9+uikbmj3BJGo+ve3P8HEiWFHEtu//20Nv9Wrq770UtkW/b/7TrVdO/v9XHaZ6qpVux4zcqTGbAR2Zs0a1fPO2/E7XL8+fs+9cKF9iVepEv/f/xdfWJVNaqrqkCGJW+XUr5/9br/4IuxISsUTRKLavNmqAdq1S6x/guhBb8cfX7q+/KWxfbuNq6hc2QYgffjhzvuaN7futWE0yCaL3Fy7EKlUyX5fc+eW/jnHj7eu2vXqqX75ZemfL5aVK62LMKh2725X64lk2TK7cLroorAjKTVPEIns1VftzxBUvX5JRQ96e/jhwge9lZVvvlE94gj7PV17rV0Jv/KKPX733bCjSw4TJlhDfvXqNpZiT738sn02WrZU/eGHuIUXU06O6v33W1vdEUdYJ4lE0aOHdTTJzAw7klLzBJHIcnJUW7Wy4no8e7SU1J4OeisrW7bY2BERa3M48EDVY49NrJJXolu+3LoSg+p119nvtLhyclRvucXO7dxZde3a4OIs6JNPrLRSo4bq8OFl97q7k5Fhn8M77ww7krjwBJHoxoyxP8ULL4Tz+j/9ZEkqf9BbIs8VNXmy6iGHWKzjxoUdTfLZvt2+2MC6LP/4Y9HnrFtnHRTAkkQYpcqlS3eMzr7++pIlt3jK71Zdv37ZJskAeYJIdHl5dmW3//5l35d61ixrENx3X9VPPy3b195TGzYk76CqRPHee/Y3r11b9YMPdn/c4sV28ZCSYh0VwrRtmw0ozJ+iJegqrlhGjbLXD/t3EUeeIJLB1Kn25+jbt+xec9w4G/malmYDulzFkpmp2qaNfe7uvXfXxv7Jk61qp1Yt1c8+CyfGWP77X0tu++5r98vKli2qBx9s7SGJ0DYXJ54gksXFF9uMnStWBP9ab75pjY2tW+/59A4u+W3erHr11fZVcPrpO6bhfuMNa4Rt1sy6HCeaH36wUgTYBHllMWbi8cft9ZKlpF1MoSUIoDOwEMgEesfY3xiYAMwEvgXOjmzvBMwA5kR+nl7Ua5WLBLFwoRXlb7wxuNeInqO/Q4dyU4/qSunVV23cwYEH7hgZ3aFDYk+i9/vvqnffbY3X+dOLjBsXTMeFlSuttJ3oA1v3QCgJAkgBfgAOBqoCs4GWBY4ZBFwfud8S+Clyvy1wYOT+EcCyol6vXCQIVetdUrnyrgvuxENOzo6eSl27htfQ5xLTrFk2V1R+d+Iwe9WVxJo1dtFzwAEWe5s2NgVGPKuB/vY3+79MpK62cVJYgghyydF2QKaqLlbVbcAIoEuBYxTYJ3J/X2A5gKrOVNXlke3zgL1EpFqAsSaOBx6AatXiv/D977/DJZfAwIFw55229Gm1ivErdcV01FG2ROyECfDyy1ClStgRFU/t2tC7N/z0EwwZAlu2wOWXwyGHwDPPwMaNpXv+OXNg8GC44QZo0SIuISeN3WWO0t6A/wOGRD2+AhhQ4JgGWDVSFvAbcMxunuez3bxGTyADyGjcuHEQyTUc999vV0Lx6qmzapXqCSdY3+1nn43PczqXqHJzVUePtu6oYI3sffrs2UqDeXk2F1nt2old3VYKhFSCKI5uwFBVTQPOBt4UkT9iEpFWwGPA32KdrKqDVDVdVdPr169fJgGXiTvugPr14e67bZLo0vjxRzjxRLsy/M9/4Oab4xOjc4mqUiU47zyYNAmmToXTT4f+/eGgg+Daa+G774r/XB9+CJ99ZiX7OnWCizlBBZkglgGNoh6nRbZFuxoYCaCqU4FUoB6AiKQB7wFXquoPAcaZeGrWhPvvh4kTYezYPX+eb76B44+HX3+1D/nFF8ctROeSQvv28M47sHAh/PWvMGwYHH44dOkCkycXfgG2fbtdrDVvbtVLCWbrVpg5E157Dd54I6AX2V3RorQ3oDKwGGjKjkbqVgWOGQv0iNw/HGuDEKBW5PiLivt65aaROt/WrTZiuHXrPZuM7uOPrXdH48aq8+fHPz7nktHKlVaFW7euVT+1b2/zecX6H3v2WTtmzJiyj7OA7GwbivLkk7ZWV+vWO69DdfTRe/7chNjN9WxgEdab6d7Itn7A+ZH7LYH/RZLBLODMyPb7gE2Rbfm3/Qp7rXKXIFRtLQRQHTq0ZOcNHWqfnqOOslknnXM727jRljtt2tT+x5o1s9HR+et/r15t7Q5nnFGm833l5loHxpEjVe+5x3rVNmy4IxGA9UQ++2xrVnn7bRumUpoJjQtLEKKlreNOEOnp6ZqRkRF2GPGVlwfHHQcrV8KiRZCaWvjxqvDII3DffXDGGVa03mefws9xriLLyYF334UnnoCMDGv7u+kmWLLE6m5mzYLWrQN56c2brYPUrFkwe7b9/PZb2LTJ9qekWG1YmzbWwSz/Z7ybW0Vkhqqmx9znCSLBTZhgjWyPP27dU3cnJwd69bLuiX/5C7zyClStWnZxOpfMVOGLL1h//xPs8+VHALye+jfurfsSVavyx61aNXZ6vLvb7o7bvNmSwKxZds2X//W7zz6WAKKTQcuWRV8TxoMniGR31lnw1VeweLH1+S5o82bo2hXGjLH+4I88AiJlH6dzSWjjRhgxwoY6fP01tK0yl7sPe4+vj+vFWmqzbRslvm3duuN+wa/YJk12TQYHHRTev2xhCaJyWQfj9sBjj9mn6NFHrSQRLTvbuvR9/bUNgkvA3hbOJRpVq1EaPBiGD7ck0aqVjau74oojqFPnCC6L0+vk5u5IFpUrQ40acXjiMuIJIhkceSRccQU895xVIzVubNt/+MFKF0uXWj3qBReEG6dzCW7tWptEYPBgq/evXh0uu8yGR7RvH/+reBFLCpUr22slm7AHyrni6tfPLkceeMAeT58OJ5wAq1fD+PGeHJzbDVUb8tC9Oxx4oF1jpaTAiy/CL7/Aq6/acCGvld2VlyCSxUEHWe+Kp56yXhX/+Afstx98/LEN5HEuQWzaZIXbzMydb7m5NpVRixbWO+fww6FRIxv4HIRVq2wA2ZAhsGCBjT/t3t1KC0cfHcxrljfeSJ1M1qyBgw+GdevsE/7hh3DAAWFH5SqgdetiJ4HMTLsqj1a/Phx6qCWCBQvsY5yvevUdCSM6cRx66J51wsvLs45/gwfDe+9Zvf/xx1tSuPRS2Hvv0r3v8sgbqcuLOnWsXPzZZ9aaVrNm2BG5cmzNmtgJIDPT+kZEa9DAvtQ7d7af+bdDDoF9991xnKpd2S9YsPPtyy+tbSBfSoqdH5008pNIrI/9ihU2bOGVVyxx1a4N111nieGII4L5/VQEXoJwzrFuHXzxhV17fPWVJYHfftv5mEaNdv7yj04C8bgy37jRpkxasMDm08tPHt9/b8N88jVsuHNJY+JE6+GdkwOnnmpJ4eKLy2YMQXng4yCcczvZuhWmTLH+DePHW5+H3FzYay/rzdOixc5JoGlT2xeG7dutVBCdNPKTyMaNVoXVowdccw0cdlg4MSYzTxDOVXC5uTZ697PPLCFMnmxrSKWkwLHH2swsHTtafX2yrCOlalVLdev6pAGl4W0QzlUwqlZNlJ8QPv98R5VRq1ZWDXPGGXDKKTu3ESQTEWv7cMHxBOFcObFihSWD/KSwdKltb9TIhsl07GjTevmXqisuTxDOJan163c0LI8fD/Pm2fY6daBDB7jnHksKhx7qg8DcnvEE4VwSWbEC3n/fZnKfMMF67uy1F5x8Mlx5pVUbtWkT3OAzV7F4gnAuwS1ZYoO+3nkH/vc/a19o1gxuv93GHSRTw7JLLoEmCBHpDDwLpABDVLV/gf2NgdexJUZTgN6q+lFkXx9szepc4GZV/STIWJ1LJAsX2vyL77wDM2bYtiOPhL594aKLrKHZq41c0AJLECKSAgwEOgFZwHQRGa2q86MOuw8YqaovikhL4COgSeR+V6AVcCDwmYgcpqq5QcXrXJhUbXbR/KQwP/JfctxxNsP7hRdaW4JzZSnIEkQ7IFNVFwOIyAigCxCdIBTIXxNzX2B55H4XYISqbgV+FJHMyPNNDTBe58pUXh5Mm2ZJ4d13bT2oSpWs6+n111vPo7S0sKN0FVmQCaIhsDTqcRZwXIFj+gKfishNwN7AGVHnflXg3IYFX0BEegI9ARrnr5HgXALLyYFJkywhvPceLF8OVapY43KfPtClS/zXHHZuT4XdSN0NGKqq/xKR44E3RaTYU2up6iBgENhI6oBidK5Utm61rqjvvms9kFavtp5HZ51l7Qnnnpu8g9Vc+RZkglgGNIp6nBbZFu1qoDOAqk4VkVSgXjHPdS6h5OXBTz/B3Lk7bnPmWIPz9u22MP1551lS6Nw5OVcYcxVLkAliOtBMRJpiX+5dgT8XOOZnoCMwVEQOB1KBbGA08G8ReQprpG4GfB1grM4VmyqsXLlzEpg71waqbdq047gmTWyq6XPPtXaFjh19ziCXXAJLEKqaIyK9gE+wLqyvquo8EekHZKjqaOB2YLCI3IY1WPdQmz1wnoiMxBq0c4AbvQeTC8O6dfbFn58E8hPC6tU7jtlvP0sEV19ti/0dcQS0bGklBueSmc/m6lzE8uU2ZUV0Ilga1c2iRg378s9PAvm3/fYLL2bnSstnc3WuEPPnw5NPwrBh1lZQpYotRnPKKTuSQOvW0LixD05zFYsnCFdhTZ5sg9DGjLFeRX/7m02DffjhliScq+g8QbgKJS/PEsLjj9uKanXr2vQVN94I9eqFHZ1zicUThKsQtm61KqQnnrBup02bwoABcNVV3t3Uud3xBOHKtbVr4eWX4ZlnbKrstm1hxAhb1L6yf/qdK5T/i7hyadkySwovvwwbNsCZZ1oJ4vTTvaHZueLyBOHKlXnzrEfSW29Ze8Oll8Kdd1rJwTlXMp4gXNJT3dEj6YMPrEfSddfBbbdZW4Nzbs94gnBJKy/PJr97/HH46ivrhfTPf8INN3iPJOfiodAEISIbsCkwdtkFqKr6ZAKuzG3bBq+/blVJixZZKWHgQOjRw3skORdPhSYIVa1ZVoE4VxRVKzHceSdkZsLRR8Pbb9vsqN4jybn4K6oEUaew/aq6Jr7hOBfbrFnWpjBxok2E9+GHtp6C90hyLjhFXXfNwKqYYv0bKnBw3CNyLsovv8B998Frr0GdOlaV1LOnlxicKwtFVTF5HxAXit9/h6eegkcftTaHv//dEkWtWmFH5lzFUezrMBGpjS3ck5q/TVUnBRGUq7hUYfhw6N3bptq+6CJ47DE49NCwI3Ou4ilWghCRa4BbsKU/ZwHtganA6cGF5iqaqVOtnWHaNBvY9uabcOqpYUflXMVVqZjH3QIcCyxR1Q5AW2BtYFG5CmXJEujaFU44AX7+2dobMjI8OTgXtuImiC2qugVARKqp6ndA86JOEpHOIrJQRDJFpHeM/U+LyKzIbZGIrI3a97iIzBORBSLynIj3VylvNmyAe+6B5s1h9Gj4xz9sXEOPHlCpuJ9M51xgitsGkSUitYD/AuNE5DdgSWEniEgKMBDoBGQB00VktKrOzz9GVW+LOv4mrGSCiJwAnAgcGdk9GTgVmFjMeF0Cy821UsJ998HKlXD55dWDvV4AABieSURBVNYY3ahR2JE556IVK0Go6oWRu31FZAKwL/BxEae1AzJVdTGAiIwAugDzd3N8N+CB/JfEGsOrYl1sqwArixOrS2yff249kmbPtiql0aOhXbuwo3LOxVKsgryItBeRmgCq+gV2JV/U/JgNgagl38mKbIv1/AcBTYHPI68xFZgA/BK5faKqC2Kc11NEMkQkIzs7uzhvxYVk0SLo0gU6drQ1Gt5+2ybY8+TgXOIqbk3vi8DGqMcbI9vipSswSlVzAUTkUOBwrNdUQ+B0ETm54EmqOkhV01U1vX79+nEMx8XLmjXWM6lVK5gwwaqSvvvOpuH2ViXnEltxE4So6h+T9qlqHkVXTy0DomuV0yLbYukKDI96fCHwlapuVNWNwFjg+GLG6hLA5s22vOehh8Jzz9nSnt9/b+MbUlOLPt85F77iJojFInKziFSJ3G4BFhdxznSgmYg0FZGqWBIYXfAgEWkB1MbGVeT7GThVRCqLSBWsgXqXKiaXeLZvh5dessRw111w3HEwcyYMGgT77x92dM65kihugrgOOAErAWQBxwE9CztBVXOAXsAn2Jf7SFWdJyL9ROT8qEO7AiOiSyjAKOAHYA4wG5itqmOKGasLQW6ureLWogVcfz0cfDBMmgRjx8KRRxZ9vnMu8cjO38vJKz09XTMyMsIOo8JRhTFj4N57Ye5cOOooeOQRn2nVuWQhIjNUNT3WvuL2YjpMRMaLyNzI4yNF5L54BumSz4QJ1lW1SxfYuhVGjIBvvoGzz/bk4Fx5UNwqpsFAH2A7gKp+i1UNuQpo+nTo1AlOPx2ysmDwYJg/Hy67zEdAO1eeFPffubqqfl1gW068g3GJbf58m121XTtbwOepp6xn0jXX+PoMzpVHxf23XiUihxBZn1pE/g8bwOYqgJ9+ggcegGHDYO+94Z//tLENNX1BWufKteImiBuBQUALEVkG/AhcHlhULiGsWAEPPwwvvwwpKTZFRu/eULdu2JE558pCcediWgycISJ7Y9VSm7E2iEIn7HPJ6bffbJDbs8/aam5XX20zrTaMOVGKc668KrQNQkT2EZE+IjJARDphiaE7kAlcWhYBurKzaZN1UW3aFPr3hwsugAULbOCbJwfnKp6iShBvAr9ho5yvBe7FZle9UFVnBRybKyM5OTbSuV8/m377vPPgoYd8gJtzFV1RCeJgVW0NICJDsIbpxvmLB7nkN3Uq3HCD9Uo69VR4910b2+Ccc0V1c92efycy02qWJ4fyITvb2hZOOMHu/+c/Owa+OeccFF2COEpE1kfuC7BX5LEAqqr7BBqdi7vcXBgyBPr0sSU/77rLGqBr1Ag7Mudcoik0QahqSlkF4oI3fbpVJ2VkwGmnwcCB0LJl2FE55xKVT4xQAaxZA9ddZ1NvL1sG//63Lf3pycE5VxhPEOVYXh688gocdphVK916q63m1q2bT6bnnCuaz6BTTs2cadVJX30FJ51k1UnebdU5VxJegihn1q6Fm26C9HRYvBhef90W7vHk4JwrKS9BlBOq8OabcOedsGqVlR4efBBq1Qo7Mudcsgq0BCEinUVkoYhkikjvGPufFpFZkdsiEVkbta+xiHwqIgtEZL6INAky1mQ2Zw6ccgp0725LfU6fDs8/78nBOVc6gZUgRCQFGAh0wtaxni4io1V1fv4xqnpb1PE3AW2jnuIN4GFVHSciNYC8oGJNVuvX2zTc+clgyBC46ipftMc5Fx9BfpW0AzJVdbGqbgNGAF0KOb4bMBxARFoClVV1HICqblTVzQHGmlRUratq8+Y24+o118CiRTYy2pODcy5egvw6aQgsjXqcFdm2CxE5CGgKfB7ZdBiwVkTeFZGZIvJEpERS8LyeIpIhIhnZ2dlxDj8xff89dOwIl18OaWkwbZrNtlqnTtiROefKm0S53uwKjIrM9wRW9XUycAdwLHAw0KPgSao6SFXTVTW9fv36ZRVraKZNg/btrQvriy9aF9Zjjw07KudceRVkglgGNIp6nBbZFktXItVLEVnArEj1VA7wX+DoQKJMEp98Aqefbm0NGRk2MjrFJ0JxzgUoyAQxHWgmIk1FpCqWBEYXPEhEWgC1sTUnos+tJSL5xYLTgfkFz60ohg+Hc8+FZs3gf/+DQw4JOyLnXEUQWIKIXPn3Aj4BFgAjVXWeiPQTkfOjDu0KjFBVjTo3F6teGi8ic7DZYwcHFWsie+45+POf4cQT4Ysv4IADwo7IOVdRSNT3clJLT0/XjIyMsMOIG1W4/35b2e2CC6wUkZoadlTOufJGRGaoanqsfT6SOgHl5tpI6EGDrAvriy9CZf9LOefKWKL0YnIRW7bApZdacrjnHvvpycE5Fwb/6kkg69dDly4wcSI88wzcckvYETnnKjJPEAli5Uo46yybV2nYMBsI55xzYfIEkQAWL4Yzz4RffoExY6Bz57Ajcs45TxChmz3bEsK2bbYM6HHHhR2Rc84Zb6QO0aRJNk135cowebInB+dcYvEEEZL337dqpQMPhClT4PDDw47IOed25gkiBK++ChddBG3aWMmhUaOiz3HOubLmCaIMqcJjj9m6DZ06wWefQd26YUflnHOxeYIoI3l5cPvt0Ls3dOsGo0dDjRphR+Wcc7vnvZjKwPbt8Ne/2viGm2+Gp5/2ld+cc4nPE0TANm2CSy6BsWPh4YehTx8QCTsq55wrmieIAK1ZA+ecA19/bXMqXXtt2BE551zxeYIISHY2nHYa/PADjBoFF14YdkTOOVcyniAC8q9/wcKFMG4cdOgQdjTOOVdy3lQagC1bYMgQm5nVk4NzLlkFmiBEpLOILBSRTBHpHWP/0yIyK3JbJCJrC+zfR0SyRGRAkHHG29tvw+rV0KtX2JE459yeC6yKSURSgIFAJyALmC4io1V1fv4xqnpb1PE3AW0LPM2DwKSgYgyCKjz/PLRsaW0QzjmXrIIsQbQDMlV1sapuA0YAXQo5vhswPP+BiBwD7A98GmCMcff11zBjBtx4o3dndc4ltyATRENgadTjrMi2XYjIQUBT4PPI40rAv4A7CnsBEekpIhkikpGdnR2XoEtr4ECoWROuuCLsSJxzrnQSpZG6KzBKVXMjj28APlLVrMJOUtVBqpququn169cPPMii/PqrtT90725JwjnnklmQ3VyXAdHzlKZFtsXSFbgx6vHxwMkicgNQA6gqIhtVdZeG7kQyZIgt/HPjjUUf65xziS7IBDEdaCYiTbHE0BX4c8GDRKQFUBuYmr9NVS+P2t8DSE/05JCTAy+9BGecAS1ahB2Nc86VXmBVTKqaA/QCPgEWACNVdZ6I9BOR86MO7QqMUFUNKpayMGYMLF3qpQfnXPkhSf69/If09HTNyMgI7fXPOAO+/96m1qjs49Odc0lCRGaoanqsfYnSSJ3UFiyA8ePhuus8OTjnyg9PEHEwcCBUrQrXXBN2JM45Fz+eIEpp/Xp4/XXo2hUSoKetc87FjSeIUnrzTdi40RunnXPljyeIUlC16qVjj4V27cKOxjnn4subVEthwgRroB46NOxInHMu/rwEUQoDBkDdunDZZWFH4pxz8ecJYg/9/DO8/76tM52aGnY0zjkXf54g9tDLL9vP664LNw7nnAuKJ4g9sHUrDB4M550HBx0UdjTOORcMTxB74D//gexsX1LUOVe+eYLYAwMGQPPm0LFj2JE451xwPEGUUEYGTJvmS4o658o/HwdRQgMHwt57w5VXhh2Jc660tm/fTlZWFlu2bAk7lMClpqaSlpZGlSpVin2OJ4gSWLUKhg+Hv/4V9t037Gicc6WVlZVFzZo1adKkCVKOqwRUldWrV5OVlUXTpk2LfZ5XMZXAq69aDyafd8m58mHLli3UrVu3XCcHABGhbt26JS4pBZogRKSziCwUkUwR2WXJUBF5WkRmRW6LRGRtZHsbEZkqIvNE5FsRCX2scm4uvPACnHYatGoVdjTOuXgp78kh3568z8CqmEQkBRgIdAKygOkiMlpV5+cfo6q3RR1/E9A28nAzcKWqfi8iBwIzROQTVV0bVLxF+fBDWLIE/vWvsCJwzrmyFWQJoh2QqaqLVXUbMALoUsjx3YDhAKq6SFW/j9xfDvwKhLrawsCB0LAhdCnsHTjnXAmsXr2aNm3a0KZNGw444AAaNmz4x+Nt27YVem5GRgY333xzoPEF2UjdEFga9TgLOC7WgSJyENAU+DzGvnZAVeCHAGIslkWL4NNP4cEHfUlR51z81K1bl1mzZgHQt29fatSowR133PHH/pycHCrv5ksnPT2d9PSYS0nHTaJ83XUFRqlqbvRGEWkAvAl0V9W8gieJSE+gJ0Djxo0DC+6FF6BKFZuYzzlXPt16K0S+q+OmTRt45pmSndOjRw9SU1OZOXMmJ554Il27duWWW25hy5Yt7LXXXrz22ms0b96ciRMn8uSTT/LBBx/Qt29ffv75ZxYvXszPP//MrbfeGpfSRZAJYhnQKOpxWmRbLF2BnfoGicg+wIfAvar6VayTVHUQMAggPT1dSxtwLBs3wmuvwSWXwP77B/EKzjm3s6ysLKZMmUJKSgrr16/nyy+/pHLlynz22Wfcc889vPPOO7uc89133zFhwgQ2bNhA8+bNuf7660s05iGWIBPEdKCZiDTFEkNX4M8FDxKRFkBtYGrUtqrAe8AbqjoqwBiLNGyYrTvt8y45V76V9Eo/SJdccgkpKSkArFu3ju7du/P9998jImzfvj3mOeeccw7VqlWjWrVq7LfffqxcuZK0tLRSxRFYI7Wq5gC9gE+ABcBIVZ0nIv1E5PyoQ7sCI1Q1ugRwKXAK0COqG2yboGLdnfwlRdu2hfbty/rVnXMV1d577/3H/X/84x906NCBuXPnMmbMmN2OZahWrdof91NSUsjJySl1HIG2QajqR8BHBbbdX+Bx3xjnDQOGBRlbcUyaBHPnwiuv+LxLzrlwrFu3joYNGwIwtIzXN/aR1IUYMADq1IFu3cKOxDlXUd1111306dOHtm3bxqVUUBKyc81O8kpPT9eMjIy4Pd+yZbYY0G23wRNPxO1pnXMJZMGCBRx++OFhh1FmYr1fEZmhqjH7y3oJYjdefhny8uD668OOxDnnwuEJIoZt22DQIDjnHDj44LCjcc65cHiCiOGdd2DlSp+11TlXsXmCiGHAADj0UDjzzLAjcc658HiCKGDmTJgyBW64ASr5b8c5V4H5V2ABAwdC9erQo0fYkTjnXLgSZbK+hLBmDfz733DFFVC7dtjROOfKu9WrV9OxY0cAVqxYQUpKCvXr28oGX3/9NVWrVi30/IkTJ1K1alVOOOGEQOLzBBHltdfg99+9cdo5VzaKmu67KBMnTqRGjRqeIIKWl2fTep98Mhx5ZNjROOfKXILM9z1jxgz+/ve/s3HjRurVq8fQoUNp0KABzz33HC+99BKVK1emZcuW9O/fn5deeomUlBSGDRvG888/z8knnxzX8D1BRHz8MSxeDI88EnYkzrmKSlW56aabeP/996lfvz5vv/029957L6+++ir9+/fnxx9/pFq1aqxdu5ZatWpx3XXXlbjUURKeICIGDIAGDeDCC8OOxDkXigSY73vr1q3MnTuXTp06AZCbm0uDBg0AOPLII7n88su54IILuOCCC8okHk8QQGYmjB0LfftCEW1CzjkXGFWlVatWTJ06dZd9H374IZMmTWLMmDE8/PDDzJkzJ/B4vJsr8OKLttZ0z55hR+Kcq8iqVatGdnb2Hwli+/btzJs3j7y8PJYuXUqHDh147LHHWLduHRs3bqRmzZps2LAhsHgqfILYvBlefRUuvtiqmJxzLiyVKlVi1KhR3H333Rx11FG0adOGKVOmkJuby1/+8hdat25N27Ztufnmm6lVqxbnnXce7733Hm3atOHLL7+MezwVvopp7Vr40598SVHnXLj69u37x/1Jkybtsn/y5Mm7bDvssMP49ttvA4sp0BKEiHQWkYUikikivWPsfzpqSdFFIrI2al93Efk+cuseVIwHHggjRsBJJwX1Cs45l5wCK0GISAowEOgEZAHTRWS0qs7PP0ZVb4s6/iagbeR+HeABIB1QYEbk3N+Citc559zOgixBtAMyVXWxqm4DRgBdCjm+GzA8cv9PwDhVXRNJCuOAzgHG6pyroMrLqppF2ZP3GWSCaAgsjXqcFdm2CxE5CGgKfF6Sc0Wkp4hkiEhGdnZ2XIJ2zlUcqamprF69utwnCVVl9erVpKamlui8RGmk7gqMUtXckpykqoOAQWBrUgcRmHOu/EpLSyMrK4uKcIGZmppKWlpaic4JMkEsAxpFPU6LbIulKxA9Rd4y4LQC506MY2zOOUeVKlVo2rRp2GEkrCCrmKYDzUSkqYhUxZLA6IIHiUgLoDYQPXTwE+BMEaktIrWBMyPbnHPOlZHAShCqmiMivbAv9hTgVVWdJyL9gAxVzU8WXYERGlUJqKprRORBLMkA9FPVNUHF6pxzbldSXhpn0tPTNSMjI+wwnHMuqYjIDFVNj7mvvCQIEckGlpTiKeoBq+IUTtCSKVZIrniTKVZIrniTKVZIrnhLE+tBqlo/1o5ykyBKS0QydpdFE00yxQrJFW8yxQrJFW8yxQrJFW9QsVb4yfqcc87F5gnCOedcTJ4gdhgUdgAlkEyxQnLFm0yxQnLFm0yxQnLFG0is3gbhnHMuJi9BOOeci8kThHPOuZgqfIIoalGjRCIijURkgojMF5F5InJL2DEVRURSRGSmiHwQdixFEZFaIjJKRL4TkQUicnzYMe2OiNwW+QzMFZHhIlKyaToDJiKvisivIjI3alsdERkXWQRsXGQandDtJtYnIp+Db0XkPRGpFWaM0WLFG7XvdhFREakXj9eq0AkialGjs4CWQDcRaRluVIXKAW5X1ZZAe+DGBI8X4BZgQdhBFNOzwMeq2gI4igSNW0QaAjcD6ap6BDaVTddwo9rFUHZdw6U3MF5VmwHjI48TwVB2jXUccISqHgksAvqUdVCFGEqM9XFEpBE2b93P8XqhCp0gKPmiRqFS1V9U9ZvI/Q3YF1jMNTYSgYikAecAQ8KOpSgisi9wCvAKgKpuU9W1hZ8VqsrAXiJSGagOLA85np2o6iSg4PxpXYDXI/dfBy4o06B2I1asqvqpquZEHn6FzSidEHbzuwV4GrgLW4UzLip6gij2okaJRkSaYEu0Tgs3kkI9g31g88IOpBiaAtnAa5EqsSEisnfYQcWiqsuAJ7ErxV+Adar6abhRFcv+qvpL5P4KYP8wgymBvwJjww6iMCLSBVimqrPj+bwVPUEkJRGpAbwD3Kqq68OOJxYRORf4VVVnhB1LMVUGjgZeVNW2wCYSpwpkJ5G6+y5YUjsQ2FtE/hJuVCUTmb054fvYi8i9WNXuW2HHsjsiUh24B7g/3s9d0RNESRY1SggiUgVLDm+p6rthx1OIE4HzReQnrOrudBEZFm5IhcoCslQ1v0Q2CksYiegM4EdVzVbV7cC7wAkhx1QcK0WkAUDk568hx1MoEekBnAtcHr0cQQI6BLtYmB35f0sDvhGRA0r7xBU9QRRrUaNEISKC1ZEvUNWnwo6nMKraR1XTVLUJ9nv9XFUT9ipXVVcAS0WkeWRTR2B+iCEV5megvYhUj3wmOpKgDeoFjAa6R+53B94PMZZCiUhnrHr0fFXdHHY8hVHVOaq6n6o2ify/ZQFHRz7TpVKhE0SkESp/UaMFwEhVnRduVIU6EbgCuxqfFbmdHXZQ5chNwFsi8i3QBngk5HhiipRyRgHfAHOw/+OEmhZCRIZjq0Q2F5EsEbka6A90EpHvsVJQ/zBjzLebWAcANYFxkf+zl0INMspu4g3mtRK75OSccy4sFboE4Zxzbvc8QTjnnIvJE4RzzrmYPEE455yLyROEc865mDxBOFcCIpIb1cV4VjxnABaRJrFm6HQuLJXDDsC5JPO7qrYJOwjnyoKXIJyLAxH5SUQeF5E5IvK1iBwa2d5ERD6PrCswXkQaR7bvH1lnYHbklj9VRoqIDI6s9fCpiOwV2ptyFZ4nCOdKZq8CVUyXRe1bp6qtsVG4z0S2PQ+8HllX4C3gucj254AvVPUobM6n/BH8zYCBqtoKWAtcHPD7cW63fCS1cyUgIhtVtUaM7T8Bp6vq4siEiitUta6IrAIaqOr2yPZfVLWeiGQDaaq6Neo5mgDjIgvqICJ3A1VU9aHg35lzu/IShHPxo7u5XxJbo+7n4u2ELkSeIJyLn8uifk6N3J/CjuVALwe+jNwfD1wPf6zbvW9ZBelccfnViXMls5eIzIp6/LGq5nd1rR2ZCXYr0C2y7SZslbo7sRXrropsvwUYFJmJMxdLFr/gXALxNgjn4iDSBpGuqqvCjsW5ePEqJuecczF5CcI551xMXoJwzjkXkycI55xzMXmCcM45F5MnCOecczF5gnDOORfT/wNvr2jwhtIPGAAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "T8i7E7WpudlJ"
      },
      "source": [
        "# Conv1D + Bidirectional LSTM"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "f5VyocpquzZ8"
      },
      "source": [
        "import tensorflow as tf\n",
        "import matplotlib.pyplot as plt\n",
        "from keras.layers import Bidirectional, LSTM, Dense, Input, Dropout, Conv1D, SpatialDropout1D\n",
        "from keras.callbacks import ModelCheckpoint\n",
        "from tensorflow.keras.optimizers import Adam\n",
        "from tensorflow.keras.callbacks import ReduceLROnPlateau"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4jCVw1Usugb5",
        "outputId": "b2b53a8a-7dec-4f73-fd97-6c33fa62f1e7"
      },
      "source": [
        "sequence_input = Input(shape=(MAXLENGTH,), dtype='int32')\n",
        "embedding_sequences = embedding_layer(sequence_input)\n",
        "x = SpatialDropout1D(0.2)(embedding_sequences)\n",
        "x = Conv1D(64, 5, activation='relu')(x)\n",
        "x = Bidirectional(LSTM(64, dropout=0.2, recurrent_dropout=0.2))(x)\n",
        "x = Dense(512, activation='relu')(x)\n",
        "x = Dropout(0.5)(x)\n",
        "x = Dense(512, activation='relu')(x)\n",
        "outputs = Dense(1, activation='sigmoid')(x)\n",
        "model = tf.keras.Model(sequence_input, outputs)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Layer lstm will not use cuDNN kernel since it doesn't meet the cuDNN kernel criteria. It will use generic GPU kernel as fallback when running on GPU\n",
            "WARNING:tensorflow:Layer lstm will not use cuDNN kernel since it doesn't meet the cuDNN kernel criteria. It will use generic GPU kernel as fallback when running on GPU\n",
            "WARNING:tensorflow:Layer lstm will not use cuDNN kernel since it doesn't meet the cuDNN kernel criteria. It will use generic GPU kernel as fallback when running on GPU\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EA4cS0Ndvbzg"
      },
      "source": [
        "model.compile(optimizer=Adam(learning_rate=0.001), loss='binary_crossentropy',\n",
        "              metrics=[METRICS])\n",
        "ReduceLROnPlateau = ReduceLROnPlateau(factor=0.1,\n",
        "                                     min_lr = 0.01,\n",
        "                                     monitor = 'val_loss',\n",
        "                                     verbose = 1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "L_sNx0Upv0d4",
        "outputId": "c87b689e-0d69-4c19-c445-a00977cc08ec"
      },
      "source": [
        "history = model.fit(x_train, y_train, batch_size=BATCH_SIZE, epochs=EPOCHS,\n",
        "                    validation_data=(x_test, y_test), callbacks=[ReduceLROnPlateau])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/15\n",
            "218/218 [==============================] - 93s 249ms/step - loss: 0.3951 - tp: 18407.4612 - fp: 8262.4749 - tn: 74120.5982 - fn: 11841.5114 - accuracy: 0.8059 - precision: 0.6543 - recall: 0.5236 - auc: 0.8565 - prc: 0.6577 - f1_m: 0.5307 - val_loss: 0.2948 - val_tp: 12123.0000 - val_fp: 4681.0000 - val_tn: 36135.0000 - val_fn: 2780.0000 - val_accuracy: 0.8661 - val_precision: 0.7214 - val_recall: 0.8135 - val_auc: 0.9306 - val_prc: 0.8087 - val_f1_m: 0.7642\n",
            "Epoch 2/15\n",
            "218/218 [==============================] - 53s 242ms/step - loss: 0.2985 - tp: 23030.1689 - fp: 7816.9635 - tn: 74439.0046 - fn: 7345.9087 - accuracy: 0.8647 - precision: 0.7444 - recall: 0.7584 - auc: 0.9281 - prc: 0.8082 - f1_m: 0.7504 - val_loss: 0.2795 - val_tp: 12316.0000 - val_fp: 4440.0000 - val_tn: 36376.0000 - val_fn: 2587.0000 - val_accuracy: 0.8739 - val_precision: 0.7350 - val_recall: 0.8264 - val_auc: 0.9403 - val_prc: 0.8423 - val_f1_m: 0.7776\n",
            "Epoch 3/15\n",
            "218/218 [==============================] - 53s 242ms/step - loss: 0.2789 - tp: 23247.1598 - fp: 7068.7717 - tn: 75309.0274 - fn: 7007.0868 - accuracy: 0.8752 - precision: 0.7651 - recall: 0.7709 - auc: 0.9376 - prc: 0.8346 - f1_m: 0.7673 - val_loss: 0.2654 - val_tp: 11677.0000 - val_fp: 3320.0000 - val_tn: 37496.0000 - val_fn: 3226.0000 - val_accuracy: 0.8825 - val_precision: 0.7786 - val_recall: 0.7835 - val_auc: 0.9438 - val_prc: 0.8527 - val_f1_m: 0.7806\n",
            "Epoch 4/15\n",
            "218/218 [==============================] - 52s 239ms/step - loss: 0.2680 - tp: 23405.8493 - fp: 6491.5845 - tn: 75864.8539 - fn: 6869.7580 - accuracy: 0.8815 - precision: 0.7838 - recall: 0.7730 - auc: 0.9428 - prc: 0.8510 - f1_m: 0.7776 - val_loss: 0.2600 - val_tp: 11992.0000 - val_fp: 3427.0000 - val_tn: 37389.0000 - val_fn: 2911.0000 - val_accuracy: 0.8863 - val_precision: 0.7777 - val_recall: 0.8047 - val_auc: 0.9469 - val_prc: 0.8634 - val_f1_m: 0.7905\n",
            "Epoch 5/15\n",
            "218/218 [==============================] - 51s 234ms/step - loss: 0.2530 - tp: 23778.5251 - fp: 5983.0091 - tn: 76347.2009 - fn: 6523.3105 - accuracy: 0.8897 - precision: 0.8006 - recall: 0.7850 - auc: 0.9490 - prc: 0.8706 - f1_m: 0.7923 - val_loss: 0.2566 - val_tp: 11718.0000 - val_fp: 3065.0000 - val_tn: 37751.0000 - val_fn: 3185.0000 - val_accuracy: 0.8878 - val_precision: 0.7927 - val_recall: 0.7863 - val_auc: 0.9476 - val_prc: 0.8666 - val_f1_m: 0.7890\n",
            "Epoch 6/15\n",
            "218/218 [==============================] - 52s 238ms/step - loss: 0.2475 - tp: 23847.7169 - fp: 5855.8813 - tn: 76554.9178 - fn: 6373.5297 - accuracy: 0.8916 - precision: 0.8026 - recall: 0.7892 - auc: 0.9513 - prc: 0.8745 - f1_m: 0.7953 - val_loss: 0.2572 - val_tp: 11793.0000 - val_fp: 3153.0000 - val_tn: 37663.0000 - val_fn: 3110.0000 - val_accuracy: 0.8876 - val_precision: 0.7890 - val_recall: 0.7913 - val_auc: 0.9476 - val_prc: 0.8678 - val_f1_m: 0.7896\n",
            "Epoch 7/15\n",
            "218/218 [==============================] - 51s 235ms/step - loss: 0.2359 - tp: 24127.3562 - fp: 5591.2557 - tn: 76855.8447 - fn: 6057.5890 - accuracy: 0.8977 - precision: 0.8140 - recall: 0.8000 - auc: 0.9556 - prc: 0.8868 - f1_m: 0.8065 - val_loss: 0.2529 - val_tp: 11841.0000 - val_fp: 3059.0000 - val_tn: 37757.0000 - val_fn: 3062.0000 - val_accuracy: 0.8901 - val_precision: 0.7947 - val_recall: 0.7945 - val_auc: 0.9497 - val_prc: 0.8735 - val_f1_m: 0.7943\n",
            "Epoch 8/15\n",
            "218/218 [==============================] - 52s 239ms/step - loss: 0.2311 - tp: 24285.6986 - fp: 5393.1644 - tn: 76952.3973 - fn: 6000.7854 - accuracy: 0.8992 - precision: 0.8199 - recall: 0.8029 - auc: 0.9577 - prc: 0.8934 - f1_m: 0.8108 - val_loss: 0.2546 - val_tp: 11406.0000 - val_fp: 2579.0000 - val_tn: 38237.0000 - val_fn: 3497.0000 - val_accuracy: 0.8910 - val_precision: 0.8156 - val_recall: 0.7653 - val_auc: 0.9495 - val_prc: 0.8738 - val_f1_m: 0.7890\n",
            "Epoch 9/15\n",
            "218/218 [==============================] - 51s 236ms/step - loss: 0.2238 - tp: 24486.1689 - fp: 5137.4612 - tn: 77217.6119 - fn: 5790.8037 - accuracy: 0.9036 - precision: 0.8283 - recall: 0.8092 - auc: 0.9603 - prc: 0.8997 - f1_m: 0.8183 - val_loss: 0.2528 - val_tp: 12096.0000 - val_fp: 3316.0000 - val_tn: 37500.0000 - val_fn: 2807.0000 - val_accuracy: 0.8901 - val_precision: 0.7848 - val_recall: 0.8116 - val_auc: 0.9504 - val_prc: 0.8738 - val_f1_m: 0.7975\n",
            "Epoch 10/15\n",
            "218/218 [==============================] - 52s 237ms/step - loss: 0.2174 - tp: 24777.0868 - fp: 5051.1963 - tn: 77255.0457 - fn: 5548.7169 - accuracy: 0.9066 - precision: 0.8329 - recall: 0.8180 - auc: 0.9627 - prc: 0.9066 - f1_m: 0.8249 - val_loss: 0.2519 - val_tp: 12311.0000 - val_fp: 3491.0000 - val_tn: 37325.0000 - val_fn: 2592.0000 - val_accuracy: 0.8908 - val_precision: 0.7791 - val_recall: 0.8261 - val_auc: 0.9507 - val_prc: 0.8762 - val_f1_m: 0.8016\n",
            "Epoch 11/15\n",
            "218/218 [==============================] - 51s 234ms/step - loss: 0.2145 - tp: 24678.7169 - fp: 4955.1279 - tn: 77499.0776 - fn: 5499.1233 - accuracy: 0.9079 - precision: 0.8330 - recall: 0.8193 - auc: 0.9634 - prc: 0.9081 - f1_m: 0.8258 - val_loss: 0.2560 - val_tp: 12074.0000 - val_fp: 3194.0000 - val_tn: 37622.0000 - val_fn: 2829.0000 - val_accuracy: 0.8919 - val_precision: 0.7908 - val_recall: 0.8102 - val_auc: 0.9497 - val_prc: 0.8738 - val_f1_m: 0.8001\n",
            "Epoch 12/15\n",
            "218/218 [==============================] - 52s 236ms/step - loss: 0.2088 - tp: 25019.3607 - fp: 4856.0776 - tn: 77525.8950 - fn: 5230.7123 - accuracy: 0.9115 - precision: 0.8390 - recall: 0.8292 - auc: 0.9655 - prc: 0.9121 - f1_m: 0.8337 - val_loss: 0.2532 - val_tp: 12006.0000 - val_fp: 3132.0000 - val_tn: 37684.0000 - val_fn: 2897.0000 - val_accuracy: 0.8918 - val_precision: 0.7931 - val_recall: 0.8056 - val_auc: 0.9498 - val_prc: 0.8744 - val_f1_m: 0.7989\n",
            "Epoch 13/15\n",
            "218/218 [==============================] - 52s 240ms/step - loss: 0.2040 - tp: 25190.3562 - fp: 4769.7260 - tn: 77555.1826 - fn: 5116.7808 - accuracy: 0.9126 - precision: 0.8421 - recall: 0.8309 - auc: 0.9671 - prc: 0.9169 - f1_m: 0.8360 - val_loss: 0.2540 - val_tp: 11682.0000 - val_fp: 2777.0000 - val_tn: 38039.0000 - val_fn: 3221.0000 - val_accuracy: 0.8924 - val_precision: 0.8079 - val_recall: 0.7839 - val_auc: 0.9498 - val_prc: 0.8743 - val_f1_m: 0.7955\n",
            "Epoch 14/15\n",
            "218/218 [==============================] - 53s 242ms/step - loss: 0.2010 - tp: 25161.3973 - fp: 4702.1324 - tn: 77651.0365 - fn: 5117.4795 - accuracy: 0.9141 - precision: 0.8458 - recall: 0.8318 - auc: 0.9680 - prc: 0.9196 - f1_m: 0.8386 - val_loss: 0.2548 - val_tp: 12161.0000 - val_fp: 3303.0000 - val_tn: 37513.0000 - val_fn: 2742.0000 - val_accuracy: 0.8915 - val_precision: 0.7864 - val_recall: 0.8160 - val_auc: 0.9498 - val_prc: 0.8735 - val_f1_m: 0.8006\n",
            "Epoch 15/15\n",
            "218/218 [==============================] - 52s 240ms/step - loss: 0.1994 - tp: 25449.0046 - fp: 4673.9361 - tn: 77585.5205 - fn: 4923.5845 - accuracy: 0.9152 - precision: 0.8453 - recall: 0.8387 - auc: 0.9687 - prc: 0.9209 - f1_m: 0.8418 - val_loss: 0.2660 - val_tp: 12491.0000 - val_fp: 3779.0000 - val_tn: 37037.0000 - val_fn: 2412.0000 - val_accuracy: 0.8889 - val_precision: 0.7677 - val_recall: 0.8382 - val_auc: 0.9496 - val_prc: 0.8717 - val_f1_m: 0.8011\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "o5rJBmIAy3W6",
        "outputId": "bcfe1298-57c8-4f4a-8f42-cfe93d243762"
      },
      "source": [
        "plot_history(history)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9UnQ4hEHyj40"
      },
      "source": [
        "#Bidirectional LSTM"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oOZytKrfynN5",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "c5ec1d8e-a002-4914-a584-68bbcc1f7b2b"
      },
      "source": [
        "sequence_input = Input(shape=(MAXLENGTH,), dtype='int32')\n",
        "embedding_sequences = embedding_layer(sequence_input)\n",
        "x = SpatialDropout1D(0.2)(embedding_sequences)\n",
        "x = Bidirectional(LSTM(64, dropout=0.2, recurrent_dropout=0.2))(x)\n",
        "x = Dense(512, activation='relu')(x)\n",
        "x = Dropout(0.2)(x)\n",
        "x = Dense(512, activation='relu')(x)\n",
        "outputs = Dense(1, activation='sigmoid')(x)\n",
        "bidir_model = tf.keras.Model(sequence_input, outputs)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Layer lstm_10 will not use cuDNN kernel since it doesn't meet the cuDNN kernel criteria. It will use generic GPU kernel as fallback when running on GPU\n",
            "WARNING:tensorflow:Layer lstm_10 will not use cuDNN kernel since it doesn't meet the cuDNN kernel criteria. It will use generic GPU kernel as fallback when running on GPU\n",
            "WARNING:tensorflow:Layer lstm_10 will not use cuDNN kernel since it doesn't meet the cuDNN kernel criteria. It will use generic GPU kernel as fallback when running on GPU\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ry2oKMzGzQ0e"
      },
      "source": [
        "bidir_model.compile(optimizer=Adam(learning_rate=0.001), loss='binary_crossentropy',\n",
        "              metrics=[METRICS])\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "vSQvuE_z1ujN",
        "outputId": "8d96aba6-9a2b-4f30-b4e7-11eb77f7f2d7"
      },
      "source": [
        "bidir_model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"model_1\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_2 (InputLayer)         [(None, 30)]              0         \n",
            "_________________________________________________________________\n",
            "embedding (Embedding)        (None, 30, 300)           91539000  \n",
            "_________________________________________________________________\n",
            "spatial_dropout1d_1 (Spatial (None, 30, 300)           0         \n",
            "_________________________________________________________________\n",
            "bidirectional_1 (Bidirection (None, 128)               186880    \n",
            "_________________________________________________________________\n",
            "dense_3 (Dense)              (None, 512)               66048     \n",
            "_________________________________________________________________\n",
            "dropout_1 (Dropout)          (None, 512)               0         \n",
            "_________________________________________________________________\n",
            "dense_4 (Dense)              (None, 512)               262656    \n",
            "_________________________________________________________________\n",
            "dense_5 (Dense)              (None, 1)                 513       \n",
            "=================================================================\n",
            "Total params: 92,055,097\n",
            "Trainable params: 516,097\n",
            "Non-trainable params: 91,539,000\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Od2D_Ewdzhs5",
        "outputId": "262676fb-a009-4cf5-9f37-ee8bc54d3556"
      },
      "source": [
        "bidir_history = bidir_model.fit(x_train, y_train, batch_size=512, epochs=EPOCHS,\n",
        "                    validation_data=(x_test, y_test))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/15\n",
            "436/436 [==============================] - 110s 239ms/step - loss: 0.2668 - tp: 35774.8375 - fp: 10303.4256 - tn: 112516.4737 - fn: 9249.4485 - accuracy: 0.8839 - precision: 0.7753 - recall: 0.7991 - auc: 0.9449 - prc: 0.8549 - f1_m: 0.7758 - val_loss: 0.2499 - val_tp: 12024.0000 - val_fp: 3304.0000 - val_tn: 37512.0000 - val_fn: 2879.0000 - val_accuracy: 0.8890 - val_precision: 0.7844 - val_recall: 0.8068 - val_auc: 0.9504 - val_prc: 0.8732 - val_f1_m: 0.7949\n",
            "Epoch 2/15\n",
            "436/436 [==============================] - 99s 227ms/step - loss: 0.2521 - tp: 23832.5240 - fp: 6287.1167 - tn: 75719.2449 - fn: 6286.2998 - accuracy: 0.8880 - precision: 0.7905 - recall: 0.7928 - auc: 0.9494 - prc: 0.8683 - f1_m: 0.7910 - val_loss: 0.2467 - val_tp: 12239.0000 - val_fp: 3381.0000 - val_tn: 37435.0000 - val_fn: 2664.0000 - val_accuracy: 0.8915 - val_precision: 0.7835 - val_recall: 0.8212 - val_auc: 0.9525 - val_prc: 0.8783 - val_f1_m: 0.8014\n",
            "Epoch 3/15\n",
            "436/436 [==============================] - 104s 239ms/step - loss: 0.2461 - tp: 23928.2540 - fp: 5959.2082 - tn: 76047.8627 - fn: 6189.8604 - accuracy: 0.8908 - precision: 0.7991 - recall: 0.7929 - auc: 0.9519 - prc: 0.8757 - f1_m: 0.7954 - val_loss: 0.2430 - val_tp: 12338.0000 - val_fp: 3415.0000 - val_tn: 37401.0000 - val_fn: 2565.0000 - val_accuracy: 0.8927 - val_precision: 0.7832 - val_recall: 0.8279 - val_auc: 0.9535 - val_prc: 0.8829 - val_f1_m: 0.8043\n",
            "Epoch 4/15\n",
            "436/436 [==============================] - 103s 236ms/step - loss: 0.2373 - tp: 24167.9199 - fp: 5752.8421 - tn: 76246.9382 - fn: 5957.4851 - accuracy: 0.8952 - precision: 0.8065 - recall: 0.8024 - auc: 0.9554 - prc: 0.8851 - f1_m: 0.8038 - val_loss: 0.2398 - val_tp: 11571.0000 - val_fp: 2440.0000 - val_tn: 38376.0000 - val_fn: 3332.0000 - val_accuracy: 0.8964 - val_precision: 0.8259 - val_recall: 0.7764 - val_auc: 0.9551 - val_prc: 0.8882 - val_f1_m: 0.7999\n",
            "Epoch 5/15\n",
            "436/436 [==============================] - 103s 235ms/step - loss: 0.2289 - tp: 24271.2357 - fp: 5461.3341 - tn: 76538.4371 - fn: 5854.1785 - accuracy: 0.8994 - precision: 0.8179 - recall: 0.8047 - auc: 0.9585 - prc: 0.8933 - f1_m: 0.8106 - val_loss: 0.2388 - val_tp: 12621.0000 - val_fp: 3358.0000 - val_tn: 37458.0000 - val_fn: 2282.0000 - val_accuracy: 0.8988 - val_precision: 0.7898 - val_recall: 0.8469 - val_auc: 0.9561 - val_prc: 0.8911 - val_f1_m: 0.8168\n",
            "Epoch 6/15\n",
            "436/436 [==============================] - 100s 230ms/step - loss: 0.2245 - tp: 24574.8696 - fp: 5365.2220 - tn: 76534.0389 - fn: 5651.0549 - accuracy: 0.9017 - precision: 0.8201 - recall: 0.8140 - auc: 0.9602 - prc: 0.8986 - f1_m: 0.8163 - val_loss: 0.2381 - val_tp: 12765.0000 - val_fp: 3630.0000 - val_tn: 37186.0000 - val_fn: 2138.0000 - val_accuracy: 0.8965 - val_precision: 0.7786 - val_recall: 0.8565 - val_auc: 0.9570 - val_prc: 0.8934 - val_f1_m: 0.8153\n",
            "Epoch 7/15\n",
            "436/436 [==============================] - 100s 229ms/step - loss: 0.2157 - tp: 24642.8330 - fp: 5100.9817 - tn: 76929.7300 - fn: 5451.6407 - accuracy: 0.9065 - precision: 0.8293 - recall: 0.8211 - auc: 0.9632 - prc: 0.9059 - f1_m: 0.8244 - val_loss: 0.2396 - val_tp: 12642.0000 - val_fp: 3430.0000 - val_tn: 37386.0000 - val_fn: 2261.0000 - val_accuracy: 0.8979 - val_precision: 0.7866 - val_recall: 0.8483 - val_auc: 0.9564 - val_prc: 0.8909 - val_f1_m: 0.8158\n",
            "Epoch 8/15\n",
            "436/436 [==============================] - 99s 228ms/step - loss: 0.2120 - tp: 24717.2403 - fp: 4883.5400 - tn: 77079.6430 - fn: 5444.7620 - accuracy: 0.9077 - precision: 0.8353 - recall: 0.8196 - auc: 0.9646 - prc: 0.9103 - f1_m: 0.8266 - val_loss: 0.2327 - val_tp: 12226.0000 - val_fp: 2840.0000 - val_tn: 37976.0000 - val_fn: 2677.0000 - val_accuracy: 0.9010 - val_precision: 0.8115 - val_recall: 0.8204 - val_auc: 0.9574 - val_prc: 0.8945 - val_f1_m: 0.8154\n",
            "Epoch 9/15\n",
            "436/436 [==============================] - 99s 226ms/step - loss: 0.2076 - tp: 24929.2952 - fp: 4857.8330 - tn: 77100.0801 - fn: 5237.9771 - accuracy: 0.9099 - precision: 0.8371 - recall: 0.8262 - auc: 0.9660 - prc: 0.9131 - f1_m: 0.8312 - val_loss: 0.2359 - val_tp: 12383.0000 - val_fp: 2933.0000 - val_tn: 37883.0000 - val_fn: 2520.0000 - val_accuracy: 0.9021 - val_precision: 0.8085 - val_recall: 0.8309 - val_auc: 0.9576 - val_prc: 0.8972 - val_f1_m: 0.8190\n",
            "Epoch 10/15\n",
            "436/436 [==============================] - 101s 232ms/step - loss: 0.2017 - tp: 25049.2929 - fp: 4696.6270 - tn: 77303.2952 - fn: 5075.9703 - accuracy: 0.9131 - precision: 0.8425 - recall: 0.8306 - auc: 0.9678 - prc: 0.9172 - f1_m: 0.8361 - val_loss: 0.2368 - val_tp: 12494.0000 - val_fp: 3078.0000 - val_tn: 37738.0000 - val_fn: 2409.0000 - val_accuracy: 0.9015 - val_precision: 0.8023 - val_recall: 0.8384 - val_auc: 0.9577 - val_prc: 0.8963 - val_f1_m: 0.8194\n",
            "Epoch 11/15\n",
            "436/436 [==============================] - 103s 235ms/step - loss: 0.1985 - tp: 25146.5881 - fp: 4740.2357 - tn: 77244.2517 - fn: 4994.1098 - accuracy: 0.9134 - precision: 0.8413 - recall: 0.8354 - auc: 0.9689 - prc: 0.9205 - f1_m: 0.8379 - val_loss: 0.2339 - val_tp: 12426.0000 - val_fp: 3021.0000 - val_tn: 37795.0000 - val_fn: 2477.0000 - val_accuracy: 0.9013 - val_precision: 0.8044 - val_recall: 0.8338 - val_auc: 0.9571 - val_prc: 0.8952 - val_f1_m: 0.8183\n",
            "Epoch 12/15\n",
            "436/436 [==============================] - 102s 233ms/step - loss: 0.1957 - tp: 25250.5767 - fp: 4570.0412 - tn: 77379.2906 - fn: 4925.2769 - accuracy: 0.9153 - precision: 0.8473 - recall: 0.8355 - auc: 0.9698 - prc: 0.9225 - f1_m: 0.8407 - val_loss: 0.2352 - val_tp: 12434.0000 - val_fp: 3066.0000 - val_tn: 37750.0000 - val_fn: 2469.0000 - val_accuracy: 0.9007 - val_precision: 0.8022 - val_recall: 0.8343 - val_auc: 0.9573 - val_prc: 0.8958 - val_f1_m: 0.8174\n",
            "Epoch 13/15\n",
            "436/436 [==============================] - 103s 237ms/step - loss: 0.1940 - tp: 25343.0870 - fp: 4583.1465 - tn: 77422.5515 - fn: 4776.4005 - accuracy: 0.9165 - precision: 0.8467 - recall: 0.8418 - auc: 0.9704 - prc: 0.9234 - f1_m: 0.8436 - val_loss: 0.2411 - val_tp: 12591.0000 - val_fp: 3161.0000 - val_tn: 37655.0000 - val_fn: 2312.0000 - val_accuracy: 0.9018 - val_precision: 0.7993 - val_recall: 0.8449 - val_auc: 0.9579 - val_prc: 0.8959 - val_f1_m: 0.8209\n",
            "Epoch 14/15\n",
            "436/436 [==============================] - 102s 234ms/step - loss: 0.1891 - tp: 25427.4577 - fp: 4497.1556 - tn: 77507.0481 - fn: 4693.5240 - accuracy: 0.9180 - precision: 0.8490 - recall: 0.8438 - auc: 0.9718 - prc: 0.9270 - f1_m: 0.8459 - val_loss: 0.2457 - val_tp: 11544.0000 - val_fp: 2064.0000 - val_tn: 38752.0000 - val_fn: 3359.0000 - val_accuracy: 0.9027 - val_precision: 0.8483 - val_recall: 0.7746 - val_auc: 0.9564 - val_prc: 0.8963 - val_f1_m: 0.8091\n",
            "Epoch 15/15\n",
            "436/436 [==============================] - 101s 232ms/step - loss: 0.1853 - tp: 25536.5103 - fp: 4359.0664 - tn: 77569.7551 - fn: 4659.8535 - accuracy: 0.9200 - precision: 0.8562 - recall: 0.8452 - auc: 0.9730 - prc: 0.9309 - f1_m: 0.8501 - val_loss: 0.2397 - val_tp: 12292.0000 - val_fp: 2761.0000 - val_tn: 38055.0000 - val_fn: 2611.0000 - val_accuracy: 0.9036 - val_precision: 0.8166 - val_recall: 0.8248 - val_auc: 0.9577 - val_prc: 0.8976 - val_f1_m: 0.8200\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "pcD8oJTP1FFm",
        "outputId": "fb66f19a-d6f3-4714-a8c3-51027c0e6ffc"
      },
      "source": [
        "plot_history(bidir_history)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kgDPc0XJUMH1"
      },
      "source": [
        "# Tuning - Bidirectional LSTM"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "E5ZxLEb9UQRZ",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "82c7df29-6ad1-4ab4-cef0-f550896ee54a"
      },
      "source": [
        "sequence_input = Input(shape=(MAXLENGTH,), dtype='int32')\n",
        "embedding_sequences = embedding_layer(sequence_input)\n",
        "x = SpatialDropout1D(0.2)(embedding_sequences)\n",
        "x = Bidirectional(LSTM(128, dropout=0.1, recurrent_dropout=0.2))(x)\n",
        "x = Dense(50, activation='relu')(x)\n",
        "x = Dropout(0.1)(x)\n",
        "x = Dense(25, activation='relu')(x)\n",
        "outputs = Dense(1, activation='sigmoid')(x)\n",
        "stacked_bidir_model = tf.keras.Model(sequence_input, outputs)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Layer lstm_11 will not use cuDNN kernel since it doesn't meet the cuDNN kernel criteria. It will use generic GPU kernel as fallback when running on GPU\n",
            "WARNING:tensorflow:Layer lstm_11 will not use cuDNN kernel since it doesn't meet the cuDNN kernel criteria. It will use generic GPU kernel as fallback when running on GPU\n",
            "WARNING:tensorflow:Layer lstm_11 will not use cuDNN kernel since it doesn't meet the cuDNN kernel criteria. It will use generic GPU kernel as fallback when running on GPU\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4m4LkNDvWmew",
        "outputId": "482a7cca-9405-41d9-8530-86424e186b6a"
      },
      "source": [
        "stacked_bidir_model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"model_9\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_10 (InputLayer)        [(None, 30)]              0         \n",
            "_________________________________________________________________\n",
            "embedding (Embedding)        (None, 30, 300)           91539000  \n",
            "_________________________________________________________________\n",
            "spatial_dropout1d_9 (Spatial (None, 30, 300)           0         \n",
            "_________________________________________________________________\n",
            "bidirectional_4 (Bidirection (None, 256)               439296    \n",
            "_________________________________________________________________\n",
            "dense_27 (Dense)             (None, 50)                12850     \n",
            "_________________________________________________________________\n",
            "dropout_9 (Dropout)          (None, 50)                0         \n",
            "_________________________________________________________________\n",
            "dense_28 (Dense)             (None, 25)                1275      \n",
            "_________________________________________________________________\n",
            "dense_29 (Dense)             (None, 1)                 26        \n",
            "=================================================================\n",
            "Total params: 91,992,447\n",
            "Trainable params: 453,447\n",
            "Non-trainable params: 91,539,000\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lvYyHO6PUZYi"
      },
      "source": [
        "stacked_bidir_model.compile(optimizer=Adam(learning_rate=0.001), loss='binary_crossentropy',\n",
        "              metrics=[METRICS])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "GpOMTwJOUdNq",
        "outputId": "2da49d20-418d-49a2-c17d-aae998e814a2"
      },
      "source": [
        "stacked_bidir_history = stacked_bidir_model.fit(x_train, y_train, batch_size=512, epochs=10,\n",
        "                    validation_data=(x_test, y_test))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/10\n",
            "436/436 [==============================] - 106s 229ms/step - loss: 0.3775 - tp: 32169.7231 - fp: 11825.1991 - tn: 111038.1854 - fn: 12811.0778 - accuracy: 0.8552 - precision: 0.7360 - recall: 0.7165 - auc: 0.9081 - prc: 0.7782 - f1_m: 0.6083 - val_loss: 0.2809 - val_tp: 11600.0000 - val_fp: 3644.0000 - val_tn: 37172.0000 - val_fn: 3303.0000 - val_accuracy: 0.8753 - val_precision: 0.7610 - val_recall: 0.7784 - val_auc: 0.9379 - val_prc: 0.8334 - val_f1_m: 0.7686\n",
            "Epoch 2/10\n",
            "436/436 [==============================] - 99s 227ms/step - loss: 0.2899 - tp: 22846.0618 - fp: 7127.0275 - tn: 74808.5812 - fn: 7343.5149 - accuracy: 0.8695 - precision: 0.7591 - recall: 0.7555 - auc: 0.9325 - prc: 0.8226 - f1_m: 0.7565 - val_loss: 0.2588 - val_tp: 11596.0000 - val_fp: 3109.0000 - val_tn: 37707.0000 - val_fn: 3307.0000 - val_accuracy: 0.8849 - val_precision: 0.7886 - val_recall: 0.7781 - val_auc: 0.9466 - val_prc: 0.8600 - val_f1_m: 0.7826\n",
            "Epoch 3/10\n",
            "436/436 [==============================] - 99s 227ms/step - loss: 0.2607 - tp: 23334.4279 - fp: 6227.0778 - tn: 75755.5057 - fn: 6808.1739 - accuracy: 0.8842 - precision: 0.7900 - recall: 0.7744 - auc: 0.9457 - prc: 0.8580 - f1_m: 0.7815 - val_loss: 0.2481 - val_tp: 11906.0000 - val_fp: 3088.0000 - val_tn: 37728.0000 - val_fn: 2997.0000 - val_accuracy: 0.8908 - val_precision: 0.7941 - val_recall: 0.7989 - val_auc: 0.9510 - val_prc: 0.8747 - val_f1_m: 0.7958\n",
            "Epoch 4/10\n",
            "436/436 [==============================] - 99s 226ms/step - loss: 0.2468 - tp: 23809.3478 - fp: 5836.8146 - tn: 76148.9359 - fn: 6330.0870 - accuracy: 0.8907 - precision: 0.8024 - recall: 0.7887 - auc: 0.9517 - prc: 0.8761 - f1_m: 0.7947 - val_loss: 0.2394 - val_tp: 11891.0000 - val_fp: 2744.0000 - val_tn: 38072.0000 - val_fn: 3012.0000 - val_accuracy: 0.8967 - val_precision: 0.8125 - val_recall: 0.7979 - val_auc: 0.9543 - val_prc: 0.8848 - val_f1_m: 0.8044\n",
            "Epoch 5/10\n",
            "436/436 [==============================] - 99s 227ms/step - loss: 0.2340 - tp: 24398.0618 - fp: 5488.0069 - tn: 76360.6453 - fn: 5878.4714 - accuracy: 0.8983 - precision: 0.8162 - recall: 0.8054 - auc: 0.9568 - prc: 0.8890 - f1_m: 0.8102 - val_loss: 0.2415 - val_tp: 12578.0000 - val_fp: 3447.0000 - val_tn: 37369.0000 - val_fn: 2325.0000 - val_accuracy: 0.8964 - val_precision: 0.7849 - val_recall: 0.8440 - val_auc: 0.9555 - val_prc: 0.8878 - val_f1_m: 0.8129\n",
            "Epoch 6/10\n",
            "436/436 [==============================] - 97s 222ms/step - loss: 0.2213 - tp: 24625.7094 - fp: 5157.6705 - tn: 76751.0481 - fn: 5590.7574 - accuracy: 0.9043 - precision: 0.8272 - recall: 0.8168 - auc: 0.9614 - prc: 0.9019 - f1_m: 0.8212 - val_loss: 0.2364 - val_tp: 11717.0000 - val_fp: 2443.0000 - val_tn: 38373.0000 - val_fn: 3186.0000 - val_accuracy: 0.8990 - val_precision: 0.8275 - val_recall: 0.7862 - val_auc: 0.9558 - val_prc: 0.8893 - val_f1_m: 0.8058\n",
            "Epoch 7/10\n",
            "436/436 [==============================] - 97s 223ms/step - loss: 0.2106 - tp: 24754.0709 - fp: 4830.7986 - tn: 77164.6430 - fn: 5375.6728 - accuracy: 0.9093 - precision: 0.8377 - recall: 0.8220 - auc: 0.9650 - prc: 0.9112 - f1_m: 0.8290 - val_loss: 0.2380 - val_tp: 12358.0000 - val_fp: 3060.0000 - val_tn: 37756.0000 - val_fn: 2545.0000 - val_accuracy: 0.8994 - val_precision: 0.8015 - val_recall: 0.8292 - val_auc: 0.9559 - val_prc: 0.8921 - val_f1_m: 0.8146\n",
            "Epoch 8/10\n",
            "436/436 [==============================] - 98s 224ms/step - loss: 0.1999 - tp: 24994.0481 - fp: 4645.4119 - tn: 77401.8421 - fn: 5083.8833 - accuracy: 0.9135 - precision: 0.8431 - recall: 0.8312 - auc: 0.9684 - prc: 0.9187 - f1_m: 0.8364 - val_loss: 0.2380 - val_tp: 11723.0000 - val_fp: 2366.0000 - val_tn: 38450.0000 - val_fn: 3180.0000 - val_accuracy: 0.9005 - val_precision: 0.8321 - val_recall: 0.7866 - val_auc: 0.9557 - val_prc: 0.8919 - val_f1_m: 0.8081\n",
            "Epoch 9/10\n",
            "436/436 [==============================] - 98s 226ms/step - loss: 0.1936 - tp: 25104.3776 - fp: 4458.1716 - tn: 77653.5400 - fn: 4909.0961 - accuracy: 0.9167 - precision: 0.8502 - recall: 0.8356 - auc: 0.9704 - prc: 0.9237 - f1_m: 0.8424 - val_loss: 0.2459 - val_tp: 12241.0000 - val_fp: 2854.0000 - val_tn: 37962.0000 - val_fn: 2662.0000 - val_accuracy: 0.9010 - val_precision: 0.8109 - val_recall: 0.8214 - val_auc: 0.9561 - val_prc: 0.8918 - val_f1_m: 0.8156\n",
            "Epoch 10/10\n",
            "436/436 [==============================] - 99s 227ms/step - loss: 0.1840 - tp: 25619.2586 - fp: 4190.4416 - tn: 77750.6568 - fn: 4564.8284 - accuracy: 0.9225 - precision: 0.8606 - recall: 0.8500 - auc: 0.9734 - prc: 0.9320 - f1_m: 0.8546 - val_loss: 0.2431 - val_tp: 12368.0000 - val_fp: 3007.0000 - val_tn: 37809.0000 - val_fn: 2535.0000 - val_accuracy: 0.9005 - val_precision: 0.8044 - val_recall: 0.8299 - val_auc: 0.9557 - val_prc: 0.8916 - val_f1_m: 0.8163\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YCo1HdT0fwuo"
      },
      "source": [
        "plot_history(stacked_bidir_history)"
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}