NN.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Team-26_ProjC2_Code.ipynb",
      "provenance": [],
      "collapsed_sections": [
        "2iUtQ67QB1hz",
        "coixU1LKB6sl",
        "i1POcSSRB0fo",
        "eXFWluEMDYhg",
        "RnRUei0lDro6",
        "b0GP5oKXEd0D",
        "sQUtbHqNFHbN",
        "6xAr2KjDJ2YA",
        "ZKTxxTj4U26h",
        "GYDAM7efjM4Q",
        "zRyRFUfmYaKs",
        "yElkQwiETANJ",
        "vU67zmoIXESr",
        "gbihG0R0ns3c",
        "f7D7j11EiZJq",
        "Qs35ZE7oL3du",
        "pqaOgdwD_Ofu"
      ],
      "toc_visible": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2iUtQ67QB1hz"
      },
      "source": [
        "#Import Libraries and mount the drive to access the drive"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "c6w1MAoBGTU0"
      },
      "source": [
        "import pandas as pd\n",
        "import glob\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn import metrics\n",
        "import numpy as np\n",
        "from keras.models import Sequential\n",
        "from keras.layers import Dense\n",
        "from keras.layers import Flatten, Reshape\n",
        "from keras.utils import to_categorical\n",
        "from keras.layers.convolutional import Conv1D\n",
        "from keras.layers.convolutional import MaxPooling1D\n",
        "from keras.optimizers import SGD, Adam, RMSprop\n",
        "from keras.layers import Dropout, InputLayer, LSTM\n",
        "from keras.layers import Bidirectional, BatchNormalization\n",
        "from keras.layers import TimeDistributed\n",
        "from keras.layers import Conv2D, MaxPooling2D, Flatten, Dense, Activation, BatchNormalization, Dropout, UpSampling1D, AveragePooling1D"
      ],
      "execution_count": 39,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "185ODFrbGWt0"
      },
      "source": [
        "import os\n",
        "from google.colab import drive\n",
        "drive.mount('/content/gdrive/')\n",
        "os.chdir(\"./gdrive/MyDrive/Neural Competition Project\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "id": "pF5C8C1NGpN5",
        "outputId": "bddaf437-3588-4f2e-d54a-a2e7eb3d7503"
      },
      "source": [
        "pwd!"
      ],
      "execution_count": 41,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'/content/gdrive/My Drive/Neural Competition Project'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 41
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "coixU1LKB6sl"
      },
      "source": [
        "# Data Preprocessing before Feeding to Neural Network Model\n",
        "\n",
        "The interpolated csv file is created by merging all the subject files with the y data interpolated to match the acc and gyrometer x sampling rate"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3KTco3NDGtDw"
      },
      "source": [
        "df=pd.read_csv('Interpolated_combined.csv')\n",
        "walking = df[df.label==0].count()[0]\n",
        "downstairs = df[df.label==1].count()[0]\n",
        "upstairs = df[df.label==2].count()[0]\n",
        "grass = df[df.label==3].count()[0]\n",
        "print(walking, downstairs,upstairs, grass)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8lzwNSWnG280"
      },
      "source": [
        "# Partitioning the x and y data from the df.\n",
        "# Dropping the time and label columns from the x dataframe \n",
        "y_train = df.label\n",
        "x_train= df.drop('label', axis=1)\n",
        "x_train= x_train.drop('time', axis=1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "JFTBif0EG7Cg",
        "outputId": "1a891237-ef1e-48b4-fb92-7c5a06920d50"
      },
      "source": [
        "print(y_train)"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0          0.0\n",
            "1          0.0\n",
            "2          0.0\n",
            "3          0.0\n",
            "4          0.0\n",
            "          ... \n",
            "1341612    0.0\n",
            "1341613    0.0\n",
            "1341614    0.0\n",
            "1341615    0.0\n",
            "1341616    0.0\n",
            "Name: label, Length: 1341617, dtype: float64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "i1POcSSRB0fo"
      },
      "source": [
        "## One hot encoding for y labels"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4_YEOCPTHE9B"
      },
      "source": [
        "y_train_cat = to_categorical(y_train,4).reshape(-1,4)\n",
        "df_y = pd.DataFrame(y_train_cat, columns = ['l1','l2','l3','l4'])"
      ],
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        },
        "id": "nXbZsiljHLFv",
        "outputId": "743fc742-1db9-4e80-ac6e-7c56b528c7f3"
      },
      "source": [
        "df_y"
      ],
      "execution_count": 7,
      "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>l1</th>\n",
              "      <th>l2</th>\n",
              "      <th>l3</th>\n",
              "      <th>l4</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</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",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1341612</th>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1341613</th>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1341614</th>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1341615</th>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1341616</th>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>1341617 rows × 4 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "          l1   l2   l3   l4\n",
              "0        1.0  0.0  0.0  0.0\n",
              "1        1.0  0.0  0.0  0.0\n",
              "2        1.0  0.0  0.0  0.0\n",
              "3        1.0  0.0  0.0  0.0\n",
              "4        1.0  0.0  0.0  0.0\n",
              "...      ...  ...  ...  ...\n",
              "1341612  1.0  0.0  0.0  0.0\n",
              "1341613  1.0  0.0  0.0  0.0\n",
              "1341614  1.0  0.0  0.0  0.0\n",
              "1341615  1.0  0.0  0.0  0.0\n",
              "1341616  1.0  0.0  0.0  0.0\n",
              "\n",
              "[1341617 rows x 4 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "U91heLhmG9qp",
        "outputId": "a0fcb3bb-1e46-4ff0-881e-ce1922fb8a82"
      },
      "source": [
        "print(x_train)"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "            acc_x     acc_y     acc_z    gyro_x    gyro_y    gyro_z\n",
            "0        4.186920  8.344455  2.908057  0.005771 -0.004480 -0.003345\n",
            "1        4.544637  8.408659  2.890000  0.007967  0.022412  0.001159\n",
            "2        4.849308  8.411614  2.900692  0.027778 -0.010670 -0.014223\n",
            "3        4.509190  8.118649  2.847298  0.021577 -0.045498 -0.021111\n",
            "4        4.226515  8.273807  2.851742  0.012534  0.000445 -0.016830\n",
            "...           ...       ...       ...       ...       ...       ...\n",
            "1341612  2.072244  8.908878 -3.500000  0.001351  0.001191  0.001031\n",
            "1341613  2.085123  8.915123 -3.520000  0.001918 -0.001147  0.000000\n",
            "1341614  2.083774  8.910000 -3.538981 -0.002015 -0.004099  0.001042\n",
            "1341615  2.111447  8.908553 -3.535724  0.000183 -0.001673  0.001856\n",
            "1341616  2.087730  8.900000 -3.518865  0.000052 -0.001267  0.000000\n",
            "\n",
            "[1341617 rows x 6 columns]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "eXFWluEMDYhg"
      },
      "source": [
        "## Standardize the Train Data\n",
        "\n",
        "Using Standard Scalar function to normalize the training data."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qfBaAxiHH8TC"
      },
      "source": [
        "from sklearn.preprocessing import StandardScaler\n",
        "scaler = StandardScaler()\n",
        "df_x_scaled = scaler.fit_transform(x_train)"
      ],
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "XNTHqdWNIOiJ",
        "outputId": "05e4e64a-c617-4881-e4cb-30fcef12e49d"
      },
      "source": [
        "df_x_scale = pd.DataFrame(df_x_scaled, columns = ['ax','ay','az','gx','gy','gz'])\n",
        "print(df_x_scale)"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "               ax        ay        az        gx        gy        gz\n",
            "0        0.620543  0.045290 -0.567099  0.000351 -0.018272  0.083457\n",
            "1        0.682938  0.054657 -0.570337  0.001033 -0.001082  0.089309\n",
            "2        0.736081  0.055088 -0.568419  0.007191 -0.022229  0.069323\n",
            "3        0.676755  0.012346 -0.577995  0.005263 -0.044493  0.060374\n",
            "4        0.627449  0.034983 -0.577198  0.002453 -0.015124  0.065937\n",
            "...           ...       ...       ...       ...       ...       ...\n",
            "1341612  0.251684  0.127635 -1.716335 -0.001023 -0.014647  0.089143\n",
            "1341613  0.253931  0.128546 -1.719921 -0.000847 -0.016142  0.087803\n",
            "1341614  0.253695  0.127799 -1.723325 -0.002069 -0.018029  0.089157\n",
            "1341615  0.258522  0.127588 -1.722741 -0.001386 -0.016478  0.090215\n",
            "1341616  0.254385  0.126340 -1.719718 -0.001427 -0.016218  0.087803\n",
            "\n",
            "[1341617 rows x 6 columns]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RnRUei0lDro6"
      },
      "source": [
        "## Creating windows in the training data. \n",
        "\n",
        "We are dividing our entire test dataset in multiple window frames. We have used the window size of 40, hence 40 samples of data will be considered as a single frame and will help to identify patterns. \n",
        "\n",
        "We will also creating window frames on our y labels which are one hot-encoded."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rQpg_q_ZIPqy"
      },
      "source": [
        "def define_windows(df_x, df_y, windowsize=120):\n",
        "  window_x=[]\n",
        "  window_y=[]\n",
        "  for i in range(0, len(df_x)-windowsize, windowsize+1):\n",
        "    window_x.append([df_x.values[i:i+windowsize]])\n",
        "    window_y.append([df_y.values[i:i+windowsize]])\n",
        "    #window_y.append([df_y[['l1','l2','l3','l4']].values[i:i+windowsize]])\n",
        "  window_x=np.array(window_x).reshape(-1,windowsize,6)\n",
        "  window_y=np.array(window_y).reshape(-1,windowsize,4)\n",
        "\n",
        "  return window_x, window_y"
      ],
      "execution_count": 11,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "pefDUfI2JKmi",
        "outputId": "d85b127b-dcfb-48dc-9d7b-6b919efc952b"
      },
      "source": [
        "wsize = 40\n",
        "x_window, y_window = define_windows(df_x_scale, df_y, wsize)\n",
        "print(x_window.shape, y_window.shape)"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(32722, 40, 6) (32722, 40, 4)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "b0GP5oKXEd0D"
      },
      "source": [
        "## Splitting the data into training and validation dataset\n",
        "\n",
        "We are splitting the data into 90% training and 10% validation dataset. We are not using to upssample our minority classes. Hence, we want to maximize our training data to include as much samples from the minority classes."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "z3jE5dSFJd5Z",
        "outputId": "3076a2dd-cf9e-4be2-9165-81ed081702cc"
      },
      "source": [
        "x_train,x_val,y_train,y_val=train_test_split(x_window, y_window, test_size=0.1,random_state=123)\n",
        "\n",
        "print(len(x_train), len(x_val))\n",
        "print(len(y_train), len(y_val))"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "29449 3273\n",
            "29449 3273\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sQUtbHqNFHbN"
      },
      "source": [
        "## Helper function to plot the data curve"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Aq4DNKCDJoP1"
      },
      "source": [
        "import matplotlib.pyplot as plt\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='validation')\n",
        "    plt.ylabel('Loss')\n",
        "    plt.xlabel('Epoch')\n",
        "    plt.legend(['Train', 'Validation'])\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='validation')\n",
        "    plt.ylabel('Accuracy')\n",
        "    plt.xlabel('Epoch')\n",
        "    plt.legend(['Train', 'Validation'])\n",
        "    plt.show()"
      ],
      "execution_count": 14,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ESnJydyZM9fy"
      },
      "source": [
        "# Function for changing the categorical prediction to numerical prediction\n",
        "def flatten_pred(y_pred):\n",
        "  y_hat=[]\n",
        "  for i in y_pred:\n",
        "    for j in i:\n",
        "      y_hat.append(np.argmax(j))\n",
        "  \n",
        "  return y_hat"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6xAr2KjDJ2YA"
      },
      "source": [
        "# CNN"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yDU682d-LDBQ"
      },
      "source": [
        "def create_1dCNN():\n",
        "  model = Sequential()\n",
        "  model.add(Conv1D(16, kernel_size=3, activation='relu', padding='same', input_shape=(40,6)))\n",
        "  #model.add(MaxPooling1D(2))\n",
        "  model.add(AveragePooling1D(2))\n",
        "  model.add(Dropout(0.2))\n",
        "\n",
        "  model.add(Conv1D(32, kernel_size=3, activation='relu', padding='same'))\n",
        "  #model.add(MaxPooling1D(2))\n",
        "  model.add(AveragePooling1D(2))\n",
        "  model.add(Dropout(0.2))\n",
        "\n",
        "  model.add(Conv1D(32, kernel_size=3, activation='relu', padding='same'))\n",
        "  model.add(UpSampling1D(size=2))\n",
        "  model.add(Dropout(0.2))\n",
        "\n",
        "  model.add(Conv1D(16, kernel_size=3, activation='relu', padding='same'))\n",
        "  model.add(UpSampling1D(size=2))\n",
        "  model.add(Dropout(0.2))\n",
        "\n",
        "  model.add(Conv1D(4, kernel_size=3, activation='softmax', padding='same'))\n",
        "  model.compile(optimizer=Adam(learning_rate=0.001),  loss='categorical_crossentropy', metrics = ['accuracy'])\n",
        "  return model\n",
        "  #return model"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_4k1ds5vRLEP",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "b95fe043-f931-4cfe-a0e4-ace9bbdd7337"
      },
      "source": [
        "print(cnn_model.summary())"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv1d (Conv1D)              (None, 40, 16)            304       \n",
            "_________________________________________________________________\n",
            "average_pooling1d (AveragePo (None, 20, 16)            0         \n",
            "_________________________________________________________________\n",
            "dropout (Dropout)            (None, 20, 16)            0         \n",
            "_________________________________________________________________\n",
            "conv1d_1 (Conv1D)            (None, 20, 32)            1568      \n",
            "_________________________________________________________________\n",
            "average_pooling1d_1 (Average (None, 10, 32)            0         \n",
            "_________________________________________________________________\n",
            "dropout_1 (Dropout)          (None, 10, 32)            0         \n",
            "_________________________________________________________________\n",
            "conv1d_2 (Conv1D)            (None, 10, 32)            3104      \n",
            "_________________________________________________________________\n",
            "up_sampling1d (UpSampling1D) (None, 20, 32)            0         \n",
            "_________________________________________________________________\n",
            "dropout_2 (Dropout)          (None, 20, 32)            0         \n",
            "_________________________________________________________________\n",
            "conv1d_3 (Conv1D)            (None, 20, 16)            1552      \n",
            "_________________________________________________________________\n",
            "up_sampling1d_1 (UpSampling1 (None, 40, 16)            0         \n",
            "_________________________________________________________________\n",
            "dropout_3 (Dropout)          (None, 40, 16)            0         \n",
            "_________________________________________________________________\n",
            "conv1d_4 (Conv1D)            (None, 40, 4)             196       \n",
            "=================================================================\n",
            "Total params: 6,724\n",
            "Trainable params: 6,724\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "None\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "AoMwnBvKLk9d",
        "outputId": "b69cc8a8-e75b-4ed0-d16d-4ec15422906d"
      },
      "source": [
        "cnn_model = create_1dCNN()\n",
        "\n",
        "history_cnn= cnn_model.fit(x_train, y_train, epochs=25, verbose=1, validation_data=(x_val, y_val))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/25\n",
            "921/921 [==============================] - 8s 6ms/step - loss: 0.7880 - accuracy: 0.7404 - val_loss: 0.5315 - val_accuracy: 0.7864\n",
            "Epoch 2/25\n",
            "921/921 [==============================] - 5s 5ms/step - loss: 0.5503 - accuracy: 0.7929 - val_loss: 0.4702 - val_accuracy: 0.8143\n",
            "Epoch 3/25\n",
            "921/921 [==============================] - 5s 5ms/step - loss: 0.5020 - accuracy: 0.8084 - val_loss: 0.4342 - val_accuracy: 0.8347\n",
            "Epoch 4/25\n",
            "921/921 [==============================] - 5s 5ms/step - loss: 0.4740 - accuracy: 0.8205 - val_loss: 0.4147 - val_accuracy: 0.8390\n",
            "Epoch 5/25\n",
            "921/921 [==============================] - 5s 5ms/step - loss: 0.4550 - accuracy: 0.8258 - val_loss: 0.3925 - val_accuracy: 0.8536\n",
            "Epoch 6/25\n",
            "921/921 [==============================] - 5s 5ms/step - loss: 0.4493 - accuracy: 0.8291 - val_loss: 0.3887 - val_accuracy: 0.8474\n",
            "Epoch 7/25\n",
            "921/921 [==============================] - 5s 5ms/step - loss: 0.4370 - accuracy: 0.8315 - val_loss: 0.3695 - val_accuracy: 0.8583\n",
            "Epoch 8/25\n",
            "921/921 [==============================] - 5s 5ms/step - loss: 0.4183 - accuracy: 0.8411 - val_loss: 0.3666 - val_accuracy: 0.8603\n",
            "Epoch 9/25\n",
            "921/921 [==============================] - 5s 5ms/step - loss: 0.4209 - accuracy: 0.8388 - val_loss: 0.3565 - val_accuracy: 0.8646\n",
            "Epoch 10/25\n",
            "921/921 [==============================] - 5s 5ms/step - loss: 0.4049 - accuracy: 0.8453 - val_loss: 0.3533 - val_accuracy: 0.8638\n",
            "Epoch 11/25\n",
            "921/921 [==============================] - 5s 5ms/step - loss: 0.4068 - accuracy: 0.8435 - val_loss: 0.3493 - val_accuracy: 0.8675\n",
            "Epoch 12/25\n",
            "921/921 [==============================] - 5s 5ms/step - loss: 0.3993 - accuracy: 0.8470 - val_loss: 0.3411 - val_accuracy: 0.8745\n",
            "Epoch 13/25\n",
            "921/921 [==============================] - 5s 5ms/step - loss: 0.3996 - accuracy: 0.8472 - val_loss: 0.3355 - val_accuracy: 0.8749\n",
            "Epoch 14/25\n",
            "921/921 [==============================] - 5s 5ms/step - loss: 0.4001 - accuracy: 0.8466 - val_loss: 0.3390 - val_accuracy: 0.8728\n",
            "Epoch 15/25\n",
            "921/921 [==============================] - 5s 5ms/step - loss: 0.3927 - accuracy: 0.8503 - val_loss: 0.3328 - val_accuracy: 0.8769\n",
            "Epoch 16/25\n",
            "921/921 [==============================] - 5s 6ms/step - loss: 0.3891 - accuracy: 0.8527 - val_loss: 0.3322 - val_accuracy: 0.8754\n",
            "Epoch 17/25\n",
            "921/921 [==============================] - 5s 6ms/step - loss: 0.3877 - accuracy: 0.8505 - val_loss: 0.3342 - val_accuracy: 0.8741\n",
            "Epoch 18/25\n",
            "921/921 [==============================] - 5s 5ms/step - loss: 0.3826 - accuracy: 0.8542 - val_loss: 0.3308 - val_accuracy: 0.8752\n",
            "Epoch 19/25\n",
            "921/921 [==============================] - 5s 5ms/step - loss: 0.3888 - accuracy: 0.8521 - val_loss: 0.3236 - val_accuracy: 0.8791\n",
            "Epoch 20/25\n",
            "921/921 [==============================] - 5s 6ms/step - loss: 0.3746 - accuracy: 0.8588 - val_loss: 0.3340 - val_accuracy: 0.8718\n",
            "Epoch 21/25\n",
            "921/921 [==============================] - 5s 5ms/step - loss: 0.3764 - accuracy: 0.8566 - val_loss: 0.3210 - val_accuracy: 0.8795\n",
            "Epoch 22/25\n",
            "921/921 [==============================] - 5s 5ms/step - loss: 0.3807 - accuracy: 0.8534 - val_loss: 0.3211 - val_accuracy: 0.8792\n",
            "Epoch 23/25\n",
            "921/921 [==============================] - 5s 5ms/step - loss: 0.3712 - accuracy: 0.8583 - val_loss: 0.3207 - val_accuracy: 0.8797\n",
            "Epoch 24/25\n",
            "921/921 [==============================] - 5s 5ms/step - loss: 0.3801 - accuracy: 0.8542 - val_loss: 0.3213 - val_accuracy: 0.8779\n",
            "Epoch 25/25\n",
            "921/921 [==============================] - 5s 5ms/step - loss: 0.3759 - accuracy: 0.8582 - val_loss: 0.3272 - val_accuracy: 0.8765\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 573
        },
        "id": "868Fv3aAMW6-",
        "outputId": "e684ad02-1cbb-4e2d-ec0d-319260e35d2f"
      },
      "source": [
        "#Plot the loss and accuracy for the model\n",
        "plot_history(history_cnn)"
      ],
      "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"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pRYAXQ3TIozu",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "f55d955e-749a-4fcd-b0bd-eaad921bce36"
      },
      "source": [
        "from sklearn.metrics import classification_report\n",
        "\n",
        "y_val_pred = cnn_model.predict(x_val)\n",
        "y_val_pred_flat = flatten_pred(y_val_pred)\n",
        "y_val_flat = flatten_pred(y_val)\n",
        "print(classification_report(y_val_pred_flat, y_val_flat))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.97      0.88      0.92    106925\n",
            "           1       0.87      0.87      0.87      5651\n",
            "           2       0.93      0.91      0.92      7726\n",
            "           3       0.43      0.85      0.57     10618\n",
            "\n",
            "    accuracy                           0.88    130920\n",
            "   macro avg       0.80      0.88      0.82    130920\n",
            "weighted avg       0.92      0.88      0.89    130920\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZKTxxTj4U26h"
      },
      "source": [
        "# LSTM \n",
        "Single layer LSTM"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GxGZtyvOSyfu"
      },
      "source": [
        "def create_LSTM():\n",
        "  model = Sequential() \n",
        "  model.add(LSTM(100,  activation='relu', return_sequences=True,  input_shape=(40, 6)))\n",
        "  model.add(Dropout(0.1))\n",
        "  model.add(Dense(4, activation=\"softmax\"))\n",
        "  #model.summary()\n",
        "  model.compile(optimizer=Adam(learning_rate=0.001), loss='categorical_crossentropy', metrics = ['accuracy'])\n",
        "  return model"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Pr8pwxqtS1zw",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "a4942a0f-77fa-482b-a78e-8cbe6c1acde4"
      },
      "source": [
        "lstm_model =create_LSTM()\n",
        "\n",
        "print(lstm_model.summary())"
      ],
      "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",
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "lstm (LSTM)                  (None, 40, 100)           42800     \n",
            "_________________________________________________________________\n",
            "dropout (Dropout)            (None, 40, 100)           0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 40, 4)             404       \n",
            "=================================================================\n",
            "Total params: 43,204\n",
            "Trainable params: 43,204\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "None\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1BnI1rS4S8Mz",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "f7b0d0a3-6441-49d9-b8b0-defd4bdaceb9"
      },
      "source": [
        "history_lstm = lstm_model.fit(x_train, y_train, epochs=10, verbose=1, validation_data=(x_val, y_val))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/10\n",
            "921/921 [==============================] - 56s 57ms/step - loss: 1.0211 - accuracy: 0.7480 - val_loss: 0.7006 - val_accuracy: 0.7504\n",
            "Epoch 2/10\n",
            "921/921 [==============================] - 53s 57ms/step - loss: 0.6241 - accuracy: 0.7794 - val_loss: 0.5272 - val_accuracy: 0.7979\n",
            "Epoch 3/10\n",
            "921/921 [==============================] - 53s 58ms/step - loss: 0.4764 - accuracy: 0.8266 - val_loss: 0.4407 - val_accuracy: 0.8433\n",
            "Epoch 4/10\n",
            "921/921 [==============================] - 53s 57ms/step - loss: 0.4269 - accuracy: 0.8444 - val_loss: 0.4142 - val_accuracy: 0.8493\n",
            "Epoch 5/10\n",
            "921/921 [==============================] - 52s 56ms/step - loss: 0.3892 - accuracy: 0.8590 - val_loss: 0.3821 - val_accuracy: 0.8607\n",
            "Epoch 6/10\n",
            "921/921 [==============================] - 53s 57ms/step - loss: 0.3688 - accuracy: 0.8665 - val_loss: 0.3741 - val_accuracy: 0.8639\n",
            "Epoch 7/10\n",
            "921/921 [==============================] - 52s 56ms/step - loss: 0.3546 - accuracy: 0.8720 - val_loss: 0.3579 - val_accuracy: 0.8713\n",
            "Epoch 8/10\n",
            "921/921 [==============================] - 52s 56ms/step - loss: 0.3396 - accuracy: 0.8772 - val_loss: 0.3561 - val_accuracy: 0.8702\n",
            "Epoch 9/10\n",
            "921/921 [==============================] - 52s 57ms/step - loss: 0.3312 - accuracy: 0.8807 - val_loss: 0.3349 - val_accuracy: 0.8800\n",
            "Epoch 10/10\n",
            "921/921 [==============================] - 52s 57ms/step - loss: 0.3210 - accuracy: 0.8836 - val_loss: 0.3403 - val_accuracy: 0.8752\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "g6Ni55nCa6uX",
        "outputId": "6d96b67d-c8d4-4bac-acb7-a2f81cdf776b"
      },
      "source": [
        "history_lstm = lstm_model.fit(x_train, y_train, epochs=15, verbose=1, validation_data=(x_val, y_val))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/15\n",
            "921/921 [==============================] - 52s 57ms/step - loss: 0.3135 - accuracy: 0.8869 - val_loss: 0.3395 - val_accuracy: 0.8778\n",
            "Epoch 2/15\n",
            "921/921 [==============================] - 53s 57ms/step - loss: 0.3032 - accuracy: 0.8910 - val_loss: 0.3450 - val_accuracy: 0.8734\n",
            "Epoch 3/15\n",
            "921/921 [==============================] - 53s 57ms/step - loss: 0.2990 - accuracy: 0.8923 - val_loss: 0.3124 - val_accuracy: 0.8866\n",
            "Epoch 4/15\n",
            "921/921 [==============================] - 52s 57ms/step - loss: 0.2905 - accuracy: 0.8956 - val_loss: 0.3242 - val_accuracy: 0.8842\n",
            "Epoch 5/15\n",
            "921/921 [==============================] - 52s 56ms/step - loss: 0.2845 - accuracy: 0.8971 - val_loss: 0.3077 - val_accuracy: 0.8908\n",
            "Epoch 6/15\n",
            "921/921 [==============================] - 52s 56ms/step - loss: 0.2815 - accuracy: 0.8994 - val_loss: 0.3113 - val_accuracy: 0.8876\n",
            "Epoch 7/15\n",
            "921/921 [==============================] - 52s 56ms/step - loss: 0.2732 - accuracy: 0.9017 - val_loss: 0.3108 - val_accuracy: 0.8906\n",
            "Epoch 8/15\n",
            "921/921 [==============================] - 52s 56ms/step - loss: 0.2687 - accuracy: 0.9038 - val_loss: 0.3060 - val_accuracy: 0.8914\n",
            "Epoch 9/15\n",
            "921/921 [==============================] - 52s 57ms/step - loss: 0.2619 - accuracy: 0.9060 - val_loss: 0.3024 - val_accuracy: 0.8937\n",
            "Epoch 10/15\n",
            "921/921 [==============================] - 53s 57ms/step - loss: 0.2572 - accuracy: 0.9075 - val_loss: 0.3031 - val_accuracy: 0.8945\n",
            "Epoch 11/15\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.2543 - accuracy: 0.9088 - val_loss: 0.3029 - val_accuracy: 0.8926\n",
            "Epoch 12/15\n",
            "921/921 [==============================] - 54s 59ms/step - loss: 0.2509 - accuracy: 0.9098 - val_loss: 0.2975 - val_accuracy: 0.8972\n",
            "Epoch 13/15\n",
            "921/921 [==============================] - 53s 58ms/step - loss: 0.2478 - accuracy: 0.9111 - val_loss: 0.3192 - val_accuracy: 0.8933\n",
            "Epoch 14/15\n",
            "921/921 [==============================] - 53s 58ms/step - loss: 0.3290 - accuracy: 0.8847 - val_loss: 0.3043 - val_accuracy: 0.8914\n",
            "Epoch 15/15\n",
            "921/921 [==============================] - 54s 59ms/step - loss: 0.2572 - accuracy: 0.9078 - val_loss: 0.3017 - val_accuracy: 0.8961\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 573
        },
        "id": "LgDsEzkQVKn8",
        "outputId": "a43ceb9c-a07d-4d5a-cfc6-822316dbfd5e"
      },
      "source": [
        "#Plot the loss and accuracy for the model\n",
        "plot_history(history_lstm)"
      ],
      "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"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "H86uLftpVOpf",
        "outputId": "7dab449c-25f0-42a6-d6f2-fac6df24f8e4"
      },
      "source": [
        "from sklearn.metrics import classification_report\n",
        "\n",
        "y_val_pred = lstm_model.predict(x_val)\n",
        "y_val_pred_flat = flatten_pred(y_val_pred)\n",
        "y_val_flat = flatten_pred(y_val)\n",
        "print(classification_report(y_val_pred_flat, y_val_flat))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.97      0.90      0.93    104363\n",
            "           1       0.79      0.90      0.84      4951\n",
            "           2       0.84      0.91      0.88      6936\n",
            "           3       0.60      0.86      0.71     14670\n",
            "\n",
            "    accuracy                           0.90    130920\n",
            "   macro avg       0.80      0.89      0.84    130920\n",
            "weighted avg       0.92      0.90      0.90    130920\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GYDAM7efjM4Q"
      },
      "source": [
        "#BidirectionalLSTM + LSTM\n",
        "LearningRate-0.01\n",
        " Dropout for 1st Layer - 0.2 \n",
        " Dropout for 2nd layer - 0.1\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cr9BXYmOWLXx"
      },
      "source": [
        "def create_biLSTM():\n",
        "  model = Sequential()\n",
        "  model.add(Bidirectional(LSTM(100,return_sequences=True, input_shape=(40,6))))\n",
        "  model.add(Dropout(0.2))\n",
        "  model.add(LSTM(50, return_sequences=True, activation=\"relu\"))\n",
        "  model.add(Dropout(0.1))\n",
        "  model.add(Dense(25, activation=\"relu\"))\n",
        "  model.add(Dense(4, activation=\"softmax\"))\n",
        "  model.compile(optimizer=Adam(learning_rate=0.01),  loss='categorical_crossentropy', metrics = ['accuracy'])\n",
        "  return model"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "VFQTcfHWWR84",
        "outputId": "e4ead9a5-d9c4-4e3a-df0a-87d551c598ce"
      },
      "source": [
        "bilstm_model =create_biLSTM()\n",
        "\n",
        "history_bilstm = bilstm_model.fit(x_train, y_train, epochs=25, verbose=1, validation_data=(x_val, y_val))"
      ],
      "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",
            "Epoch 1/25\n",
            "921/921 [==============================] - 95s 65ms/step - loss: 0.6708 - accuracy: 0.7926 - val_loss: 0.3244 - val_accuracy: 0.8788\n",
            "Epoch 2/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.2999 - accuracy: 0.8946 - val_loss: 0.2889 - val_accuracy: 0.9011\n",
            "Epoch 3/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.2411 - accuracy: 0.9169 - val_loss: 0.2600 - val_accuracy: 0.9136\n",
            "Epoch 4/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.2167 - accuracy: 0.9271 - val_loss: 0.2078 - val_accuracy: 0.9288\n",
            "Epoch 5/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.2040 - accuracy: 0.9314 - val_loss: 0.3795 - val_accuracy: 0.8702\n",
            "Epoch 6/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.2776 - accuracy: 0.9061 - val_loss: 0.2421 - val_accuracy: 0.9197\n",
            "Epoch 7/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.2168 - accuracy: 0.9272 - val_loss: 0.2271 - val_accuracy: 0.9255\n",
            "Epoch 8/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.1842 - accuracy: 0.9383 - val_loss: 0.2096 - val_accuracy: 0.9306\n",
            "Epoch 9/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 5.2271 - accuracy: 0.8084 - val_loss: 0.8186 - val_accuracy: 0.7381\n",
            "Epoch 10/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.8000 - accuracy: 0.7520 - val_loss: 0.8182 - val_accuracy: 0.7381\n",
            "Epoch 11/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.7975 - accuracy: 0.7531 - val_loss: 0.8186 - val_accuracy: 0.7381\n",
            "Epoch 12/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.7911 - accuracy: 0.7512 - val_loss: 0.8183 - val_accuracy: 0.7381\n",
            "Epoch 13/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.7855 - accuracy: 0.7551 - val_loss: 0.8183 - val_accuracy: 0.7381\n",
            "Epoch 14/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.8029 - accuracy: 0.7479 - val_loss: 0.8185 - val_accuracy: 0.7381\n",
            "Epoch 15/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.7952 - accuracy: 0.7509 - val_loss: 0.8182 - val_accuracy: 0.7381\n",
            "Epoch 16/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.7924 - accuracy: 0.7498 - val_loss: 0.8188 - val_accuracy: 0.7381\n",
            "Epoch 17/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.7882 - accuracy: 0.7529 - val_loss: 0.8184 - val_accuracy: 0.7381\n",
            "Epoch 18/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.7894 - accuracy: 0.7527 - val_loss: 0.8202 - val_accuracy: 0.7381\n",
            "Epoch 19/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.8027 - accuracy: 0.7471 - val_loss: 0.8204 - val_accuracy: 0.7381\n",
            "Epoch 20/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.7881 - accuracy: 0.7535 - val_loss: 0.8181 - val_accuracy: 0.7381\n",
            "Epoch 21/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.7846 - accuracy: 0.7532 - val_loss: 0.8186 - val_accuracy: 0.7381\n",
            "Epoch 22/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.7912 - accuracy: 0.7539 - val_loss: 0.8195 - val_accuracy: 0.7381\n",
            "Epoch 23/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.7916 - accuracy: 0.7515 - val_loss: 0.8189 - val_accuracy: 0.7381\n",
            "Epoch 24/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.7962 - accuracy: 0.7491 - val_loss: 0.8183 - val_accuracy: 0.7381\n",
            "Epoch 25/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.7837 - accuracy: 0.7541 - val_loss: 0.8187 - val_accuracy: 0.7381\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ucBZCWYvQiS8",
        "outputId": "902cdd1d-8af2-450d-92fd-1c0cb3be7c59"
      },
      "source": [
        "print(bilstm_model.summary())"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_4\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "bidirectional_4 (Bidirection (None, 40, 200)           85600     \n",
            "_________________________________________________________________\n",
            "dropout_8 (Dropout)          (None, 40, 200)           0         \n",
            "_________________________________________________________________\n",
            "lstm_9 (LSTM)                (None, 40, 50)            50200     \n",
            "_________________________________________________________________\n",
            "dropout_9 (Dropout)          (None, 40, 50)            0         \n",
            "_________________________________________________________________\n",
            "dense_8 (Dense)              (None, 40, 25)            1275      \n",
            "_________________________________________________________________\n",
            "dense_9 (Dense)              (None, 40, 4)             104       \n",
            "=================================================================\n",
            "Total params: 137,179\n",
            "Trainable params: 137,179\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "None\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 573
        },
        "id": "tZC8z0BlWd0O",
        "outputId": "c55b0520-cd5b-4264-c849-1b4044e5fa7a"
      },
      "source": [
        "#Plot the loss and accuracy for the model\n",
        "plot_history(history_bilstm)"
      ],
      "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"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7ZYSNm5vWix9",
        "outputId": "b2c3ad04-6a83-4f28-f94f-7d4511e979aa"
      },
      "source": [
        "from sklearn.metrics import classification_report\n",
        "\n",
        "y_val_pred = lstm_model.predict(x_val)\n",
        "y_val_pred_flat = flatten_pred(y_val_pred)\n",
        "y_val_flat = flatten_pred(y_val)\n",
        "print(classification_report(y_val_pred_flat, y_val_flat, zero_division = 1))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       1.00      0.74      0.85    130920\n",
            "           1       0.00      1.00      0.00         0\n",
            "           2       0.00      1.00      0.00         0\n",
            "           3       0.00      1.00      0.00         0\n",
            "\n",
            "    accuracy                           0.74    130920\n",
            "   macro avg       0.25      0.93      0.21    130920\n",
            "weighted avg       1.00      0.74      0.85    130920\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "zRyRFUfmYaKs"
      },
      "source": [
        "#BidirectionalLSTM + LSTM\n",
        "\n",
        "LearningRate-0.05\n",
        "Dropout for 1st Layer - 0.2\n",
        "Dropout for 2nd layer - 0.1"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fvbrI7fyYbfh"
      },
      "source": [
        "def create_LSTM3():\n",
        "  model = Sequential()\n",
        "  model.add(Bidirectional(LSTM(100,return_sequences=True, input_shape=(40,6))))\n",
        "  model.add(Dropout(0.2))\n",
        "  model.add(LSTM(50, return_sequences=True, activation=\"relu\"))\n",
        "  model.add(Dropout(0.1))\n",
        "  model.add(Dense(25, activation=\"relu\"))\n",
        "  model.add(Dense(4, activation=\"softmax\"))\n",
        "  model.compile(optimizer=Adam(learning_rate=0.05),  loss='categorical_crossentropy', metrics = ['accuracy'])\n",
        "  return model"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "SWmCWngJYnai",
        "outputId": "af5c6283-5a2c-4f56-8d04-589e05321e15"
      },
      "source": [
        "lstm3_model =create_LSTM3()\n",
        "history_lstm3 = lstm3_model.fit(x_train, y_train, epochs=25, verbose=1, validation_data=(x_val, y_val))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Layer lstm_1 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",
            "Epoch 1/25\n",
            "921/921 [==============================] - 94s 64ms/step - loss: 2623594548.8258 - accuracy: 0.7327 - val_loss: 0.8220 - val_accuracy: 0.7381\n",
            "Epoch 2/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.7963 - accuracy: 0.7517 - val_loss: 0.8189 - val_accuracy: 0.7381\n",
            "Epoch 3/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.7925 - accuracy: 0.7535 - val_loss: 0.8196 - val_accuracy: 0.7381\n",
            "Epoch 4/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 1838.0487 - accuracy: 0.7487 - val_loss: 0.8239 - val_accuracy: 0.7381\n",
            "Epoch 5/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.7965 - accuracy: 0.7540 - val_loss: 0.8191 - val_accuracy: 0.7381\n",
            "Epoch 6/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.8018 - accuracy: 0.7510 - val_loss: 0.8191 - val_accuracy: 0.7381\n",
            "Epoch 7/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.7988 - accuracy: 0.7491 - val_loss: 0.8223 - val_accuracy: 0.7381\n",
            "Epoch 8/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.7971 - accuracy: 0.7508 - val_loss: 0.8191 - val_accuracy: 0.7381\n",
            "Epoch 9/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.7936 - accuracy: 0.7516 - val_loss: 0.8189 - val_accuracy: 0.7381\n",
            "Epoch 10/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.7810 - accuracy: 0.7590 - val_loss: 0.8193 - val_accuracy: 0.7381\n",
            "Epoch 11/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.7814 - accuracy: 0.7585 - val_loss: 0.8195 - val_accuracy: 0.7381\n",
            "Epoch 12/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.7925 - accuracy: 0.7517 - val_loss: 0.8196 - val_accuracy: 0.7381\n",
            "Epoch 13/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.7935 - accuracy: 0.7519 - val_loss: 0.8193 - val_accuracy: 0.7381\n",
            "Epoch 14/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.7978 - accuracy: 0.7489 - val_loss: 0.8229 - val_accuracy: 0.7381\n",
            "Epoch 15/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.7886 - accuracy: 0.7553 - val_loss: 0.8275 - val_accuracy: 0.7381\n",
            "Epoch 16/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.8034 - accuracy: 0.7492 - val_loss: 0.8207 - val_accuracy: 0.7381\n",
            "Epoch 17/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.7973 - accuracy: 0.7526 - val_loss: 0.8239 - val_accuracy: 0.7381\n",
            "Epoch 18/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.7887 - accuracy: 0.7552 - val_loss: 0.8206 - val_accuracy: 0.7381\n",
            "Epoch 19/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.7876 - accuracy: 0.7542 - val_loss: 0.8218 - val_accuracy: 0.7381\n",
            "Epoch 20/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.7942 - accuracy: 0.7517 - val_loss: 0.8187 - val_accuracy: 0.7381\n",
            "Epoch 21/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.7978 - accuracy: 0.7526 - val_loss: 0.8198 - val_accuracy: 0.7381\n",
            "Epoch 22/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.8098 - accuracy: 0.7503 - val_loss: 0.8181 - val_accuracy: 0.7381\n",
            "Epoch 23/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 130608.1011 - accuracy: 0.7527 - val_loss: 0.9145 - val_accuracy: 0.7380\n",
            "Epoch 24/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 5.5969 - accuracy: 0.7530 - val_loss: 0.8964 - val_accuracy: 0.7381\n",
            "Epoch 25/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.8664 - accuracy: 0.7508 - val_loss: 0.8213 - val_accuracy: 0.7381\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "MzrN4Z7wYsSu",
        "outputId": "dfddf299-1bee-4335-9474-e0177df37e89"
      },
      "source": [
        "lstm3_model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "bidirectional (Bidirectional (None, 40, 200)           85600     \n",
            "_________________________________________________________________\n",
            "dropout (Dropout)            (None, 40, 200)           0         \n",
            "_________________________________________________________________\n",
            "lstm_1 (LSTM)                (None, 40, 50)            50200     \n",
            "_________________________________________________________________\n",
            "dropout_1 (Dropout)          (None, 40, 50)            0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 40, 25)            1275      \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 40, 4)             104       \n",
            "=================================================================\n",
            "Total params: 137,179\n",
            "Trainable params: 137,179\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 573
        },
        "id": "goToQHOwYylF",
        "outputId": "fc7aa30d-ada8-4d99-c28b-6930bc23164e"
      },
      "source": [
        "#Plot the loss and accuracy for the model\n",
        "plot_history(history_lstm3)"
      ],
      "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"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "8ANZ8-pjY4c0",
        "outputId": "7875afe2-67ee-4c16-f00c-837783e49d3d"
      },
      "source": [
        "from sklearn.metrics import classification_report\n",
        "\n",
        "y_val_pred = lstm3_model.predict(x_val)\n",
        "y_val_pred_flat = flatten_pred(y_val_pred)\n",
        "y_val_flat = flatten_pred(y_val)\n",
        "print(classification_report(y_val_pred_flat, y_val_flat))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       1.00      0.74      0.85    130920\n",
            "           1       0.00      0.00      0.00         0\n",
            "           2       0.00      0.00      0.00         0\n",
            "           3       0.00      0.00      0.00         0\n",
            "\n",
            "    accuracy                           0.74    130920\n",
            "   macro avg       0.25      0.18      0.21    130920\n",
            "weighted avg       1.00      0.74      0.85    130920\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Recall and F-score are ill-defined and being set to 0.0 in labels with no true samples. Use `zero_division` parameter to control this behavior.\n",
            "  _warn_prf(average, modifier, msg_start, len(result))\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yElkQwiETANJ"
      },
      "source": [
        "#BidirectionalLSTM + LSTM \n",
        "Learning Rate - 0.001\n",
        "Dropout for 1st Layer - 0.1\n",
        "Dropout for 2nd layer - 0.1"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "UV538J0lQ269",
        "outputId": "af802aa4-9215-4666-a644-6de610b3518a"
      },
      "source": [
        "x_train.shape"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(29449, 40, 6)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 25
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qzr6EnRfTldg"
      },
      "source": [
        "def create_LSTM_dropout_tuning():\n",
        "  model = Sequential()\n",
        "  model.add(Bidirectional(LSTM(100,return_sequences=True, input_shape=(40,6))))\n",
        "  model.add(Dropout(0.1))\n",
        "  model.add(LSTM(50, return_sequences=True, activation=\"relu\"))\n",
        "  model.add(Dropout(0.1))\n",
        "  model.add(Dense(25, activation=\"relu\"))\n",
        "  model.add(Dense(4, activation=\"softmax\"))\n",
        "  model.compile(optimizer=Adam(learning_rate=0.001),  loss='categorical_crossentropy', metrics = ['accuracy'])\n",
        "  return model"
      ],
      "execution_count": 27,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rPLHahdeT0Gw",
        "outputId": "3d8b82f5-909e-4fa6-fce1-e5fa1db8f9a8"
      },
      "source": [
        "\n",
        "lstm_dropout_tuning =create_LSTM_dropout_tuning()\n",
        "history_lstm_dropout1 = lstm_dropout_tuning.fit(x_train, y_train, epochs=25, verbose=1, validation_data=(x_val, y_val))"
      ],
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Layer lstm_5 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",
            "Epoch 1/25\n",
            "921/921 [==============================] - 66s 67ms/step - loss: 0.6327 - accuracy: 0.7764 - val_loss: 0.3259 - val_accuracy: 0.8859\n",
            "Epoch 2/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.3125 - accuracy: 0.8875 - val_loss: 0.2830 - val_accuracy: 0.9055\n",
            "Epoch 3/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.2626 - accuracy: 0.9116 - val_loss: 0.2331 - val_accuracy: 0.9218\n",
            "Epoch 4/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.2251 - accuracy: 0.9233 - val_loss: 0.2100 - val_accuracy: 0.9310\n",
            "Epoch 5/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.2053 - accuracy: 0.9312 - val_loss: 0.1904 - val_accuracy: 0.9386\n",
            "Epoch 6/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.1863 - accuracy: 0.9364 - val_loss: 0.1813 - val_accuracy: 0.9386\n",
            "Epoch 7/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.1732 - accuracy: 0.9403 - val_loss: 0.2092 - val_accuracy: 0.9342\n",
            "Epoch 8/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.1669 - accuracy: 0.9436 - val_loss: 0.1715 - val_accuracy: 0.9420\n",
            "Epoch 9/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.1548 - accuracy: 0.9483 - val_loss: 0.1714 - val_accuracy: 0.9428\n",
            "Epoch 10/25\n",
            "921/921 [==============================] - 59s 65ms/step - loss: 0.1399 - accuracy: 0.9531 - val_loss: 0.1739 - val_accuracy: 0.9462\n",
            "Epoch 11/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.1372 - accuracy: 0.9536 - val_loss: 0.1649 - val_accuracy: 0.9449\n",
            "Epoch 12/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.1274 - accuracy: 0.9565 - val_loss: 0.1790 - val_accuracy: 0.9417\n",
            "Epoch 13/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.1204 - accuracy: 0.9597 - val_loss: 0.1587 - val_accuracy: 0.9478\n",
            "Epoch 14/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.1140 - accuracy: 0.9613 - val_loss: 0.1723 - val_accuracy: 0.9451\n",
            "Epoch 15/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.1034 - accuracy: 0.9651 - val_loss: 0.1495 - val_accuracy: 0.9522\n",
            "Epoch 16/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.0966 - accuracy: 0.9674 - val_loss: 0.1718 - val_accuracy: 0.9472\n",
            "Epoch 17/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.0955 - accuracy: 0.9689 - val_loss: 0.1689 - val_accuracy: 0.9514\n",
            "Epoch 18/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.0864 - accuracy: 0.9718 - val_loss: 0.1657 - val_accuracy: 0.9524\n",
            "Epoch 19/25\n",
            "921/921 [==============================] - 59s 65ms/step - loss: 0.0856 - accuracy: 0.9715 - val_loss: 0.1640 - val_accuracy: 0.9526\n",
            "Epoch 20/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.0846 - accuracy: 0.9717 - val_loss: 0.1888 - val_accuracy: 0.9447\n",
            "Epoch 21/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.0796 - accuracy: 0.9738 - val_loss: 0.1799 - val_accuracy: 0.9508\n",
            "Epoch 22/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.0709 - accuracy: 0.9768 - val_loss: 0.1787 - val_accuracy: 0.9483\n",
            "Epoch 23/25\n",
            "921/921 [==============================] - 59s 65ms/step - loss: 0.0708 - accuracy: 0.9761 - val_loss: 0.2223 - val_accuracy: 0.9396\n",
            "Epoch 24/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.0709 - accuracy: 0.9768 - val_loss: 0.1856 - val_accuracy: 0.9466\n",
            "Epoch 25/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.0697 - accuracy: 0.9774 - val_loss: 0.1659 - val_accuracy: 0.9547\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "YGmzCMceUUEz",
        "outputId": "b1a0ec25-7ae8-429c-f4f1-800b06454292"
      },
      "source": [
        "lstm_dropout_tuning.summary()"
      ],
      "execution_count": 53,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_2\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "bidirectional_2 (Bidirection (None, 40, 200)           85600     \n",
            "_________________________________________________________________\n",
            "dropout_4 (Dropout)          (None, 40, 200)           0         \n",
            "_________________________________________________________________\n",
            "lstm_5 (LSTM)                (None, 40, 50)            50200     \n",
            "_________________________________________________________________\n",
            "dropout_5 (Dropout)          (None, 40, 50)            0         \n",
            "_________________________________________________________________\n",
            "dense_4 (Dense)              (None, 40, 25)            1275      \n",
            "_________________________________________________________________\n",
            "dense_5 (Dense)              (None, 40, 4)             104       \n",
            "=================================================================\n",
            "Total params: 137,179\n",
            "Trainable params: 137,179\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "v3S6OhhAdRpk",
        "outputId": "a346d0ec-8d38-4ece-cb37-ff83780a5e08"
      },
      "source": [
        "from sklearn.metrics import classification_report\n",
        "\n",
        "y_val_pred = lstm_dropout_tuning.predict(x_val)\n",
        "y_val_pred_flat = flatten_pred(y_val_pred)\n",
        "y_val_flat = flatten_pred(y_val)\n",
        "print(classification_report(y_val_pred_flat, y_val_flat))"
      ],
      "execution_count": 35,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.97      0.97      0.97     97193\n",
            "           1       0.94      0.90      0.92      5877\n",
            "           2       0.96      0.94      0.95      7604\n",
            "           3       0.88      0.92      0.90     20246\n",
            "\n",
            "    accuracy                           0.95    130920\n",
            "   macro avg       0.94      0.93      0.93    130920\n",
            "weighted avg       0.96      0.95      0.95    130920\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vU67zmoIXESr"
      },
      "source": [
        "#BidirectionalLSTM +LSTM\n",
        "\n",
        "Learning Rate - 0.001\n",
        "Dropout for 1st Layer - 0.2\n",
        "Dropout for 2nd layer - 0.2\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "IBk9lGeKXRzg"
      },
      "source": [
        "def create_biLSTM1():\n",
        "  model = Sequential()\n",
        "  model.add(Bidirectional(LSTM(100,return_sequences=True, input_shape=(40,6))))\n",
        "  model.add(Dropout(0.2))\n",
        "  model.add(LSTM(50, return_sequences=True, activation=\"relu\"))\n",
        "  model.add(Dropout(0.2))\n",
        "  model.add(Dense(25, activation=\"relu\"))\n",
        "  model.add(Dense(4, activation=\"softmax\"))\n",
        "  model.compile(optimizer=Adam(learning_rate=0.001),  loss='categorical_crossentropy', metrics = ['accuracy'])\n",
        "  return model"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "bJfHYSP8XXXR",
        "outputId": "0943258b-5787-4891-e8ba-ed83e04eecbe"
      },
      "source": [
        "biLSTM1_model =create_biLSTM1()\n",
        "history_biLSTM1 = biLSTM1_model.fit(x_train, y_train, epochs=25, verbose=1, validation_data=(x_val, y_val))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Layer lstm_1 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",
            "Epoch 1/25\n",
            "921/921 [==============================] - 96s 66ms/step - loss: 0.5903 - accuracy: 0.7948 - val_loss: 0.3510 - val_accuracy: 0.8810\n",
            "Epoch 2/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.3254 - accuracy: 0.8819 - val_loss: 0.2757 - val_accuracy: 0.9008\n",
            "Epoch 3/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.2590 - accuracy: 0.9112 - val_loss: 0.2377 - val_accuracy: 0.9227\n",
            "Epoch 4/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.2296 - accuracy: 0.9215 - val_loss: 0.2276 - val_accuracy: 0.9182\n",
            "Epoch 5/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.2103 - accuracy: 0.9282 - val_loss: 0.2099 - val_accuracy: 0.9277\n",
            "Epoch 6/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.1914 - accuracy: 0.9336 - val_loss: 0.1850 - val_accuracy: 0.9361\n",
            "Epoch 7/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.1819 - accuracy: 0.9386 - val_loss: 0.1839 - val_accuracy: 0.9388\n",
            "Epoch 8/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.1658 - accuracy: 0.9457 - val_loss: 0.1834 - val_accuracy: 0.9394\n",
            "Epoch 9/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.1607 - accuracy: 0.9455 - val_loss: 0.1948 - val_accuracy: 0.9390\n",
            "Epoch 10/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.1530 - accuracy: 0.9487 - val_loss: 0.1705 - val_accuracy: 0.9417\n",
            "Epoch 11/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.1378 - accuracy: 0.9544 - val_loss: 0.1672 - val_accuracy: 0.9476\n",
            "Epoch 12/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.1326 - accuracy: 0.9543 - val_loss: 0.1728 - val_accuracy: 0.9395\n",
            "Epoch 13/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.1317 - accuracy: 0.9555 - val_loss: 0.1653 - val_accuracy: 0.9446\n",
            "Epoch 14/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.1139 - accuracy: 0.9631 - val_loss: 0.1578 - val_accuracy: 0.9492\n",
            "Epoch 15/25\n",
            "921/921 [==============================] - 58s 64ms/step - loss: 0.1038 - accuracy: 0.9657 - val_loss: 0.1794 - val_accuracy: 0.9438\n",
            "Epoch 16/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.1050 - accuracy: 0.9648 - val_loss: 0.1750 - val_accuracy: 0.9473\n",
            "Epoch 17/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.0964 - accuracy: 0.9682 - val_loss: 0.1825 - val_accuracy: 0.9404\n",
            "Epoch 18/25\n",
            "921/921 [==============================] - 61s 66ms/step - loss: 0.0914 - accuracy: 0.9702 - val_loss: 0.1743 - val_accuracy: 0.9461\n",
            "Epoch 19/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.0945 - accuracy: 0.9686 - val_loss: 0.1819 - val_accuracy: 0.9432\n",
            "Epoch 20/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.0813 - accuracy: 0.9727 - val_loss: 0.1815 - val_accuracy: 0.9476\n",
            "Epoch 21/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.0800 - accuracy: 0.9740 - val_loss: 0.1660 - val_accuracy: 0.9457\n",
            "Epoch 22/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.0755 - accuracy: 0.9751 - val_loss: 0.1682 - val_accuracy: 0.9492\n",
            "Epoch 23/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.0749 - accuracy: 0.9761 - val_loss: 0.1807 - val_accuracy: 0.9422\n",
            "Epoch 24/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.0673 - accuracy: 0.9790 - val_loss: 0.1850 - val_accuracy: 0.9496\n",
            "Epoch 25/25\n",
            "921/921 [==============================] - 58s 63ms/step - loss: 0.0620 - accuracy: 0.9798 - val_loss: 0.1857 - val_accuracy: 0.9501\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ey0lbEiWUPAO",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "deb20f3b-6b11-4957-92d8-2c69b1527154"
      },
      "source": [
        "biLSTM1_model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "bidirectional (Bidirectional (None, 40, 200)           85600     \n",
            "_________________________________________________________________\n",
            "dropout (Dropout)            (None, 40, 200)           0         \n",
            "_________________________________________________________________\n",
            "lstm_1 (LSTM)                (None, 40, 50)            50200     \n",
            "_________________________________________________________________\n",
            "dropout_1 (Dropout)          (None, 40, 50)            0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 40, 25)            1275      \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 40, 4)             104       \n",
            "=================================================================\n",
            "Total params: 137,179\n",
            "Trainable params: 137,179\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 573
        },
        "id": "xLMPUq1KXo3R",
        "outputId": "6516894c-2106-439d-94bb-40934b66b5d8"
      },
      "source": [
        "#Plot the loss and accuracy for the model\n",
        "plot_history(history_biLSTM1)"
      ],
      "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": "iVBORw0KGgoAAAANSUhEUgAAAYwAAAEWCAYAAAB1xKBvAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3dd3hUZfbA8e8xlCC9KhIgKN2VGmWxIggiKgiCgquAumBDsbBrWfSHCuq6uIss6ooNYRFURESlKYKyVkIRpQQQKaGGTqgp5/fHe0MmIWUScjNJ5nyeZ57M3DL33Elyz7zlvq+oKsYYY0xuzgh1AMYYY4oHSxjGGGOCYgnDGGNMUCxhGGOMCYolDGOMMUGxhGGMMSYoljCMMcYExRKGMYCILBSRfSJSNtSxGFNUWcIwYU9EooHLAAW6F+JxSxXWsYwpCJYwjIH+wA/ABGBA2kIRqSsi00UkQUT2iMi4gHWDRGS1iBwSkVUi0sZbriLSMGC7CSIy0nveQUTiReRREdkBvCMiVUXkM+8Y+7znUQH7VxORd0Rkm7d+hrf8VxG5PmC70iKyW0Ra+/YpmbBnCcMYlzAme4+rReQsEYkAPgM2AdFAHWAqgIj0AUZ4+1XClUr2BHmss4FqQH1gMO5/8B3vdT3gKDAuYPtJwJnA+UAt4F/e8onArQHbdQO2q+qyIOMwJs/ExpIy4UxELgUWALVVdbeIrAFex5U4ZnrLkzPtMxeYpaovZ/F+CjRS1fXe6wlAvKoOF5EOwDygkqoeyyaeVsACVa0qIrWBrUB1Vd2XabtzgDigjqoeFJFpwE+q+mK+PwxjcmElDBPuBgDzVHW39/o9b1ldYFPmZOGpC/yWz+MlBCYLETlTRF4XkU0ichD4BqjilXDqAnszJwsAVd0GfAvcKCJVgGtwJSRjfGONbiZsiUg54CYgwmtTACgLVAF2AvVEpFQWSWMLcF42b3sEV4WU5mwgPuB15iL9I0AToJ2q7vBKGMsA8Y5TTUSqqOr+LI71LvBn3P/x96q6NfuzNeb0WQnDhLMbgBSgOdDKezQDFnnrtgMviEh5EYkUkUu8/d4EholIW3Eaikh9b91y4BYRiRCRrsAVucRQEddusV9EqgH/l7ZCVbcDs4FXvcbx0iJyecC+M4A2wFBcm4YxvrKEYcLZAOAdVd2sqjvSHrhG537A9UBDYDOulHAzgKp+CIzCVV8dwl24q3nvOdTbbz/wJ29dTsYA5YDduHaTOZnW3wYkAWuAXcCDaStU9SjwEdAAmJ7Hczcmz6zR25hiTESeAhqr6q25bmzMabI2DGOKKa8K605cKcQY31mVlDHFkIgMwjWKz1bVb0IdjwkPViVljDEmKFbCMMYYE5QS04ZRo0YNjY6ODnUYxhhTrCxZsmS3qtYMZtsSkzCio6OJjY0NdRjGGFOsiMimYLe1KiljjDFBsYRhjDEmKJYwjDHGBMXXNgxvLJ2XgQjgTVV9IdP6+sDbQE1gL3CrqsZ7614ErsUltS+AoZrHPsBJSUnEx8dz7FiWI0mbfIiMjCQqKorSpUuHOhRjTCHzLWF4wzO/AnTGjcOzWERmquqqgM1GAxNV9V0R6Qg8D9wmIhcDlwAtvO3+hxvEbWFeYoiPj6dixYpER0cjIqd3QgZVZc+ePcTHx9OgQYNQh2OMKWR+VkldBKxX1Q2qegI3W1mPTNs0B77yni8IWK9AJFAGN9x0adxw03ly7NgxqlevbsmigIgI1atXtxKbMWHKz4RRBzd0QZp4b1mgn4Fe3vOeQEURqa6q3+MSyHbvMVdVV2c+gIgMFpFYEYlNSEjIMghLFgXLPk9jwleo78MYBowTkYG4mca2Aiki0hA3L0GUt90XInKZqi4K3FlVxwPjAWJiYmyME2NMiXLiBBw6lP3j4EH3s1YtGDzY/3j8TBhbcVNMponylp3kTTPZC0BEKgA3qup+b2C1H1Q10Vs3G2iPm9im2NizZw+dOnUCYMeOHURERFCzpruh8qeffqJMmTLZ7hsbG8vEiRMZO3ZsocRqjAmNgwdh8WL48Uf44QdYuRIOHHCJ4MSJ4N7jj38s/gljMdBIRBrgEkVf4JbADUSkBm7O4lTgcVyPKXAT1gwSkedxU1VegZtoplipXr06y5cvB2DEiBFUqFCBYcOGnVyfnJxMqVJZ/wpiYmKIiYkplDiNMYUjJcUlhB9+cAnixx9h1SpI6//ZpAlceCFUqwYVK6Y/KlXK+Drz8sLqtOhbwlDVZBEZAszFdat9W1VXisgzQKyqzgQ6AM+LiOKqpO7zdp8GdAR+wTWAz1HVT/2KtTANHDiQyMhIli1bxiWXXELfvn0ZOnQox44do1y5crzzzjs0adKEhQsXMnr0aD777DNGjBjB5s2b2bBhA5s3b+bBBx/kgQceCPWpGGNyoArbtsFPP6WXHmJj4fBht756dWjXDm66yf286CKoWjW0MefG1zYMVZ0FzMq07KmA59NwySHzfinAXQUZy4MPgvdlv8C0agVj8lHuiY+P57vvviMiIoKDBw+yaNEiSpUqxZdffskTTzzBRx99dMo+a9asYcGCBRw6dIgmTZpwzz332L0QxhQRhw+7ksOKFfDLL+k/9+xx60uXhtat4Y47XHJo1w7OOw+KWx+SUDd6h6U+ffoQEREBwIEDBxgwYADr1q1DREhKSspyn2uvvZayZctStmxZatWqxc6dO4mKispyW2OMP1JSYMOGjElhxQr47bf0aqXy5eGCC6BXL2jRAmJi3JfLyMjQxl4QwiZh5Kck4Jfy5cuffP7kk09y5ZVX8vHHH7Nx40Y6dOiQ5T5ly5Y9+TwiIoLk5GS/wzSmRDt2DPbtS3/s3ZvxdVaPjRvhyBG3/xlnQMOGLhn07++SRIsWEB3t1pVEYZMwiqoDBw5Qp467PWXChAmhDcaYEkYVtm+H1athzRr3M+2xfXvO+1aq5NoU0h5Nm0Lnzi4ptGgBzZvDmWcWznkUFZYwQuyvf/0rAwYMYOTIkVx77bWhDseYYik5GX7//dTEsGaN66KaplIlaNYMrr7alQ6qV8+YFNIelStDNh0Yw1qJmdM7JiZGM0+gtHr1apo1axaiiEou+1xNqB0+7HodffMNLFrknh89mr6+dm1XImjWLOOjdu3i19DsNxFZoqpB9eG3HGqMKfL27YNvv3XJ4ZtvXPfU5GTXVtCqlbtprWVLlxSaNoUqVUIdcclkCcMYU+Ts2OGSQ1qCWLHCtUeULu3uV/jLX+Cyy+Dii131kSkcljCMMb5LTXWlhF27ICEh58euXS5hgGtUvvhiGDECLr/c3b9QrlxITyWsWcIwxhSonTvT72z+8Ud3Q9vu3e4ehqxUrgw1a7pHdLQbGqNJE5cg2rQpvGEvTO4sYRhj8u3YMVi2LGOC2LjRrStVyrUrXHcdnH12elKoWdONrlqzJtSoATmMwWmKGEsYxpigpN3lHDg20vLlkDY4Qd26btTUIUPczzZtrPqopCmh9yMWHVdeeSVz587NsGzMmDHcc889WW7foUMH0roHd+vWjf3795+yzYgRIxg9enSOx50xYwarVqXPhvvUU0/x5Zdf5jV8E4ZOnHAjqE6bBs8+C7fc4noiVagAjRvDrbfCW2+59oWHH4bp02HrVti8GT74AB55BC65xJJFSWQlDJ/169ePqVOncvXVV59cNnXqVF588cVc9501a1au22RnxowZXHfddTRv3hyAZ555Jt/vZUoeVTffwvr17ga3VavSb3Zbv951WU1Tv767q7lTJ9dt9cIL4fzz7ca2cGS/cp/17t2b4cOHc+LECcqUKcPGjRvZtm0bU6ZM4eGHH+bo0aP07t2bp59++pR9o6OjiY2NpUaNGowaNYp3332XWrVqUbduXdq2bQvAG2+8wfjx4zlx4gQNGzZk0qRJLF++nJkzZ/L1118zcuRIPvroI5599lmuu+46evfuzfz58xk2bBjJyclceOGFvPbaa5QtW5bo6GgGDBjAp59+SlJSEh9++CFNmzYt7I/MnIakJNfovGPHqY/t2zO+ThsTCSAiwt353Ly5GzSvWTP3vEkTN5ieMRBOCSNE45tXq1aNiy66iNmzZ9OjRw+mTp3KTTfdxBNPPEG1atVISUmhU6dOrFixghYtWmT5HkuWLGHq1KksX76c5ORk2rRpczJh9OrVi0GDBgEwfPhw3nrrLe6//366d+9+MkEEOnbsGAMHDmT+/Pk0btyY/v3789prr/Hggw8CUKNGDZYuXcqrr77K6NGjefPNN0/3UzI+2r8fPv3UVR99953rjZSVqlXdXc5nn+3aF84+2z0aNHCJoWFDa3w2uQufhBFCadVSaQnjrbfe4oMPPmD8+PEkJyezfft2Vq1alW3CWLRoET179uRMb6Sz7t27n1z366+/Mnz4cPbv309iYmKGqq+sxMXF0aBBAxo3bgzAgAEDeOWVV04mjF69egHQtm1bpk+fftrnbgre3r3wyScuSXzxhStVREVBjx6u4TktGaQliLPOgoDBjo3Jt/BJGCEc37xHjx489NBDLF26lCNHjlCtWjVGjx7N4sWLqVq1KgMHDuTYsWP5eu+BAwcyY8YMWrZsyYQJE1i4cOFpxZo2jLoNoV607N4NM2a4JDF/vmtjqF8fHngAevd2dz+X1CG1TdHh65+YiHQVkTgRWS8ij2Wxvr6IzBeRFSKyUESiAtbVE5F5IrJaRFaJSLSfsfqpQoUKXHnlldxxxx3069ePgwcPUr58eSpXrszOnTuZPXt2jvtffvnlzJgxg6NHj3Lo0CE+/TR9ttpDhw5Ru3ZtkpKSmDx58snlFStW5NChQ6e8V5MmTdi4cSPr168HYNKkSVxxxRUFdKamIO3aBa+/Dldd5UoKgwbBunWuZ9LixW501tGjXRWTJQtTGHwrYYhIBPAK0BmIBxaLyExVXRWw2Whgoqq+KyIdgeeB27x1E4FRqvqFiFQAUv2KtTD069ePnj17MnXqVJo2bUrr1q1p2rQpdevW5ZJLLslx3zZt2nDzzTfTsmVLatWqxYUXXnhy3bPPPku7du2oWbMm7dq1O5kk+vbty6BBgxg7dizTpqXPghsZGck777xDnz59TjZ633333f6ctMmzAwfgww9h8mQ3hlJqKjRqBI8+6koSrVrZaKsmdHwb3lxE2gMjVPVq7/XjAKr6fMA2K4GuqrpFRAQ4oKqVRKQ5MF5VLw32eDa8eeGxz7VgJSXBvHkwcSLMnOnunm7cGG6+Gfr0gT/8wZKE8U9RGd68DrAl4HU80C7TNj8DvYCXgZ5ARRGpDjQG9ovIdKAB8CXwmKpmGI1GRAYDgwHq1avnxzkY4wtVWLrUJYkpU9yge9Wrw5//DLfd5u51sCRhippQN3oPA8aJyEDgG2ArkIKL6zKgNbAZeB8YCLwVuLOqjgfGgythFFbQxuTXli2uumniRHeTXJky0L27SxJdu1rXVlO0+ZkwtgJ1A15HectOUtVtuBIGXjvFjaq6X0TigeWqusFbNwP4I5kSRjBUFbGvagWmpMzQWJgOHICPP4ZJk2DBAle6uPRS16Ddp4+7R8KY4sDPhLEYaCQiDXCJoi9wS+AGIlID2KuqqcDjwNsB+1YRkZqqmgB0BDI2UAQhMjKSPXv2UL16dUsaBUBV2bNnD5GRkaEOpUhKTEwfZmPlSvdYtSp99NaGDd28DrfeCueeG8pIjckf3xKGqiaLyBBgLhABvK2qK0XkGSBWVWcCHYDnRURxVVL3efumiMgwYL7XGL4EeCOvMURFRREfH09CQkLBnJQhMjKSqKio3DcswQ4fdokhMCmsXJmeGMDdKNe0KbRv79olOnZ03V/te4spznzrJVXYsuolZUxBUHVThH7yievFtGRJ+royZVxiOP9892je3P0891wbnM8UD0Wll5QxxVZSkrsPYuZMlyg2bXKlg/bt4emn4YILLDGY8GN/6sZ4Dh6EOXNcgpg1yw3sFxkJnTvDk0+6mePOOivUURoTOpYwTFiLj08vRSxY4EoWNWpAz55uML+rrrLhvY1JYwnDhJ3kZPjsM3jtNXeHNbjhN4YOdUmifXs3P4QxJiNLGCZsbNsGb7zhHlu3Qp06rpvrTTe5hmvrwWRMzixhmBJNFb76ypUmZsyAlBTo0gXGjXNtEtZgbUzw7N/FlEh798KECe5u6rVr3ThNDz0Ed93lbqAzptg4ccINF3DggOuJkdXzWrXg3nt9D8UShikxVOGnn1xp4v333aiv7du7cZv69HE9noqF1FQ38UXDhtaYUlSkjRZZrZqb19YvH3/sJjnZsyc9KQQzudpll1nCMCYn+/e7adqXLXOPxYthzRrXq2nAALjnHmjZMtRR5kFSEkydCi+84G4fb9XKXTw6dQp1ZOFryxY3CNi777qiatmy8M9/uj+ugmz0SkqCxx+Hl16CZs3cH27lyumPKlWyf16xYuF9sVDVEvFo27atmpIpNVU1Pl71009Vn31WtVcv1QYNVN3XPvc4+2zVa65RfeUV1QMH8vDmiYm+xR20I0dc4NHR7mQuuEB11CjV+vXd62uvVV25MtRRFl2pqapr16ru318w75eYqDpxomqnTqoi7ndw+eWq48e7PzJwf4R79xbM8bZuVb30Uve+992neuxYwbxvkHBDNQV1nQ35hb6gHpYwSo6dO1U/+kj10UdVu3RRrVkzY3Jo2FC1Tx/V555TnT1bdfv2fBxk+3bVP/3JXRCeftpddArbgQOqf/+76llnuRNr395lxbRYjh5VffFF1cqVVc84Q/Wuu1R37Dj94x45ojpliuqNN6o++aTq8eOn/56hkJKiOmOG+9zS/jgaN1a95RbVf/5T9ZtvVA8dCv69FixQHThQtUIF917nnqs6YoTqb79l3G70aNVSpVxC/+670zuHr75SrVVL9cwzVSdPPr33yidLGKbYSE1VXbdO9Z13VO+4w/2/p/3vly6t2qqV6u23q44dq7poUR5LD1lJTlb9979VK1VSLVPGfXME1X793IW0MOzapTp8uGqVKu7YXbqoLlyYfdJKSFC9/353kapQQXXkSNXDh/N2zJQUd3G6/XbVihXdcWvVcj/btnXf0P1w5Ijq6tXu+AXl+HHVt99WbdbMxR8d7S7io0ap9uypWrdu+h+RiGrz5qr9+7s/ou++y/jZrVvnkmZa6a5iRdU773TJJqcvET/+6Iq5ERGqzz+f9/NLSXH7nXGGatOmIS1BWsIwRVZSkmpsrOqYMaq9e7uqpLT/7apVVa+/3n3p/u47H0rmP/yg2rq1O1jnzu4imZrq/nFBtV27fBZXgrR5s+rQoarlyrkL2Y03ug8jWHFxqjfc4GKNilJ9993cL1QrV6o+9lj6RbRiRZc0vvrK7Tt9umq1aqrly7usXVAlrdRU99716rnjnnOO6j33qM6Zk/8SzcGDqi+9pFqnjnvPli3dt/KkpFO33bFD9fPPXenxuusy/qFFRKi2aOF+32lJpXNn1f/+N2+JeP9+1ZtuSv97Crb0t3ev+0MH1ZtvducVQpYwTJGydatre+jcOb20n/bF8NZbVf/zH9Vffy3YL6EZ7NmjOniwuzCcc47qBx+cemGcPt1VC9Stq7p8ecEePy7OFZ9Kl3alhAEDVFetyv/7LVyoGhPjPsTWrd3FP9COHar/+pdqmzbpF8hu3Vw1VFYXxC1bVK+4wm3bt+/ptwXExalefbWebI8ZO9bV+Zcv75ZVquQulO+9F9yxdu5U/dvf0ktkHTq4usi8JLe0hrBPPnElimuucZ/h88+788+v1FTXthEZ6aoW583LefslS1zJpFQp97mEoio0E0sYpkjYuVP1oYfc/5KI+1J3333uunU6/6NBS0lxVRc1ariL5kMP5VyntXSp+/Zavry7sJyuLVtU//xnd+zISNUhQ1Q3bjz991V15zZ5cvo3+OuvV33rLXchjIjQk1VNY8YE9803OdlV6UREuEyen7r5xETVJ55wVX2VKqm+/HLGb/9Hj6p+9pn7TNKqw0qXdt8kxo079Y/it99U7703/Q+oVy9XFVQU/fKLq/oSUX38cdUTJzKuT01VfeMN1bJlXenwdNs+CpAlDBNSe/a4/5ny5V0V7cCBGdsNC8XPP6tecon7E7/44uBLDVu3um+eIqr/+Ef+vgHu2aP6l7+4C13p0q4aqiAaq7Ny5IjqCy+4CzS4EtLjj+e/BPP99+l188884xJJblJTXS+FtGqv/v1zr9pLSXEXzb/+VbVJEz1Z7Gzb1lUj9e3r/njKlHEJZs2a/J1PYTp8WHXQoPS/ubQvB4cPu3+CtKqrXbtCG2cmljBMSOzf7zqVVKrkrrf9+oXg//zgQdWHH3YXvOrV3bfuvNZ1HT7sumGBq0oKts49MdF13apc2X0A/fur/v57nk8hXxISVBcvLph6vf37XU8jUL3sMtVNm7LfNi7ONdqDK0IuWpS/Y65e7RJf+/bus6tY0SXdrVvz936hNHWqi79KFdXXXnOfC6g+9VRwCbiQFZmEAXQF4oD1wGNZrK8PzAdWAAuBqEzrKwHxwLjcjmUJI3QSE11VcNWqerKL+i+/FHIQ+/apTpjg2ijAfdPbvTv/75eS4uq6wdXv5/ReJ064C0Naw+r114fgAyhgqamuUb1CBXfhmzYt4/rERFeSKV3afUMYOzbrxuf8SEgIvjtsUbV+fXo7U7VqqrNmhTqibBWJhIGbx/s34FygDPAz0DzTNh8CA7znHYFJmda/DLxnCaNoOnLEdXdPq47u1i1vnX5OS0qKO9jIka7qKa3evlUrV61SUCZPdvXO553nvgVnjmHKFLcO3M1X//tfwR27KFi3Lv3CN2iQSxTTpqVXPw0Y4F91W3F3/LhrtyiodiufFJWE0R6YG/D6ceDxTNusBOp6zwU4GLCuLTAVGGgJo2g5ftzdmJz2Zb5Tp0Jqw0tIcBfw225Lz1Jp9d5/+5u7WPtR5P/uO3e8ypVV5851377nzEnvonvBBa4xtwj0ePHF8eOurQHcZ5DWpbWkJccwlZeE4edYUnWALQGv44F2mbb5GeiFK0n0BCqKSHVgH/AScCtwlY8xmjxISYHJk+Gpp9wc15de6l536JBpw6Sk9FE00x5HjrjR/848M/1Rvnz689KlTx2bJyXFDRA1Zw7Mnu2eq7qhZ6++Grp2dWOV+z1vavv2blTD66+Hbt2gTRsXS3S0G2eoX7+SPUhgmTLw97+nz1V7yy1uLCUbGz7shPo3PgwYJyIDgW+ArUAKcC8wS1XjJYcBvkRkMDAYoF69er4HG65U4fPP4cnHkqi48nv+dvaXXNNlK3UqHkCeOwiPZkoOR4/m/SAREacmkW3b3DjlItCunZvtqGtXaNu28C/Q9evDt9/CbbfB99/D2LEweLAbjC5cXHWVe5iw5WfC2ArUDXgd5S07SVW34UoYiEgF4EZV3S8i7YHLROReoAJQRkQSVfWxTPuPB8YDxMTEqG9nEsZip21kwWNzafjbHBbJfCpwCE2IQH49K33EzKpV3bfttNeVKmUcabNyZZcIjh+Hw4ddaSPwkd2yCy90JYjOnV2pItQqVnSzMKWmwhlnhDoaYwqdnwljMdBIRBrgEkVf4JbADUSkBrBXVVNxbRxvA6jqnwK2GQjEZE4WxidHjsDXX7P7vbkc+3gOMYfjiAEOVatPuRtvgWuuRjp2dEkgXFmyMGHKt4ShqskiMgSYi+sx9baqrhSRZ3CNLDOBDsDzIqK4Kqn7/IrHZEPVzb0wdy7MmYN+/Q1y4jjliWRZRAdWX3sPlzzblYqtGtuk18aEOXGN5MVfTEyMxsbGhjqM4uXAATcV3RdfALCjWnPe3381X0R05YJ7L2PYk+WKRE2QMcY/IrJEVWOC2TbUjd4mVOLjoVs3dPVq5nd5kQe+vZm4/fUYOBBeGwF16+b2BsaYcGMJIxz9+it6zTUk7T5A/0qzeH9eZ264AaaNgubNQx2cMaaosoQRbr7+muTrerDv+Jl0TvqGMhe04ttP4eKLQx2YMaaos+4eYWTva++T1LELaxPPoVuV73loQit++MGShTEmOFbCCAPHjsH/bvwnV816hEVyGQuGzuCrZ6tRsWKoIzPGFCeWMEowVZgxPZV9dzzCHQfH8N05vanzxSSeah4Z6tCMMcWQVUmVUL/+Ct06HiO5983ccXAMm3sO5eIt73OuJQtjTD5Zwihh9uyBIUOgQ4u9PPm/LvRhGikvvkS96WPsDmVjzGmxKqkSQtWNHPvAA1DlwCZ+qXINZx/+DaZMIaJv31CHZ4wpAewrZwmwb58bYfu226B7veXEVW9P7dRtyNy5YMnCGFNALGEUcwsXQsuW8NFHMOnPX/POhsspXTYC/ve/LCaqMMaY/LOEUUwdPw5//St07AjlysEvY+Zz6+RrkKgoN1/DH/4Q6hCNMSWMtWEUQ6tWwZ/+BMuXw113wZhr5hLZ9wZo1Ai+/BJq1Qp1iMaYEsgSRjGiCq+8An/5C1SoAJ98At1LzYKePd0gUF98ATVqhDpMY0wJZVVSxcSOHXDttXD//XDllfDLL9CdmXDDDXDBBTB/viULY4yvLGEUAzNnupywYAGMG+fm1z77u+lw443QurWrhqpWLdRhGmNKOEsYheXwYTh4MM+73HUX9OgBUVGwZAncdx/Ihx/ATTe5Oa/nzYMqVXwK2hhj0lnCKCzdu8NZZ7kr/m+/5br59u0QEwNvvOF6Q/34ozdXxXvvuZsuLr7YTasaznNrG2MKla8JQ0S6ikiciKwXkceyWF9fROaLyAoRWSgiUd7yViLyvYis9Nbd7GecvouNha++ghYtXAZo3Bhuvtktz8K+fdClC2zZ4tqx//53KFMGmDjR3Z13+eUwezY23KwxpjD5ljBEJAJ4BbgGaA70E5HM87mNBiaqagvgGeB5b/kRoL+qng90BcaISPGtd3n5ZXdx/+IL2LjRdXOaM8dVKXXq5EoK3tzqhw/DddfB2rUwY4ZbDcDbb8PAge7Gi88/h/LlQ3U2xpgw5WcJ4yJgvapuUNUTwFSgR6ZtmgNfec8XpK1X1bWqus57vg3YBdT0MVb/bNsG778Pd9wBlSrBOefACy+44sM//gFr1kDXrtCqFckT/svNvZL44QdX83TVVUN1cBkAABePSURBVN57vP463HknXH21awE/88yQnpIxJjz5mTDqAFsCXsd7ywL9DPTynvcEKopI9cANROQioAxwSsW/iAwWkVgRiU1ISCiwwAvUq69CcrLrDxuoUiUYNgx+/x3eeQdNTqbU7bfx6rzz+F/vMdx4daLb7pVX4O67XZ/ajz92t3UbY0wIhLrRexhwhYgsA64AtgIpaStFpDYwCbhdVVMz76yq41U1RlVjatYsggWQo0fhP/9xDd7nnZf1NmXKoAMG8kCHX7iOT6FBA9p/8BDUq+faOYYMcd2kpk+HSJvLwhgTOrkmDBG5XkTyk1i2AnUDXkd5y05S1W2q2ktVWwN/85bt945bCfgc+Juq/pCP44fe5MlugoqHHspxsxEjYNyrZ9Bs2HXU/e1rNxbUlVfChx+6ey0+/NBr9TbGmNAR9Rpbs91A5L9Ae+Aj4G1VXRPUG4uUAtYCnXCJYjFwi6quDNimBrBXVVNFZBSQoqpPiUgZYDbwqaqOCeZ4MTExGptNr6OQUHV325UuDUuXgkiWm40dC0OHuiaON9/MtFlCAlSvbhMfGWN8IyJLVDUmmG1zvRKp6q1Aa1wbwgSvu+tgEcmxT6eqJgNDgLnAauADVV0pIs+ISHdvsw5AnIisBc4CRnnLbwIuBwaKyHLv0SqYEyoyvvwSVq6EBx/MNllMnuySxQ03uHbtUzarWdOShTGmyMi1hHFyQ9cYfRvwIC4BNATGquq//QsveEWuhHHtte7W7E2boGzZU1Z//rlrmrj8cpg1y5onjDGhUaAlDBHpLiIfAwuB0sBFqnoN0BJ45HQCLbHi4lwWuPfeLJPFokXQuze0auXutbBkYYwpDoIZ3vxG4F+q+k3gQlU9IiJ3+hNWMTd2rGukvvvuU1b9/DNcfz3Ur+9u1q5UKQTxGWNMPgSTMEYA29NeiEg54CxV3aiq8/0KrNjauxcmTHAzHGWayGj9enfvXcWKbszAotgT2BhjshNMi+qHQOA9ECneMpOVN9+EI0dca3aAbdugc2d3D9+8ee42C2OMKU6CKWGU8ob2AEBVT3jdXk1mSUnw73+78Z5atsyw6tZbYfduNwZhs2Yhis8YY05DMCWMhIBusIhID2C3fyEVYx9/DPHxrittgG+/dZMfPfusG2/QGGOKo2Bu3DsPmAycAwhufKj+qrre//CCVyS61bZv7262W7s2w/0T114LP/3kBqq1QWaNMUVJXrrV5lolpaq/AX8UkQre68TTjK9k+uEH9xg7NkOyWLbM9bAdOdKShTGmeAumDQMRuRY4H4gU73ZkVX3Gx7iKn5dfdrPf3X57hsXPP++6zt53X4jiMsaYAhLMjXv/AW4G7sdVSfUB6vscV/ESH+8GCPzzn6FChZOL16yBadNcsrBpt40xxV0wjd4Xq2p/YJ+qPo0biLCxv2EVM+PGucEGhwzJsPjvf3d3cWdqAzfGmGIpmIRxzPt5RETOAZKA2v6FVMwcPgzjx0PPnhAdfXLxpk3w3//CoEGn3L9njDHFUjBtGJ9682n/A1gKKPCGr1EVJ5Mmwb59p8x58Y9/uNFnhw0LUVzGGFPAckwY3sRJ871JjT4Skc+ASFU9UCjRFXWpqa6xOyYGLr745OIdO9wN3/37Q926OexvjDHFSI5VUt60qK8EvD5uySLA3LmuZTvTnBf/+pe76fvRR0MYmzHGFLBg2jDmi8iNItnMAhTOxoyB2rWhT5+Ti/buhVdfhZtugkaNQhibMcYUsGASxl24wQaPi8hBETkkIgd9jqvoW7nSjSJ4330Z5tseNw4SE+Hxx0MYmzHG+CCYKVorquoZqlpGVSt5r4OaxUFEuopInIisF5HHslhfX0Tmi8gKEVkoIlEB6waIyDrvMSBvp1UIXn7Z9Zm9666TixIT3eLrr4cWLUIYmzHG+CDXXlIicnlWyzNPqJTFfhG49o/OQDywWERmquqqgM1GAxNV9V0R6Qg8D9wmItWA/wNicL2ylnj77gvmpHy3e7frHXXbbVCjxsnFr7/uqqSeeCKEsRljjE+C6Vb7l4DnkcBFwBKgYy77XQSsV9UNACIyFegBBCaM5sDD3vMFwAzv+dXAF6q619v3C6ArMCWIeP336qtw7FiGOS+OHYPRo93I5n/8YwhjM8YYnwQz+OD1ga9FpC4wJoj3roMb2TZNPNAu0zY/A72Al4GeQEURqZ7NvnWCOKb/Dh1yjd3du8P5559cPGGC6047eXLoQjPGGD8F0+idWTxQUFMADQOuEJFlwBXAVtyMfkERkcEiEisisQkJCQUUUi5ee83dqPe3v51clJTkhgFp1w6uvLJwwjDGmMIWTBvGv3HtCOASTCvcHd+52QoE3rYW5S07SVW34UoYeMOn36iq+0VkK9Ah074LMx9AVccD48HNhxFETKfnyBF46SXo0gUuuujk4qlT3VwXY8dmuB3DGGNKlGDaMAJnJUoGpqjqt0HstxhoJCINcImiL3BL4AYiUgPY690g+DjwtrdqLvCciFT1Xnfx1ofWm2/Crl0wfPjJRampbgjzCy5wEyUZY0xJFUzCmAYcU9UUcL2fRORMVT2S006qmiwiQ3AX/wjgbVVdKSLPALGqOhNXinheRBT4BrjP23eviDyLSzoAz6Q1gIfM8ePw4otw+eVw2WUnF8+YAatXw5QpGeZNMsaYEieYKVp/AK5Km2nPqzqap6oX57hjIfN9itbXX4e773Y363XuDLgRzWNi4OBBN0JIRIR/hzfGGD8U6BStuMEGT07LqqqJInJmvqMrjpKS4IUXXKv2VVedXDxvHixd6mqqLFkYY0q6YCpRDotIm7QXItIWOOpfSEXQe++5Vu3hwzO0ao8aBVFR7v49Y4wp6YIpYTwIfCgi23BTtJ6Nm7I1PKSkwHPPQcuWGVq1Fy1yj5dfzjCUlDHGlFjB3Li3WESaAk28RXGqmuRvWEXItGmwdq2bszugdPHcc1CzppvG2xhjwkGuVVIich9QXlV/VdVfgQoicq//oRUBqakwciQ0awa9ep1cvGwZzJnjJtk7M7xac4wxYSyYNoxB3ox7AHgDAA7yL6Qi5NNP4ddf3WiCAX1m5851PwMGqjXGmBIvmIQRETh5kjcKbcmvtVd1pYvzzoO+fTOsiotz8yZVqxai2IwxJgSCafSeA7wvIq97r+8CZvsXUhExbx7Exro+s6UyfkxxcdCkSTb7GWNMCRVMCeNR4Cvgbu/xC1DOz6BCThWefRbq1s2yz+zatZYwjDHhJ5gZ91KBH4GNuDkuOgKr/Q0rxL7+Gr79Fh599JQ+s3v2uEfjxiGKzRhjQiTbKikRaQz08x67gfcBVLXkD+A9ciScfTbccccpq+Li3E8rYRhjwk1ObRhrgEXAdaq6HkBEHiqUqELp++9h/nw3fV65U2veLGEYY8JVTlVSvYDtwAIReUNEOuHu9C7ZRo2C6tWz7TO7di2ULg3R0YUbljHGhFq2CUNVZ6hqX6Apbr7tB4FaIvKaiHQprAAL1bJl8Pnn7o68ChWy3CQuzvW0LRVM/zJjjClBgmn0Pqyq73lze0cBy3A9p0qeUaOgcmUYMiTbTaxLrTEmXOVpyh9V3aeq41W1k18BhczKlfDRR/DAAy5pZCElBdavt4RhjAlPNkdcmueeg/LlYejQbDfZtAlOnLCEYYwJT5YwANatg6lT4d57XYN3NtJ6SNk9GMaYcORrwhCRriISJyLrReSxLNbXE5EFIrJMRFaISDdveWkReVdEfhGR1SLyuJ9x8sIL7ga9hx/OcTPrUmuMCWe+JQxvkMJXgGuA5kA/EWmeabPhwAeq2hroC7zqLe8DlFXVC4C2wF0iEu1LoJs2wcSJMGiQu1kvB3FxULUq1KjhSyTGGFOk+dk59CJgvapuABCRqUAPYFXANgpU8p5XBrYFLC8vIqVw41adAA76EmXt2vDqq9C1a66bpo0hJSX/bhRjjDmFn1VSdYAtAa/jvWWBRgC3ikg8MAu431s+DTiMu3FwMzBaVfdmPoCIDBaRWBGJTUhIyF+UZcq40kXdurluGhdn7RfGmPAV6kbvfsAEVY0CugGTROQMXOkkBTgHaAA8IiLnZt7Z6+Ibo6oxNWvW9DXQxETYutXaL4wx4cvPhLEVCPzaHuUtC3Qn8AGAqn4PRAI1gFuAOaqapKq7gG+BGB9jzdXate6nJQxjTLjyM2EsBhqJSAMRKYNr1J6ZaZvNQCcAEWmGSxgJ3vKO3vLywB9xgyGGjCUMY0y48y1hqGoyMASYi5s/4wNVXSkiz4hId2+zR4BBIvIzMAUYqKqK611VQURW4hLPO6q6wq9YgxEX5xq7zzsvlFEYY0zo+DqEnqrOwjVmBy57KuD5KuCSLPZLxHWtLTLi4qB+/SxHPDfGmLAQ6kbvYsMGHTTGhDtLGEFQtXm8jTHGEkYQtm933WrtHgxjTDizhBEEG0PKGGMsYQTFEoYxxljCCMrata53VJ3MA5sYY0wYsYQRhLQxpM6wT8sYE8bsEhgE61JrjDGWMHJ1/Dj8/rslDGOMsYSRiw0bIDXVutQaY4wljFxYDyljjHEsYeTCEoYxxjiWMHIRF+em+q5UKfdtjTGmJLOEkYu1a639whhjwBJGrqxLrTHGOJYwcrB3L+zebQnDGGPAEkaOrMHbGGPS+ZowRKSriMSJyHoReSyL9fVEZIGILBORFSLSLWBdCxH5XkRWisgvIhLpZ6xZSZvH29owjDHGxylaRSQCNzd3ZyAeWCwiM71pWdMMx831/ZqINMdN5xotIqWA/wK3qerPIlIdSPIr1uzExUGpUtCgQWEf2Rhjih4/SxgXAetVdYOqngCmAj0ybaNAWofVysA273kXYIWq/gygqntUNcXHWLMUFwfnnQelSxf2kY0xpujxM2HUAbYEvI73lgUaAdwqIvG40sX93vLGgIrIXBFZKiJ/zeoAIjJYRGJFJDYhIaFgo8d6SBljTKBQN3r3AyaoahTQDZgkImfgqsouBf7k/ewpIp0y76yq41U1RlVjatasWaCBpaTA+vXWfmGMMWn8TBhbgboBr6O8ZYHuBD4AUNXvgUigBq408o2q7lbVI7jSRxsfYz3F5s1upForYRhjjONnwlgMNBKRBiJSBugLzMy0zWagE4CINMMljARgLnCBiJzpNYBfAayiEFmXWmOMyci3XlKqmiwiQ3AX/wjgbVVdKSLPALGqOhN4BHhDRB7CNYAPVFUF9onIP3FJR4FZqvq5X7FmxRKGMcZk5FvCAFDVWbjqpMBlTwU8XwVcks2+/8V1rQ2JtWuhcmUo4KYRY4wptkLd6F1kpfWQEgl1JMYYUzRYwsiGdak1xpiMLGFk4fBhiI+3hGGMMYEsYWRh3Tr30+7BMMaYdJYwsmA9pIwx5lSWMLKQljAaNQptHMYYU5RYwshCXBzUqwflyoU6EmOMKTosYWRh7VqrjjLGmMwsYWSial1qjTEmK5YwMtmxAw4dsoRhjDGZWcLIJK3B27rUGmNMRpYwMkmbx9tKGMYYk5EljEzi4lzvqLp1c9/WGGPCiSWMTOLi3P0XZ9gnY4wxGdhlMZO4OGu/MMaYrFjCCHDiBPz+u7VfGGNMVixhBNiwAVJSLGEYY0xWLGEEsEEHjTEme74mDBHpKiJxIrJeRB7LYn09EVkgIstEZIWIdMtifaKIDPMzzjR2D4YxxmTPt4QhIhHAK8A1QHOgn4g0z7TZcOADVW0N9AVezbT+n8Bsv2LMbO1aqFULqlQprCMaY0zx4WcJ4yJgvapuUNUTwFSgR6ZtFKjkPa8MbEtbISI3AL8DK32MMQMbQ8oYY7LnZ8KoA2wJeB3vLQs0ArhVROKBWcD9ACJSAXgUeDqnA4jIYBGJFZHYhISE0w7YEoYxxmQv1I3e/YAJqhoFdAMmicgZuETyL1VNzGlnVR2vqjGqGlOzZs3TCmTfPkhIsPYLY4zJTikf33srEDjARpS3LNCdQFcAVf1eRCKBGkA7oLeIvAhUAVJF5JiqjvMrWBtDyhhjcuZnwlgMNBKRBrhE0Re4JdM2m4FOwAQRaQZEAgmqelnaBiIyAkj0M1mAdak1xpjc+FYlparJwBBgLrAa1xtqpYg8IyLdvc0eAQaJyM/AFGCgqqpfMeUkLg4iIuDcc0NxdGOMKfr8LGGgqrNwjdmBy54KeL4KuCSX9xjhS3CZxMW5ZFG6dGEczRhjip9QN3oXGTaPtzHG5MwSBpCaCuvWWcIwxpicWMIANm+GY8csYRhjTE4sYWBjSBljTDAsYWD3YBhjTDAsYeBKGJUqwVlnhToSY4wpuixhkD4tq0ioIzHGmKLLEgY26KAxxgQj7BPGkSOwZYslDGOMyU3YJ4zEROjXD9q3D3UkxhhTtPk6NEhxUKsWvPdeqKMwxpiiL+xLGMYYY4JjCcMYY0xQLGEYY4wJiiUMY4wxQbGEYYwxJiiWMIwxxgTFEoYxxpigWMIwxhgTFFHVUMdQIEQkAdh0Gm9RA9hdQOEUN3bu4Suczz+czx3Sz7++qtYMZocSkzBOl4jEqmpMqOMIBTv38Dx3CO/zD+dzh/ydv1VJGWOMCYolDGOMMUGxhJFufKgDCCE79/AVzucfzucO+Th/a8MwxhgTFCthGGOMCYolDGOMMUEJ+4QhIl1FJE5E1ovIY6GOp7CJyEYR+UVElotIbKjj8ZOIvC0iu0Tk14Bl1UTkCxFZ5/2sGsoY/ZTN+Y8Qka3e73+5iHQLZYx+EZG6IrJARFaJyEoRGeotL/G//xzOPc+/+7BuwxCRCGAt0BmIBxYD/VR1VUgDK0QishGIUdUSfwOTiFwOJAITVfUP3rIXgb2q+oL3haGqqj4ayjj9ks35jwASVXV0KGPzm4jUBmqr6lIRqQgsAW4ABlLCf/85nPtN5PF3H+4ljIuA9aq6QVVPAFOBHiGOyfhEVb8B9mZa3AN413v+Lu4fqUTK5vzDgqpuV9Wl3vNDwGqgDmHw+8/h3PMs3BNGHWBLwOt48vlBFmMKzBORJSIyONTBhMBZqrrde74DOCuUwYTIEBFZ4VVZlbgqmcxEJBpoDfxImP3+M5075PF3H+4Jw8ClqtoGuAa4z6u2CEvq6mfDrY72NeA8oBWwHXgptOH4S0QqAB8BD6rqwcB1Jf33n8W55/l3H+4JYytQN+B1lLcsbKjqVu/nLuBjXDVdONnp1fGm1fXuCnE8hUpVd6pqiqqmAm9Qgn//IlIad8GcrKrTvcVh8fvP6tzz87sP94SxGGgkIg1EpAzQF5gZ4pgKjYiU9xrBEJHyQBfg15z3KnFmAgO85wOAT0IYS6FLu1h6elJCf/8iIsBbwGpV/WfAqhL/+8/u3PPzuw/rXlIAXleyMUAE8LaqjgpxSIVGRM7FlSoASgHvleTzF5EpQAfcsM47gf8DZgAfAPVww+PfpKolsmE4m/PvgKuSUGAjcFdAnX6JISKXAouAX4BUb/ETuLr8Ev37z+Hc+5HH333YJwxjjDHBCfcqKWOMMUGyhGGMMSYoljCMMcYExRKGMcaYoFjCMMYYExRLGMbkgYikBIzuubwgRzgWkejAkWSNKWpKhToAY4qZo6raKtRBGBMKVsIwpgB484q86M0t8pOINPSWR4vIV94Ab/NFpJ63/CwR+VhEfvYeF3tvFSEib3jzFswTkXIhOyljMrGEYUzelMtUJXVzwLoDqnoBMA43egDAv4F3VbUFMBkY6y0fC3ytqi2BNsBKb3kj4BVVPR/YD9zo8/kYEzS709uYPBCRRFWtkMXyjUBHVd3gDfS2Q1Wri8hu3OQ1Sd7y7apaQ0QSgChVPR7wHtHAF6rayHv9KFBaVUf6f2bG5M5KGMYUHM3meV4cD3iegrUzmiLEEoYxBefmgJ/fe8+/w42CDPAn3CBwAPOBe8BNFSwilQsrSGPyy769GJM35URkecDrOaqa1rW2qoiswJUS+nnL7gfeEZG/AAnA7d7yocB4EbkTV5K4BzeJjTFFlrVhGFMAvDaMGFXdHepYjPGLVUkZY4wJipUwjDHGBMVKGMYYY4JiCcMYY0xQLGEYY4wJiiUMY4wxQbGEYYwxJij/D6h1Kk2G4RG4AAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "dFc3yrqeXuGw",
        "outputId": "f30849a3-8696-4303-e3b6-356ee10f6a92"
      },
      "source": [
        "from sklearn.metrics import classification_report\n",
        "\n",
        "y_val_pred = biLSTM1_model.predict(x_val)\n",
        "y_val_pred_flat = flatten_pred(y_val_pred)\n",
        "y_val_flat = flatten_pred(y_val)\n",
        "print(classification_report(y_val_pred_flat, y_val_flat))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.97      0.96      0.97     97633\n",
            "           1       0.92      0.95      0.93      5511\n",
            "           2       0.95      0.93      0.94      7653\n",
            "           3       0.86      0.90      0.88     20123\n",
            "\n",
            "    accuracy                           0.95    130920\n",
            "   macro avg       0.92      0.93      0.93    130920\n",
            "weighted avg       0.95      0.95      0.95    130920\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gbihG0R0ns3c"
      },
      "source": [
        "#BidirectionalLSTM + LSTM\n",
        "\n",
        "Learning Rate - 0.001\n",
        "Dropout for 1st Layer - 0.2\n",
        "Dropout for 2nd layer - 0.1"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "vnvj8atap_IX",
        "outputId": "5d862068-8bdc-415b-d37e-cdc6ad252cb9"
      },
      "source": [
        "x_train.shape"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(29449, 40, 6)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 16
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "F52u9zfonvq2"
      },
      "source": [
        "def create_LSTM():\n",
        "  model = Sequential()\n",
        "  model.add(Bidirectional(LSTM(100,return_sequences=True, input_shape=(40,6))))\n",
        "  model.add(Dropout(0.2))\n",
        "  model.add(LSTM(50, return_sequences=True, activation=\"relu\"))\n",
        "  model.add(Dropout(0.1))\n",
        "  model.add(Dense(25, activation=\"relu\"))\n",
        "  model.add(Dense(4, activation=\"softmax\"))\n",
        "  model.compile(optimizer=Adam(learning_rate=0.001),  loss='categorical_crossentropy', metrics = ['accuracy'])\n",
        "  return model"
      ],
      "execution_count": 17,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "02Q-n-hDpulg",
        "outputId": "c79769b0-8490-414e-e5e7-7f1ea57ee924"
      },
      "source": [
        "lstm_model.summary()\n",
        "lstm_model =create_LSTM()\n",
        "history_lstm = lstm_model.fit(x_train, y_train, epochs=25, verbose=1, validation_data=(x_val, y_val))"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "bidirectional (Bidirectional (None, 40, 200)           85600     \n",
            "_________________________________________________________________\n",
            "dropout (Dropout)            (None, 40, 200)           0         \n",
            "_________________________________________________________________\n",
            "lstm_1 (LSTM)                (None, 40, 50)            50200     \n",
            "_________________________________________________________________\n",
            "dropout_1 (Dropout)          (None, 40, 50)            0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 40, 25)            1275      \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 40, 4)             104       \n",
            "=================================================================\n",
            "Total params: 137,179\n",
            "Trainable params: 137,179\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "WARNING:tensorflow:Layer lstm_3 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",
            "Epoch 1/25\n",
            "921/921 [==============================] - 65s 67ms/step - loss: 0.6113 - accuracy: 0.7859 - val_loss: 0.3276 - val_accuracy: 0.8856\n",
            "Epoch 2/25\n",
            "921/921 [==============================] - 60s 66ms/step - loss: 0.3060 - accuracy: 0.8914 - val_loss: 0.2488 - val_accuracy: 0.9178\n",
            "Epoch 3/25\n",
            "921/921 [==============================] - 61s 66ms/step - loss: 0.2542 - accuracy: 0.9113 - val_loss: 0.2672 - val_accuracy: 0.8997\n",
            "Epoch 4/25\n",
            "921/921 [==============================] - 61s 66ms/step - loss: 0.2217 - accuracy: 0.9243 - val_loss: 0.2224 - val_accuracy: 0.9259\n",
            "Epoch 5/25\n",
            "921/921 [==============================] - 60s 66ms/step - loss: 0.1973 - accuracy: 0.9329 - val_loss: 0.1951 - val_accuracy: 0.9351\n",
            "Epoch 6/25\n",
            "921/921 [==============================] - 61s 66ms/step - loss: 0.1848 - accuracy: 0.9379 - val_loss: 0.1836 - val_accuracy: 0.9394\n",
            "Epoch 7/25\n",
            "921/921 [==============================] - 61s 66ms/step - loss: 0.1735 - accuracy: 0.9408 - val_loss: 0.1918 - val_accuracy: 0.9353\n",
            "Epoch 8/25\n",
            "921/921 [==============================] - 61s 66ms/step - loss: 0.1593 - accuracy: 0.9455 - val_loss: 0.1788 - val_accuracy: 0.9399\n",
            "Epoch 9/25\n",
            "921/921 [==============================] - 61s 66ms/step - loss: 0.1514 - accuracy: 0.9490 - val_loss: 0.1812 - val_accuracy: 0.9379\n",
            "Epoch 10/25\n",
            "921/921 [==============================] - 61s 66ms/step - loss: 0.1425 - accuracy: 0.9520 - val_loss: 0.1762 - val_accuracy: 0.9435\n",
            "Epoch 11/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.1387 - accuracy: 0.9533 - val_loss: 0.1799 - val_accuracy: 0.9443\n",
            "Epoch 12/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.1254 - accuracy: 0.9581 - val_loss: 0.1654 - val_accuracy: 0.9490\n",
            "Epoch 13/25\n",
            "921/921 [==============================] - 61s 66ms/step - loss: 0.1217 - accuracy: 0.9578 - val_loss: 0.1717 - val_accuracy: 0.9473\n",
            "Epoch 14/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.1232 - accuracy: 0.9578 - val_loss: 0.1972 - val_accuracy: 0.9421\n",
            "Epoch 15/25\n",
            "921/921 [==============================] - 59s 64ms/step - loss: 0.1088 - accuracy: 0.9631 - val_loss: 0.1685 - val_accuracy: 0.9468\n",
            "Epoch 16/25\n",
            "921/921 [==============================] - 60s 66ms/step - loss: 0.1019 - accuracy: 0.9656 - val_loss: 0.1647 - val_accuracy: 0.9479\n",
            "Epoch 17/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.0953 - accuracy: 0.9678 - val_loss: 0.1671 - val_accuracy: 0.9456\n",
            "Epoch 18/25\n",
            "921/921 [==============================] - 61s 66ms/step - loss: 0.0891 - accuracy: 0.9709 - val_loss: 0.1730 - val_accuracy: 0.9441\n",
            "Epoch 19/25\n",
            "921/921 [==============================] - 61s 66ms/step - loss: 0.0866 - accuracy: 0.9714 - val_loss: 0.1644 - val_accuracy: 0.9489\n",
            "Epoch 20/25\n",
            "921/921 [==============================] - 61s 66ms/step - loss: 0.0795 - accuracy: 0.9740 - val_loss: 0.1641 - val_accuracy: 0.9510\n",
            "Epoch 21/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.0762 - accuracy: 0.9748 - val_loss: 0.1714 - val_accuracy: 0.9473\n",
            "Epoch 22/25\n",
            "921/921 [==============================] - 60s 65ms/step - loss: 0.0733 - accuracy: 0.9760 - val_loss: 0.1794 - val_accuracy: 0.9475\n",
            "Epoch 23/25\n",
            "921/921 [==============================] - 60s 66ms/step - loss: 0.0700 - accuracy: 0.9771 - val_loss: 0.1887 - val_accuracy: 0.9477\n",
            "Epoch 24/25\n",
            "921/921 [==============================] - 60s 66ms/step - loss: 0.0655 - accuracy: 0.9789 - val_loss: 0.2102 - val_accuracy: 0.9459\n",
            "Epoch 25/25\n",
            "921/921 [==============================] - 61s 66ms/step - loss: 0.0660 - accuracy: 0.9791 - val_loss: 0.1830 - val_accuracy: 0.9452\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dB3p2XswNubz"
      },
      "source": [
        "## Evaluating performace of the model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 573
        },
        "id": "1LE0NvxQRETY",
        "outputId": "591f3518-d7e7-49fa-d7a8-b2f975091d4d"
      },
      "source": [
        "#Plot the loss and accuracy for the model\n",
        "plot_history(history_lstm)"
      ],
      "execution_count": 21,
      "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"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Ii4uvjNCpnBN",
        "outputId": "b49e5c62-7b3e-4713-f665-f6e721cb0eea"
      },
      "source": [
        "from sklearn.metrics import classification_report\n",
        "\n",
        "y_val_pred = lstm_model.predict(x_val)\n",
        "y_val_pred_flat = flatten_pred(y_val_pred)\n",
        "y_val_flat = flatten_pred(y_val)\n",
        "print(classification_report(y_val_pred_flat, y_val_flat))"
      ],
      "execution_count": 55,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.96      0.96      0.96     96620\n",
            "           1       0.93      0.92      0.93      5725\n",
            "           2       0.93      0.95      0.94      7383\n",
            "           3       0.87      0.87      0.87     21192\n",
            "\n",
            "    accuracy                           0.95    130920\n",
            "   macro avg       0.92      0.93      0.93    130920\n",
            "weighted avg       0.95      0.95      0.95    130920\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "f7D7j11EiZJq"
      },
      "source": [
        "# Prediction"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Qs35ZE7oL3du"
      },
      "source": [
        "## Read the test data files"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "p4oK2RDeMf-z"
      },
      "source": [
        "# Read all test data files \n",
        "x_test_009 = pd.read_csv('Test_Data_allx/subject_009_01x_meas_time.csv')\n",
        "x_test_010 = pd.read_csv('Test_Data_allx/subject_010_01x_meas_time.csv')\n",
        "x_test_011 = pd.read_csv('Test_Data_allx/subject_011_01x_meas_time.csv')\n",
        "x_test_012 = pd.read_csv('Test_Data_allx/subject_012_01x_meas_time.csv')\n",
        "\n",
        "x_test_009 = x_test_009.drop('time', axis=1)\n",
        "x_test_010 = x_test_010.drop('time', axis=1)\n",
        "x_test_011 = x_test_011.drop('time', axis=1)\n",
        "x_test_012 = x_test_012.drop('time', axis=1)"
      ],
      "execution_count": 22,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "UOliYNc8L7mg"
      },
      "source": [
        "## Standardize the Test Data set\n",
        "\n",
        "Standardize the test data set using .transform function.\n",
        "We are standardizing the test data set to the same scale to the training dataset."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OdJZhGzCMydb"
      },
      "source": [
        "x_test_009_scaled = scaler.transform(x_test_009)\n",
        "x_test_009 = pd.DataFrame(x_test_009_scaled, columns = ['ax','ay','az','gx','gy','gz'])\n",
        "\n",
        "x_test_010_scaled = scaler.transform(x_test_010)\n",
        "x_test_010 = pd.DataFrame(x_test_010_scaled, columns = ['ax','ay','az','gx','gy','gz'])\n",
        "\n",
        "x_test_011_scaled = scaler.transform(x_test_011)\n",
        "x_test_011 = pd.DataFrame(x_test_011_scaled, columns = ['ax','ay','az','gx','gy','gz'])\n",
        "\n",
        "x_test_012_scaled = scaler.transform(x_test_012)\n",
        "x_test_012 = pd.DataFrame(x_test_012_scaled, columns = ['ax','ay','az','gx','gy','gz'])"
      ],
      "execution_count": 37,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ChP5BYcXMR1y"
      },
      "source": [
        "## Creatiing Windows for the test dataset"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9-HevOl4M1KE"
      },
      "source": [
        "# Defining Function to Create window for test dataset\n",
        "def define_test_windows(df_x, windowsize=40):\n",
        "  window_x=[]\n",
        "  for i in range(0, len(df_x)-windowsize, windowsize):\n",
        "    window_x.append([df_x.values[i:i+windowsize]])\n",
        "  window_x=np.array(window_x).reshape(-1,windowsize,6)\n",
        "  \n",
        "  return window_x\n"
      ],
      "execution_count": 38,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DkDXXCbfM7G7"
      },
      "source": [
        "#Creating frames for all test subjects\n",
        "x_09 = define_test_windows(x_test_009, wsize)\n",
        "x_10 = define_test_windows(x_test_010, wsize)\n",
        "x_11 = define_test_windows(x_test_011, wsize)\n",
        "x_12 = define_test_windows(x_test_012, wsize)"
      ],
      "execution_count": 39,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fi69HlgAUvEU",
        "outputId": "cf408ec4-c21f-4a90-a640-ac8822b14a8e"
      },
      "source": [
        "x_09.shape"
      ],
      "execution_count": 40,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(949, 40, 6)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 40
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2isaTmUtMWkQ"
      },
      "source": [
        "## Perform prediction and transform the predicted categorical values to integer labels"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dNpt1vuqNAOH"
      },
      "source": [
        "def predictions(model):\n",
        "  y_pred_009 = model.predict(x_09)\n",
        "  y_pred_010 = model.predict(x_10)\n",
        "  y_pred_011 = model.predict(x_11)\n",
        "  y_pred_012 = model.predict(x_12)\n",
        "\n",
        "  #prediction for Subject 09\n",
        "  y_009 = y_pred_009.tolist()\n",
        "  yhat_009 = flatten_pred(y_009)\n",
        "\n",
        "  #prediction for Subject 10\n",
        "  y_010 = y_pred_010.tolist()\n",
        "  yhat_010 = flatten_pred(y_010)\n",
        "\n",
        "  #prediction for Subject 11\n",
        "  y_011 = y_pred_011.tolist()\n",
        "  yhat_011 = flatten_pred(y_011)\n",
        "\n",
        "  #prediction for Subject 12\n",
        "  y_012 = y_pred_012.tolist()\n",
        "  yhat_012 = flatten_pred(y_012)\n",
        "\n",
        "  print(y_pred_009.shape, y_pred_010.shape, y_pred_011.shape, y_pred_012.shape)\n",
        "\n",
        "  return yhat_009, yhat_010, yhat_011, yhat_012"
      ],
      "execution_count": 41,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "3J_bDg_aNDy1",
        "outputId": "13d8a141-c8a9-4fc0-81c2-5a318835ed60"
      },
      "source": [
        "#Predictions for all test subjects\n",
        "yhat_009, yhat_010, yhat_011, yhat_012 = predictions(lstm_model)"
      ],
      "execution_count": 42,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(949, 40, 4) (1227, 40, 4) (1294, 40, 4) (1132, 40, 4)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RVq3FbA8lged"
      },
      "source": [
        "yhat_009_df=pd.DataFrame(yhat_009, columns=['label'])\n",
        "yhat_010_df=pd.DataFrame(yhat_010, columns=['label'])\n",
        "yhat_011_df=pd.DataFrame(yhat_011, columns=['label'])\n",
        "yhat_012_df=pd.DataFrame(yhat_012, columns=['label'])"
      ],
      "execution_count": 43,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kgQgs5LYkTmQ"
      },
      "source": [
        "# Read all test data files \n",
        "x_meas9 = pd.read_csv('Test_Data_allx/subject_009_01x_meas_time.csv')\n",
        "x_meas10 = pd.read_csv('Test_Data_allx/subject_010_01x_meas_time.csv')\n",
        "x_meas11 = pd.read_csv('Test_Data_allx/subject_011_01x_meas_time.csv')\n",
        "x_meas12 = pd.read_csv('Test_Data_allx/subject_012_01x_meas_time.csv')"
      ],
      "execution_count": 44,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tZJ9sfi1mLrY"
      },
      "source": [
        "x_meas9 = x_meas9.join(yhat_009_df, how='left')\n",
        "x_meas10 = x_meas10.join(yhat_010_df, how='left')\n",
        "x_meas11 = x_meas11.join(yhat_011_df, how='left')\n",
        "x_meas12 = x_meas12.join(yhat_012_df, how='left')\n",
        "\n",
        "x_meas9['label'] = x_meas9['label'].fillna(0)\n",
        "x_meas10['label'] = x_meas10['label'].fillna(0)\n",
        "x_meas11['label'] = x_meas11['label'].fillna(0)\n",
        "x_meas12['label'] = x_meas12['label'].fillna(0)"
      ],
      "execution_count": 45,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OdRZNLyRmwMo"
      },
      "source": [
        "y_time_009 = pd.read_csv('Test_Data_allx/subject_009_01y_time.csv')\n",
        "y_time_010 = pd.read_csv('Test_Data_allx/subject_010_01y_time.csv')\n",
        "y_time_011 = pd.read_csv('Test_Data_allx/subject_011_01y_time.csv')\n",
        "y_time_012 = pd.read_csv('Test_Data_allx/subject_012_01y_time.csv')"
      ],
      "execution_count": 46,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uRzZuXKdmiIq"
      },
      "source": [
        "xmeasfinal9 = pd.merge(x_meas9, y_time_009, on='time', how='inner')\n",
        "xmeasfinal10 = pd.merge(x_meas10, y_time_010, on='time', how='inner')\n",
        "xmeasfinal11 = pd.merge(x_meas11, y_time_011, on='time', how='inner')\n",
        "xmeasfinal12 = pd.merge(x_meas12, y_time_012, on='time', how='inner')"
      ],
      "execution_count": 47,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FA5XxkCEn1RM"
      },
      "source": [
        "# Change the test result directory before running this cell\n",
        "xmeasfinal9['label'].to_csv(\"TestResults/new/tuning/subject_009_01__y_prediction.csv\", index=False, header=False)\n",
        "xmeasfinal10['label'].to_csv(\"TestResults/new/tuning/subject_010_01__y_prediction.csv\", index=False, header=False)\n",
        "xmeasfinal11['label'].to_csv(\"TestResults/new/tuning/subject_011_01__y_prediction.csv\", index=False, header=False)\n",
        "xmeasfinal12['label'].to_csv(\"TestResults/new/tuning/subject_012_01__y_prediction.csv\", index=False, header=False)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pqaOgdwD_Ofu"
      },
      "source": [
        "#Illustration of Test dataset\n",
        "\n",
        "Taking the labels of subject 9 and checking the labels and identifying the prediction pattern"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "aPF0zqTi_SVl",
        "outputId": "6c182a64-5f3e-405f-d58a-1629b2aa470b"
      },
      "source": [
        "x_measfinal9 = pd.read_csv('TestResults/new/subject_009_01__y_prediction.csv')\n",
        "x_measfinal10 = pd.read_csv('TestResults/new/subject_010_01__y_prediction.csv')\n",
        "x_measfinal11 = pd.read_csv('TestResults/new/subject_011_01__y_prediction.csv')\n",
        "x_measfinal12 = pd.read_csv('TestResults/new/subject_012_01__y_prediction.csv')\n",
        "print(x_measfinal9.value_counts())\n",
        "print(x_measfinal10.value_counts())\n",
        "print(x_measfinal11.value_counts())\n",
        "print(x_measfinal12.value_counts())"
      ],
      "execution_count": 38,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.0\n",
            "0.0    8586\n",
            "3.0     422\n",
            "1.0     340\n",
            "2.0     149\n",
            "dtype: int64\n",
            "0.0\n",
            "0.0    8007\n",
            "3.0    3348\n",
            "2.0     467\n",
            "1.0     447\n",
            "dtype: int64\n",
            "0.0\n",
            "0.0    8979\n",
            "3.0    2747\n",
            "2.0     738\n",
            "1.0     475\n",
            "dtype: int64\n",
            "0.0\n",
            "0.0    8869\n",
            "3.0     988\n",
            "2.0     852\n",
            "1.0     620\n",
            "dtype: int64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 592
        },
        "id": "oUtulaLlErR-",
        "outputId": "b597d197-5ac5-4f06-f67a-87463257c0af"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "plt.figure(figsize=(20, 10))\n",
        "plt.plot(x_measfinal9)\n",
        "plt.show()"
      ],
      "execution_count": 33,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x720 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 513
        },
        "id": "ciLCX1GDG1Ev",
        "outputId": "fd0c5f0e-7d5d-4da5-8813-124d48788a65"
      },
      "source": [
        "plt.figure(figsize=(15, 8))\n",
        "plt.plot(x_measfinal9[7000:8000])\n",
        "plt.ylabel('Label')\n",
        "plt.xlabel('# of Samples for Subject 9')\n",
        "plt.title('Trend plot for subsample for Subject 9')\n",
        "plt.show()"
      ],
      "execution_count": 36,
      "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"
          }
        }
      ]
    }
  ]
}