05_Interpretable_Modeling.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "05_Interpretable_Modeling.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyN9QR9DghRv7kG4dC6kVdWY",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/BradenAnderson/sales-predictions/blob/main/05_Interpretable_Modeling.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vUHabgnNEflf"
      },
      "source": [
        "## Braden Anderson\r\n",
        "## Sales Predictions Project part 5\r\n",
        "## Interpretable (Linear) Modeling"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "tOiCDaeOF-nu",
        "outputId": "5ab765bc-34ab-4861-e90d-ed4f730fd977"
      },
      "source": [
        "from google.colab import drive\r\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Cho0G2MlGkrQ"
      },
      "source": [
        "import missingno\r\n",
        "import pickle\r\n",
        "import os\r\n",
        "import pandas as pd\r\n",
        "import numpy as np\r\n",
        "import seaborn as sns\r\n",
        "import matplotlib.pyplot as plt\r\n",
        "\r\n",
        "from scipy import stats\r\n",
        "\r\n",
        "from sklearn.impute import SimpleImputer\r\n",
        "from sklearn.experimental import enable_iterative_imputer\r\n",
        "from sklearn.impute import IterativeImputer\r\n",
        "from sklearn.model_selection import cross_val_score, GridSearchCV\r\n",
        "from sklearn.neighbors import KNeighborsRegressor\r\n",
        "from sklearn.linear_model import BayesianRidge, LinearRegression, Lasso, Ridge\r\n",
        "from sklearn.tree import DecisionTreeRegressor\r\n",
        "from sklearn.ensemble import ExtraTreesRegressor, RandomForestRegressor, BaggingRegressor \r\n",
        "from sklearn.preprocessing import OrdinalEncoder, OneHotEncoder, StandardScaler, PowerTransformer\r\n",
        "from sklearn.compose import ColumnTransformer, TransformedTargetRegressor\r\n",
        "from sklearn.feature_selection import SelectKBest, f_regression\r\n",
        "from sklearn.metrics import r2_score, mean_squared_error, SCORERS\r\n",
        "from sklearn.pipeline import Pipeline\r\n",
        "\r\n",
        "\r\n",
        "%matplotlib inline"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "p8YiuTAJGnYK"
      },
      "source": [
        "filename = '/content/drive/MyDrive/Programming/Colab Notebooks/Coding_Dojo/Sales_Project/Current/Sales_Training_Data_Cleaned_Imputed.csv'\r\n",
        "\r\n",
        "sales_df = pd.read_csv(filename)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SsKWOQaEHOm5"
      },
      "source": [
        "sales_df.drop(columns=['Item_Identifier', 'Outlet_Establishment_Year'], inplace=True)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "U4DmnD_VHO5E"
      },
      "source": [
        "\r\n",
        "nominal_categorical_feat = ['Item_Type', 'Outlet_Identifier', 'Outlet_Location_Type', 'Outlet_Type', 'Item_Id_Binned']\r\n",
        "ordinal_categorical_feat = ['Item_Fat_Content', 'Outlet_Size', 'Years_of_operation']\r\n",
        "numeric_feat = ['Item_Weight', 'Item_Visibility', 'Item_MRP']\r\n",
        "\r\n",
        "X = sales_df.loc[:, sales_df.columns != 'Item_Outlet_Sales']\r\n",
        "y = sales_df.loc[:, 'Item_Outlet_Sales']\r\n",
        "\r\n",
        "\r\n",
        "preprocess = ColumnTransformer(transformers=[(\"one_hot_encoder\", OneHotEncoder(drop=\"first\", sparse=False), nominal_categorical_feat),\r\n",
        "                                             (\"ordinal_encoder\", OrdinalEncoder(categories=[['Not_Edible', 'Low Fat', 'Regular'], ['Small', 'Medium', 'High'],\r\n",
        "                                                                                            [4, 6, 9, 11, 14, 15, 16, 26, 28]]), ordinal_categorical_feat)],\r\n",
        "                               remainder='passthrough')\r\n",
        "\r\n",
        "\r\n",
        "\r\n",
        "column_list = list(range(5,38))\r\n",
        "\r\n",
        "reg_strategy = TransformedTargetRegressor()\r\n",
        "bcox_transformer = PowerTransformer(method='box-cox')\r\n",
        "\r\n",
        "model_pipeline = Pipeline([(\"preprocessing\", preprocess),\r\n",
        "                           ('feature_selector', SelectKBest(f_regression)),\r\n",
        "                           ('regress', reg_strategy)])\r\n",
        "\r\n",
        "parameter_grid = [{'feature_selector__k' : column_list,\r\n",
        "                   'regress__regressor' : [LinearRegression(), Lasso(), Ridge()],\r\n",
        "                   'regress__regressor__fit_intercept' : [False, True],\r\n",
        "                   'regress__transformer' : [None, bcox_transformer]}]\r\n",
        "\r\n",
        "\r\n",
        "score_types = {'r2' :'r2', 'MSE' : 'neg_mean_squared_error'}\r\n",
        "\r\n",
        "gs = GridSearchCV(estimator=model_pipeline, param_grid=parameter_grid, scoring=score_types, refit='MSE', cv=5, n_jobs=-1)\r\n",
        "\r\n",
        "gs.fit(X, y)\r\n",
        "\r\n",
        "PATH = '/content/drive/MyDrive/Programming/Colab Notebooks/Coding_Dojo/Sales_Project/Current/'\r\n",
        "gridsearch_result_filename = 'gridsearch_models_5_Linear.pkl'\r\n",
        "full_path = PATH + gridsearch_result_filename\r\n",
        "with open(full_path, 'wb') as file:\r\n",
        "  pickle.dump(gs, file)\r\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QU9MpOohKmgu"
      },
      "source": [
        "full_path = '/content/drive/MyDrive/Programming/Colab Notebooks/Coding_Dojo/Sales_Project/Current/gridsearch_models_5_Linear.pkl'\r\n",
        "\r\n",
        "with open(full_path, 'rb') as file:\r\n",
        "  gs_results = pickle.load(file)\r\n",
        "\r\n",
        "search_results = gs_results.cv_results_\r\n",
        "top_estimator = gs_results.best_estimator_\r\n",
        "top_score = gs_results.best_score_\r\n",
        "top_parameters = gs_results.best_params_\r\n",
        "\r\n",
        "gs_result_df = pd.DataFrame(search_results)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "INMZU9RnVh91",
        "outputId": "4a2b489c-1834-4921-efd1-1fb090abcb41"
      },
      "source": [
        "results_simplified_df= gs_result_df.loc[:,:].sort_values(by=['rank_test_MSE'], ignore_index=True)\r\n",
        "\r\n",
        "results_simplified_df.rename(columns= {'param_feature_selector__k' : 'num_features_in_model',\r\n",
        "                                       'param_regress__regressor' : 'model_type',\r\n",
        "                                       'param_regress__transformer' : 'target_transformation',\r\n",
        "                                       'param_regress__regressor__fit_intercept' : 'fit_intercept'}, inplace=True)\r\n",
        "\r\n",
        "results_simplified_df = results_simplified_df.loc[:, ['num_features_in_model', 'model_type', 'target_transformation', 'fit_intercept',\r\n",
        "                                                      'params', 'mean_test_r2', 'rank_test_r2', 'mean_test_MSE', 'rank_test_MSE']]\r\n",
        "\r\n",
        "\r\n",
        "results_simplified_df['target_transformation'] = results_simplified_df['target_transformation'].fillna(value=\"No_Transformation\")\r\n",
        "results_simplified_df.loc[ (results_simplified_df['target_transformation'] != 'No_Transformation') , 'target_transformation'] = \"Box-Cox\"\r\n",
        "\r\n",
        "# For easier viewing, trim model_type down so it only shows the name of the model used.\r\n",
        "results_simplified_df['model_type'] = results_simplified_df['model_type'].astype(str)\r\n",
        "results_simplified_df['model_type'] = results_simplified_df['model_type'].map(lambda model_string : model_string.split('(')[0])\r\n",
        "\r\n",
        "# Negative MSE is not very interpretable. Take the absolute value and square root to get a more meaningful number. \r\n",
        "results_simplified_df['RMSE'] = results_simplified_df['mean_test_MSE'].abs().pow(1./2)\r\n",
        "\r\n",
        "pd.set_option('display.max_rows', 500)\r\n",
        "\r\n",
        "results_simplified_df.head(50)\r\n"
      ],
      "execution_count": 51,
      "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>num_features_in_model</th>\n",
              "      <th>model_type</th>\n",
              "      <th>target_transformation</th>\n",
              "      <th>fit_intercept</th>\n",
              "      <th>params</th>\n",
              "      <th>mean_test_r2</th>\n",
              "      <th>rank_test_r2</th>\n",
              "      <th>mean_test_MSE</th>\n",
              "      <th>rank_test_MSE</th>\n",
              "      <th>RMSE</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>16</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 16, 'regress__regresso...</td>\n",
              "      <td>0.581160</td>\n",
              "      <td>1</td>\n",
              "      <td>-1.218403e+06</td>\n",
              "      <td>1</td>\n",
              "      <td>1103.812926</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>17</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 17, 'regress__regresso...</td>\n",
              "      <td>0.581146</td>\n",
              "      <td>2</td>\n",
              "      <td>-1.218451e+06</td>\n",
              "      <td>2</td>\n",
              "      <td>1103.834566</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>17</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 17, 'regress__regresso...</td>\n",
              "      <td>0.581146</td>\n",
              "      <td>3</td>\n",
              "      <td>-1.218451e+06</td>\n",
              "      <td>3</td>\n",
              "      <td>1103.834566</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>18</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 18, 'regress__regresso...</td>\n",
              "      <td>0.581100</td>\n",
              "      <td>4</td>\n",
              "      <td>-1.218591e+06</td>\n",
              "      <td>4</td>\n",
              "      <td>1103.898306</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>18</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 18, 'regress__regresso...</td>\n",
              "      <td>0.581100</td>\n",
              "      <td>5</td>\n",
              "      <td>-1.218591e+06</td>\n",
              "      <td>5</td>\n",
              "      <td>1103.898306</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>16</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 16, 'regress__regresso...</td>\n",
              "      <td>0.581058</td>\n",
              "      <td>6</td>\n",
              "      <td>-1.218700e+06</td>\n",
              "      <td>6</td>\n",
              "      <td>1103.947666</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>17</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 17, 'regress__regresso...</td>\n",
              "      <td>0.581042</td>\n",
              "      <td>7</td>\n",
              "      <td>-1.218753e+06</td>\n",
              "      <td>7</td>\n",
              "      <td>1103.971297</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>18</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 18, 'regress__regresso...</td>\n",
              "      <td>0.580997</td>\n",
              "      <td>8</td>\n",
              "      <td>-1.218893e+06</td>\n",
              "      <td>8</td>\n",
              "      <td>1104.035055</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>15</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 15, 'regress__regresso...</td>\n",
              "      <td>0.580945</td>\n",
              "      <td>9</td>\n",
              "      <td>-1.219014e+06</td>\n",
              "      <td>9</td>\n",
              "      <td>1104.089725</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>15</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 15, 'regress__regresso...</td>\n",
              "      <td>0.580849</td>\n",
              "      <td>10</td>\n",
              "      <td>-1.219294e+06</td>\n",
              "      <td>10</td>\n",
              "      <td>1104.216623</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10</th>\n",
              "      <td>14</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 14, 'regress__regresso...</td>\n",
              "      <td>0.580823</td>\n",
              "      <td>12</td>\n",
              "      <td>-1.219339e+06</td>\n",
              "      <td>11</td>\n",
              "      <td>1104.236746</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>11</th>\n",
              "      <td>17</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 17, 'regress__regresso...</td>\n",
              "      <td>0.580835</td>\n",
              "      <td>11</td>\n",
              "      <td>-1.219373e+06</td>\n",
              "      <td>12</td>\n",
              "      <td>1104.252180</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>12</th>\n",
              "      <td>12</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 12, 'regress__regresso...</td>\n",
              "      <td>0.580795</td>\n",
              "      <td>14</td>\n",
              "      <td>-1.219433e+06</td>\n",
              "      <td>13</td>\n",
              "      <td>1104.279501</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13</th>\n",
              "      <td>18</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 18, 'regress__regresso...</td>\n",
              "      <td>0.580812</td>\n",
              "      <td>13</td>\n",
              "      <td>-1.219447e+06</td>\n",
              "      <td>14</td>\n",
              "      <td>1104.285863</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>14</th>\n",
              "      <td>22</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 22, 'regress__regresso...</td>\n",
              "      <td>0.580772</td>\n",
              "      <td>15</td>\n",
              "      <td>-1.219503e+06</td>\n",
              "      <td>15</td>\n",
              "      <td>1104.311278</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>15</th>\n",
              "      <td>22</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 22, 'regress__regresso...</td>\n",
              "      <td>0.580772</td>\n",
              "      <td>16</td>\n",
              "      <td>-1.219503e+06</td>\n",
              "      <td>16</td>\n",
              "      <td>1104.311278</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16</th>\n",
              "      <td>14</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 14, 'regress__regresso...</td>\n",
              "      <td>0.580745</td>\n",
              "      <td>21</td>\n",
              "      <td>-1.219565e+06</td>\n",
              "      <td>17</td>\n",
              "      <td>1104.339243</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>17</th>\n",
              "      <td>21</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 21, 'regress__regresso...</td>\n",
              "      <td>0.580750</td>\n",
              "      <td>17</td>\n",
              "      <td>-1.219573e+06</td>\n",
              "      <td>18</td>\n",
              "      <td>1104.342736</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>18</th>\n",
              "      <td>21</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 21, 'regress__regresso...</td>\n",
              "      <td>0.580750</td>\n",
              "      <td>18</td>\n",
              "      <td>-1.219573e+06</td>\n",
              "      <td>19</td>\n",
              "      <td>1104.342736</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>19</th>\n",
              "      <td>19</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 19, 'regress__regresso...</td>\n",
              "      <td>0.580747</td>\n",
              "      <td>19</td>\n",
              "      <td>-1.219578e+06</td>\n",
              "      <td>20</td>\n",
              "      <td>1104.345085</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>20</th>\n",
              "      <td>19</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 19, 'regress__regresso...</td>\n",
              "      <td>0.580747</td>\n",
              "      <td>19</td>\n",
              "      <td>-1.219578e+06</td>\n",
              "      <td>20</td>\n",
              "      <td>1104.345085</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21</th>\n",
              "      <td>12</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 12, 'regress__regresso...</td>\n",
              "      <td>0.580738</td>\n",
              "      <td>22</td>\n",
              "      <td>-1.219593e+06</td>\n",
              "      <td>22</td>\n",
              "      <td>1104.351964</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>22</th>\n",
              "      <td>20</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 20, 'regress__regresso...</td>\n",
              "      <td>0.580734</td>\n",
              "      <td>23</td>\n",
              "      <td>-1.219613e+06</td>\n",
              "      <td>23</td>\n",
              "      <td>1104.360811</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>23</th>\n",
              "      <td>20</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 20, 'regress__regresso...</td>\n",
              "      <td>0.580734</td>\n",
              "      <td>24</td>\n",
              "      <td>-1.219613e+06</td>\n",
              "      <td>24</td>\n",
              "      <td>1104.360811</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>24</th>\n",
              "      <td>11</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 11, 'regress__regresso...</td>\n",
              "      <td>0.580719</td>\n",
              "      <td>25</td>\n",
              "      <td>-1.219647e+06</td>\n",
              "      <td>25</td>\n",
              "      <td>1104.376266</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>25</th>\n",
              "      <td>22</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 22, 'regress__regresso...</td>\n",
              "      <td>0.580675</td>\n",
              "      <td>26</td>\n",
              "      <td>-1.219790e+06</td>\n",
              "      <td>26</td>\n",
              "      <td>1104.441068</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>26</th>\n",
              "      <td>19</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 19, 'regress__regresso...</td>\n",
              "      <td>0.580655</td>\n",
              "      <td>27</td>\n",
              "      <td>-1.219850e+06</td>\n",
              "      <td>27</td>\n",
              "      <td>1104.468027</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>27</th>\n",
              "      <td>21</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 21, 'regress__regresso...</td>\n",
              "      <td>0.580654</td>\n",
              "      <td>28</td>\n",
              "      <td>-1.219855e+06</td>\n",
              "      <td>28</td>\n",
              "      <td>1104.470374</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>28</th>\n",
              "      <td>23</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 23, 'regress__regresso...</td>\n",
              "      <td>0.580643</td>\n",
              "      <td>29</td>\n",
              "      <td>-1.219874e+06</td>\n",
              "      <td>29</td>\n",
              "      <td>1104.478868</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>29</th>\n",
              "      <td>23</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 23, 'regress__regresso...</td>\n",
              "      <td>0.580643</td>\n",
              "      <td>30</td>\n",
              "      <td>-1.219874e+06</td>\n",
              "      <td>30</td>\n",
              "      <td>1104.478868</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>30</th>\n",
              "      <td>11</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 11, 'regress__regresso...</td>\n",
              "      <td>0.580639</td>\n",
              "      <td>32</td>\n",
              "      <td>-1.219876e+06</td>\n",
              "      <td>31</td>\n",
              "      <td>1104.479968</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>31</th>\n",
              "      <td>20</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 20, 'regress__regresso...</td>\n",
              "      <td>0.580639</td>\n",
              "      <td>31</td>\n",
              "      <td>-1.219894e+06</td>\n",
              "      <td>32</td>\n",
              "      <td>1104.487957</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>32</th>\n",
              "      <td>22</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 22, 'regress__regresso...</td>\n",
              "      <td>0.580613</td>\n",
              "      <td>33</td>\n",
              "      <td>-1.219974e+06</td>\n",
              "      <td>33</td>\n",
              "      <td>1104.524201</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>33</th>\n",
              "      <td>21</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 21, 'regress__regresso...</td>\n",
              "      <td>0.580588</td>\n",
              "      <td>34</td>\n",
              "      <td>-1.220049e+06</td>\n",
              "      <td>34</td>\n",
              "      <td>1104.558269</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>34</th>\n",
              "      <td>13</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 13, 'regress__regresso...</td>\n",
              "      <td>0.580575</td>\n",
              "      <td>35</td>\n",
              "      <td>-1.220090e+06</td>\n",
              "      <td>35</td>\n",
              "      <td>1104.576722</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>35</th>\n",
              "      <td>23</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 23, 'regress__regresso...</td>\n",
              "      <td>0.580547</td>\n",
              "      <td>36</td>\n",
              "      <td>-1.220158e+06</td>\n",
              "      <td>36</td>\n",
              "      <td>1104.607567</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>36</th>\n",
              "      <td>20</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 20, 'regress__regresso...</td>\n",
              "      <td>0.580504</td>\n",
              "      <td>37</td>\n",
              "      <td>-1.220300e+06</td>\n",
              "      <td>37</td>\n",
              "      <td>1104.671882</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>37</th>\n",
              "      <td>13</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 13, 'regress__regresso...</td>\n",
              "      <td>0.580497</td>\n",
              "      <td>38</td>\n",
              "      <td>-1.220314e+06</td>\n",
              "      <td>38</td>\n",
              "      <td>1104.678118</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>38</th>\n",
              "      <td>23</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 23, 'regress__regresso...</td>\n",
              "      <td>0.580492</td>\n",
              "      <td>39</td>\n",
              "      <td>-1.220321e+06</td>\n",
              "      <td>39</td>\n",
              "      <td>1104.681400</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>39</th>\n",
              "      <td>19</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 19, 'regress__regresso...</td>\n",
              "      <td>0.580446</td>\n",
              "      <td>40</td>\n",
              "      <td>-1.220472e+06</td>\n",
              "      <td>40</td>\n",
              "      <td>1104.749859</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>40</th>\n",
              "      <td>24</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 24, 'regress__regresso...</td>\n",
              "      <td>0.580304</td>\n",
              "      <td>41</td>\n",
              "      <td>-1.220868e+06</td>\n",
              "      <td>41</td>\n",
              "      <td>1104.928963</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>41</th>\n",
              "      <td>24</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 24, 'regress__regresso...</td>\n",
              "      <td>0.580304</td>\n",
              "      <td>42</td>\n",
              "      <td>-1.220868e+06</td>\n",
              "      <td>42</td>\n",
              "      <td>1104.928963</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>42</th>\n",
              "      <td>25</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 25, 'regress__regresso...</td>\n",
              "      <td>0.580211</td>\n",
              "      <td>43</td>\n",
              "      <td>-1.221132e+06</td>\n",
              "      <td>43</td>\n",
              "      <td>1105.048304</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>43</th>\n",
              "      <td>25</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 25, 'regress__regresso...</td>\n",
              "      <td>0.580211</td>\n",
              "      <td>44</td>\n",
              "      <td>-1.221132e+06</td>\n",
              "      <td>44</td>\n",
              "      <td>1105.048304</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>44</th>\n",
              "      <td>24</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 24, 'regress__regresso...</td>\n",
              "      <td>0.580210</td>\n",
              "      <td>45</td>\n",
              "      <td>-1.221143e+06</td>\n",
              "      <td>45</td>\n",
              "      <td>1105.053310</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>45</th>\n",
              "      <td>24</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 24, 'regress__regresso...</td>\n",
              "      <td>0.580148</td>\n",
              "      <td>46</td>\n",
              "      <td>-1.221328e+06</td>\n",
              "      <td>46</td>\n",
              "      <td>1105.136900</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>46</th>\n",
              "      <td>26</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 26, 'regress__regresso...</td>\n",
              "      <td>0.580132</td>\n",
              "      <td>47</td>\n",
              "      <td>-1.221376e+06</td>\n",
              "      <td>47</td>\n",
              "      <td>1105.158993</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>47</th>\n",
              "      <td>26</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 26, 'regress__regresso...</td>\n",
              "      <td>0.580132</td>\n",
              "      <td>48</td>\n",
              "      <td>-1.221376e+06</td>\n",
              "      <td>48</td>\n",
              "      <td>1105.158993</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>48</th>\n",
              "      <td>25</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 25, 'regress__regresso...</td>\n",
              "      <td>0.580118</td>\n",
              "      <td>49</td>\n",
              "      <td>-1.221403e+06</td>\n",
              "      <td>49</td>\n",
              "      <td>1105.171248</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>49</th>\n",
              "      <td>25</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 25, 'regress__regresso...</td>\n",
              "      <td>0.580069</td>\n",
              "      <td>50</td>\n",
              "      <td>-1.221551e+06</td>\n",
              "      <td>50</td>\n",
              "      <td>1105.238172</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   num_features_in_model        model_type  ... rank_test_MSE         RMSE\n",
              "0                     16  LinearRegression  ...             1  1103.812926\n",
              "1                     17  LinearRegression  ...             2  1103.834566\n",
              "2                     17  LinearRegression  ...             3  1103.834566\n",
              "3                     18  LinearRegression  ...             4  1103.898306\n",
              "4                     18  LinearRegression  ...             5  1103.898306\n",
              "5                     16             Ridge  ...             6  1103.947666\n",
              "6                     17             Ridge  ...             7  1103.971297\n",
              "7                     18             Ridge  ...             8  1104.035055\n",
              "8                     15  LinearRegression  ...             9  1104.089725\n",
              "9                     15             Ridge  ...            10  1104.216623\n",
              "10                    14  LinearRegression  ...            11  1104.236746\n",
              "11                    17             Ridge  ...            12  1104.252180\n",
              "12                    12  LinearRegression  ...            13  1104.279501\n",
              "13                    18             Ridge  ...            14  1104.285863\n",
              "14                    22  LinearRegression  ...            15  1104.311278\n",
              "15                    22  LinearRegression  ...            16  1104.311278\n",
              "16                    14             Ridge  ...            17  1104.339243\n",
              "17                    21  LinearRegression  ...            18  1104.342736\n",
              "18                    21  LinearRegression  ...            19  1104.342736\n",
              "19                    19  LinearRegression  ...            20  1104.345085\n",
              "20                    19  LinearRegression  ...            20  1104.345085\n",
              "21                    12             Ridge  ...            22  1104.351964\n",
              "22                    20  LinearRegression  ...            23  1104.360811\n",
              "23                    20  LinearRegression  ...            24  1104.360811\n",
              "24                    11  LinearRegression  ...            25  1104.376266\n",
              "25                    22             Ridge  ...            26  1104.441068\n",
              "26                    19             Ridge  ...            27  1104.468027\n",
              "27                    21             Ridge  ...            28  1104.470374\n",
              "28                    23  LinearRegression  ...            29  1104.478868\n",
              "29                    23  LinearRegression  ...            30  1104.478868\n",
              "30                    11             Ridge  ...            31  1104.479968\n",
              "31                    20             Ridge  ...            32  1104.487957\n",
              "32                    22             Ridge  ...            33  1104.524201\n",
              "33                    21             Ridge  ...            34  1104.558269\n",
              "34                    13  LinearRegression  ...            35  1104.576722\n",
              "35                    23             Ridge  ...            36  1104.607567\n",
              "36                    20             Ridge  ...            37  1104.671882\n",
              "37                    13             Ridge  ...            38  1104.678118\n",
              "38                    23             Ridge  ...            39  1104.681400\n",
              "39                    19             Ridge  ...            40  1104.749859\n",
              "40                    24  LinearRegression  ...            41  1104.928963\n",
              "41                    24  LinearRegression  ...            42  1104.928963\n",
              "42                    25  LinearRegression  ...            43  1105.048304\n",
              "43                    25  LinearRegression  ...            44  1105.048304\n",
              "44                    24             Ridge  ...            45  1105.053310\n",
              "45                    24             Ridge  ...            46  1105.136900\n",
              "46                    26  LinearRegression  ...            47  1105.158993\n",
              "47                    26  LinearRegression  ...            48  1105.158993\n",
              "48                    25             Ridge  ...            49  1105.171248\n",
              "49                    25             Ridge  ...            50  1105.238172\n",
              "\n",
              "[50 rows x 10 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 51
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HnNT5wg_ikoi"
      },
      "source": [
        "# Thank you Joey Gao!! \r\n",
        "# https://github.com/scikit-learn/scikit-learn/issues/12525\r\n",
        "\r\n",
        "def get_column_names_from_ColumnTransformer(column_transformer):    \r\n",
        "\r\n",
        "    col_name = []\r\n",
        "\r\n",
        "    for transformer_in_columns in column_transformer.transformers_:#the last transformer is ColumnTransformer's 'remainder' # <--- I modified this. In my case the last transformer was not the remainder. It was the ordinal encoder.\r\n",
        "\r\n",
        "        raw_col_name = transformer_in_columns[2]\r\n",
        "\r\n",
        "        if isinstance(transformer_in_columns[1],Pipeline): \r\n",
        "            transformer = transformer_in_columns[1].steps[-1][1]\r\n",
        "        else:\r\n",
        "            transformer = transformer_in_columns[1]\r\n",
        "        try:\r\n",
        "            names = transformer.get_feature_names()\r\n",
        "        except AttributeError: # if no 'get_feature_names' function, use raw column name\r\n",
        "            names = raw_col_name\r\n",
        "\r\n",
        "        if isinstance(names,np.ndarray): # eg.\r\n",
        "            col_name += names.tolist()\r\n",
        "        elif isinstance(names,list):\r\n",
        "            col_name += names    \r\n",
        "        elif isinstance(names,str):\r\n",
        "            col_name.append(names)\r\n",
        "            \r\n",
        "    return col_name"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "id": "zuXj1u7iZ4Dx",
        "outputId": "a9ae0ee4-d764-4d4a-c1db-0a8d23b5877c"
      },
      "source": [
        "# Get a boolean array that indicates which features SelectKBest chose when the best model was being fit.\r\n",
        "best_model_features_selected = gs_results.best_estimator_.named_steps['feature_selector'].get_support()\r\n",
        "\r\n",
        "# Get the column transformer instance used when creating the best model. \r\n",
        "best_model_ColumnTransformer =  gs_results.best_estimator_.named_steps['preprocessing']\r\n",
        "\r\n",
        "# Get the best models column names from the column transformer. \r\n",
        "best_model_column_names = get_column_names_from_ColumnTransformer(best_model_ColumnTransformer)\r\n",
        "\r\n",
        "# Convert the best models column names to a np.array so we can index the list using the boolean matrix\r\n",
        "# that indicates which features were used when fitting the best model. \r\n",
        "best_model_column_names = np.array(best_model_column_names)\r\n",
        "\r\n",
        "# Get the names of the features used in the best model.\r\n",
        "best_model_feature_names = best_model_column_names[best_model_features_selected]\r\n",
        "\r\n",
        "# Convert the names of best features from a numpy array back to a python list.\r\n",
        "best_model_feature_names = list(best_model_feature_names)\r\n",
        "\r\n",
        "# Get the coefficients of the best model. \r\n",
        "best_model_coefs = gs_results.best_estimator_._final_estimator.regressor_.coef_\r\n",
        "\r\n",
        "# Convert the coefficients of the best model to a python list.\r\n",
        "best_model_coefs = list(best_model_coefs)\r\n",
        "\r\n",
        "# Build a string that represents the ridge regression equation. \r\n",
        "regression_equation = \"Sales =\" \r\n",
        "for feat, coef in enumerate(zip(best_model_feature_names, best_model_coefs)): \r\n",
        "  term = \" \" + str(coef[1])[:7] + \"*(\" + str(coef[0]) + \") + \"\r\n",
        "  regression_equation = regression_equation + term\r\n",
        "\r\n",
        "regression_equation = regression_equation[:-2]\r\n",
        "\r\n",
        "regression_equation\r\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'Sales = 0.01270*(x0_Fruits and Vegetables) +  0.08816*(x1_OUT017) +  0.56196*(x1_OUT018) +  -0.6845*(x1_OUT019) +  0.98404*(x1_OUT027) +  0.11462*(x1_OUT035) +  -0.2990*(x1_OUT049) +  -0.0383*(x2_Tier 2) +  -0.7043*(x2_Tier 3) +  0.94062*(x3_Supermarket Type1) +  0.56196*(x3_Supermarket Type2) +  0.98404*(x3_Supermarket Type3) +  0.01272*(x4_Food) +  0.35044*(Outlet_Size) +  -0.0573*(2) +  0.00885*(4) '"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 32
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vfZQwtRUKnJ4"
      },
      "source": [
        "\r\n",
        "nominal_categorical_feat = ['Item_Type', 'Outlet_Identifier', 'Item_Id_Binned']\r\n",
        "ordinal_categorical_feat = ['Item_Fat_Content', 'Outlet_Size', 'Outlet_Location_Type', 'Outlet_Type', 'Years_of_operation']\r\n",
        "numeric_feat = ['Item_Weight', 'Item_Visibility', 'Item_MRP']\r\n",
        "\r\n",
        "X = sales_df.loc[:, sales_df.columns != 'Item_Outlet_Sales']\r\n",
        "y = sales_df.loc[:, 'Item_Outlet_Sales']\r\n",
        "\r\n",
        "\r\n",
        "preprocess = ColumnTransformer(transformers=[(\"one_hot_encoder\", OneHotEncoder(drop=\"first\", sparse=False), nominal_categorical_feat),\r\n",
        "                                             (\"ordinal_encoder\", OrdinalEncoder(categories=[['Not_Edible', 'Low Fat', 'Regular'], ['Small', 'Medium', 'High'],\r\n",
        "                                                                                            ['Tier 1', 'Tier 2', 'Tier 3'], ['Grocery Store', 'Supermarket Type1', 'Supermarket Type2', 'Supermarket Type3'],\r\n",
        "                                                                                            [4, 6, 9, 11, 14, 15, 16, 26, 28]]), ordinal_categorical_feat)],\r\n",
        "                               remainder='passthrough')\r\n",
        "\r\n",
        "\r\n",
        "column_list = list(range(5,35))\r\n",
        "column_list = column_list + ['all']\r\n",
        "  \r\n",
        "\r\n",
        "reg_strategy = TransformedTargetRegressor()\r\n",
        "bcox_transformer = PowerTransformer(method='box-cox')\r\n",
        "\r\n",
        "model_pipeline = Pipeline([(\"preprocessing\", preprocess),\r\n",
        "                           ('feature_selector', SelectKBest(f_regression)),\r\n",
        "                           ('regress', reg_strategy)])\r\n",
        "\r\n",
        "parameter_grid = [{'feature_selector__k' : column_list,\r\n",
        "                   'regress__regressor' : [LinearRegression(), Lasso(), Ridge()],\r\n",
        "                   'regress__regressor__fit_intercept' : [False, True],\r\n",
        "                   'regress__transformer' : [None, bcox_transformer]}]\r\n",
        "\r\n",
        "\r\n",
        "score_types = {'r2' :'r2', 'MSE' : 'neg_mean_squared_error'}\r\n",
        "\r\n",
        "gs = GridSearchCV(estimator=model_pipeline, param_grid=parameter_grid, scoring=score_types, refit='MSE', cv=5, n_jobs=-1)\r\n",
        "\r\n",
        "gs.fit(X, y)\r\n",
        "\r\n",
        "PATH = '/content/drive/MyDrive/Programming/Colab Notebooks/Coding_Dojo/Sales_Project/Current/'\r\n",
        "gridsearch_result_filename = 'gridsearch_models_6_Linear.pkl'\r\n",
        "full_path = PATH + gridsearch_result_filename\r\n",
        "with open(full_path, 'wb') as file:\r\n",
        "  pickle.dump(gs, file)\r\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zoOJDUYqMcz5"
      },
      "source": [
        "full_path = '/content/drive/MyDrive/Programming/Colab Notebooks/Coding_Dojo/Sales_Project/Current/gridsearch_models_6_Linear.pkl'\r\n",
        "\r\n",
        "with open(full_path, 'rb') as file:\r\n",
        "  gs_results2 = pickle.load(file)\r\n",
        "\r\n",
        "search_results2 = gs_results2.cv_results_\r\n",
        "top_estimator2 = gs_results2.best_estimator_\r\n",
        "top_score2 = gs_results2.best_score_\r\n",
        "top_parameters2 = gs_results2.best_params_\r\n",
        "\r\n",
        "gs_result_df2 = pd.DataFrame(search_results2)\r\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "e_gfMs_8tJd-",
        "outputId": "3f00f818-1a39-4d57-eab9-1fabedb4a722"
      },
      "source": [
        "results_simplified_df2 = gs_result_df2.loc[:,:].sort_values(by=['rank_test_MSE'], ignore_index=True)\r\n",
        "\r\n",
        "results_simplified_df2.rename(columns= {'param_feature_selector__k' : 'num_features_in_model',\r\n",
        "                                       'param_regress__regressor' : 'model_type',\r\n",
        "                                       'param_regress__transformer' : 'target_transformation',\r\n",
        "                                       'param_regress__regressor__fit_intercept' : 'fit_intercept'}, inplace=True)\r\n",
        "\r\n",
        "results_simplified_df2 = results_simplified_df2.loc[:, ['num_features_in_model', 'model_type', 'target_transformation', 'fit_intercept',\r\n",
        "                                                      'params', 'mean_test_r2', 'rank_test_r2', 'mean_test_MSE', 'rank_test_MSE']]\r\n",
        "\r\n",
        "\r\n",
        "results_simplified_df2['target_transformation'] = results_simplified_df2['target_transformation'].fillna(value=\"No_Transformation\")\r\n",
        "results_simplified_df2.loc[ (results_simplified_df2['target_transformation'] != 'No_Transformation') , 'target_transformation'] = \"Box-Cox\"\r\n",
        "\r\n",
        "# For easier viewing, trim model_type down so it only shows the name of the model used.\r\n",
        "results_simplified_df2['model_type'] = results_simplified_df2['model_type'].astype(str)\r\n",
        "results_simplified_df2['model_type'] = results_simplified_df2['model_type'].map(lambda model_string : model_string.split('(')[0])\r\n",
        "\r\n",
        "# Negative MSE is not very interpretable. Take the absolute value and square root to get a more meaningful number. \r\n",
        "results_simplified_df2['RMSE'] = results_simplified_df2['mean_test_MSE'].abs().pow(1./2)\r\n",
        "\r\n",
        "pd.set_option('display.max_rows', 500)\r\n",
        "\r\n",
        "results_simplified_df2.head(50)"
      ],
      "execution_count": 52,
      "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>num_features_in_model</th>\n",
              "      <th>model_type</th>\n",
              "      <th>target_transformation</th>\n",
              "      <th>fit_intercept</th>\n",
              "      <th>params</th>\n",
              "      <th>mean_test_r2</th>\n",
              "      <th>rank_test_r2</th>\n",
              "      <th>mean_test_MSE</th>\n",
              "      <th>rank_test_MSE</th>\n",
              "      <th>RMSE</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>13</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 13, 'regress__regresso...</td>\n",
              "      <td>0.581160</td>\n",
              "      <td>1</td>\n",
              "      <td>-1.218403e+06</td>\n",
              "      <td>1</td>\n",
              "      <td>1103.812926</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>14</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 14, 'regress__regresso...</td>\n",
              "      <td>0.581146</td>\n",
              "      <td>2</td>\n",
              "      <td>-1.218451e+06</td>\n",
              "      <td>2</td>\n",
              "      <td>1103.834566</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>15</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 15, 'regress__regresso...</td>\n",
              "      <td>0.581100</td>\n",
              "      <td>3</td>\n",
              "      <td>-1.218591e+06</td>\n",
              "      <td>3</td>\n",
              "      <td>1103.898306</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>13</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 13, 'regress__regresso...</td>\n",
              "      <td>0.580981</td>\n",
              "      <td>4</td>\n",
              "      <td>-1.218937e+06</td>\n",
              "      <td>4</td>\n",
              "      <td>1104.054906</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>12</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 12, 'regress__regresso...</td>\n",
              "      <td>0.580951</td>\n",
              "      <td>6</td>\n",
              "      <td>-1.218995e+06</td>\n",
              "      <td>5</td>\n",
              "      <td>1104.081241</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>14</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 14, 'regress__regresso...</td>\n",
              "      <td>0.580955</td>\n",
              "      <td>5</td>\n",
              "      <td>-1.219020e+06</td>\n",
              "      <td>6</td>\n",
              "      <td>1104.092393</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>15</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 15, 'regress__regresso...</td>\n",
              "      <td>0.580912</td>\n",
              "      <td>7</td>\n",
              "      <td>-1.219155e+06</td>\n",
              "      <td>7</td>\n",
              "      <td>1104.153533</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>11</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 11, 'regress__regresso...</td>\n",
              "      <td>0.580829</td>\n",
              "      <td>8</td>\n",
              "      <td>-1.219321e+06</td>\n",
              "      <td>8</td>\n",
              "      <td>1104.228762</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>12</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 12, 'regress__regresso...</td>\n",
              "      <td>0.580801</td>\n",
              "      <td>9</td>\n",
              "      <td>-1.219451e+06</td>\n",
              "      <td>9</td>\n",
              "      <td>1104.287437</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>19</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 19, 'regress__regresso...</td>\n",
              "      <td>0.580772</td>\n",
              "      <td>10</td>\n",
              "      <td>-1.219503e+06</td>\n",
              "      <td>10</td>\n",
              "      <td>1104.311278</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10</th>\n",
              "      <td>19</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 19, 'regress__regresso...</td>\n",
              "      <td>0.580772</td>\n",
              "      <td>11</td>\n",
              "      <td>-1.219503e+06</td>\n",
              "      <td>11</td>\n",
              "      <td>1104.311278</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>11</th>\n",
              "      <td>18</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 18, 'regress__regresso...</td>\n",
              "      <td>0.580750</td>\n",
              "      <td>12</td>\n",
              "      <td>-1.219573e+06</td>\n",
              "      <td>12</td>\n",
              "      <td>1104.342736</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>12</th>\n",
              "      <td>18</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 18, 'regress__regresso...</td>\n",
              "      <td>0.580750</td>\n",
              "      <td>13</td>\n",
              "      <td>-1.219573e+06</td>\n",
              "      <td>13</td>\n",
              "      <td>1104.342736</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13</th>\n",
              "      <td>16</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 16, 'regress__regresso...</td>\n",
              "      <td>0.580747</td>\n",
              "      <td>14</td>\n",
              "      <td>-1.219578e+06</td>\n",
              "      <td>14</td>\n",
              "      <td>1104.345085</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>14</th>\n",
              "      <td>17</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 17, 'regress__regresso...</td>\n",
              "      <td>0.580734</td>\n",
              "      <td>15</td>\n",
              "      <td>-1.219613e+06</td>\n",
              "      <td>15</td>\n",
              "      <td>1104.360811</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>15</th>\n",
              "      <td>17</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 17, 'regress__regresso...</td>\n",
              "      <td>0.580734</td>\n",
              "      <td>16</td>\n",
              "      <td>-1.219613e+06</td>\n",
              "      <td>16</td>\n",
              "      <td>1104.360811</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16</th>\n",
              "      <td>19</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 19, 'regress__regresso...</td>\n",
              "      <td>0.580703</td>\n",
              "      <td>17</td>\n",
              "      <td>-1.219708e+06</td>\n",
              "      <td>17</td>\n",
              "      <td>1104.403766</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>17</th>\n",
              "      <td>18</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 18, 'regress__regresso...</td>\n",
              "      <td>0.580682</td>\n",
              "      <td>18</td>\n",
              "      <td>-1.219774e+06</td>\n",
              "      <td>18</td>\n",
              "      <td>1104.433598</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>18</th>\n",
              "      <td>17</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 17, 'regress__regresso...</td>\n",
              "      <td>0.580666</td>\n",
              "      <td>19</td>\n",
              "      <td>-1.219816e+06</td>\n",
              "      <td>19</td>\n",
              "      <td>1104.452859</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>19</th>\n",
              "      <td>20</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 20, 'regress__regresso...</td>\n",
              "      <td>0.580643</td>\n",
              "      <td>21</td>\n",
              "      <td>-1.219874e+06</td>\n",
              "      <td>20</td>\n",
              "      <td>1104.478868</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>20</th>\n",
              "      <td>20</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 20, 'regress__regresso...</td>\n",
              "      <td>0.580643</td>\n",
              "      <td>22</td>\n",
              "      <td>-1.219874e+06</td>\n",
              "      <td>21</td>\n",
              "      <td>1104.478868</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21</th>\n",
              "      <td>19</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 19, 'regress__regresso...</td>\n",
              "      <td>0.580644</td>\n",
              "      <td>20</td>\n",
              "      <td>-1.219885e+06</td>\n",
              "      <td>22</td>\n",
              "      <td>1104.483971</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>22</th>\n",
              "      <td>18</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 18, 'regress__regresso...</td>\n",
              "      <td>0.580626</td>\n",
              "      <td>23</td>\n",
              "      <td>-1.219940e+06</td>\n",
              "      <td>23</td>\n",
              "      <td>1104.509144</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>23</th>\n",
              "      <td>20</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 20, 'regress__regresso...</td>\n",
              "      <td>0.580575</td>\n",
              "      <td>24</td>\n",
              "      <td>-1.220075e+06</td>\n",
              "      <td>24</td>\n",
              "      <td>1104.570177</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>24</th>\n",
              "      <td>17</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 17, 'regress__regresso...</td>\n",
              "      <td>0.580558</td>\n",
              "      <td>25</td>\n",
              "      <td>-1.220142e+06</td>\n",
              "      <td>25</td>\n",
              "      <td>1104.600197</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>25</th>\n",
              "      <td>16</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 16, 'regress__regresso...</td>\n",
              "      <td>0.580555</td>\n",
              "      <td>26</td>\n",
              "      <td>-1.220153e+06</td>\n",
              "      <td>26</td>\n",
              "      <td>1104.605507</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>26</th>\n",
              "      <td>11</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 11, 'regress__regresso...</td>\n",
              "      <td>0.580542</td>\n",
              "      <td>27</td>\n",
              "      <td>-1.220183e+06</td>\n",
              "      <td>27</td>\n",
              "      <td>1104.618990</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>27</th>\n",
              "      <td>20</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 20, 'regress__regresso...</td>\n",
              "      <td>0.580525</td>\n",
              "      <td>28</td>\n",
              "      <td>-1.220225e+06</td>\n",
              "      <td>28</td>\n",
              "      <td>1104.637842</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>28</th>\n",
              "      <td>21</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 21, 'regress__regresso...</td>\n",
              "      <td>0.580304</td>\n",
              "      <td>29</td>\n",
              "      <td>-1.220868e+06</td>\n",
              "      <td>29</td>\n",
              "      <td>1104.928963</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>29</th>\n",
              "      <td>21</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 21, 'regress__regresso...</td>\n",
              "      <td>0.580304</td>\n",
              "      <td>30</td>\n",
              "      <td>-1.220868e+06</td>\n",
              "      <td>30</td>\n",
              "      <td>1104.928963</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>30</th>\n",
              "      <td>21</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 21, 'regress__regresso...</td>\n",
              "      <td>0.580238</td>\n",
              "      <td>31</td>\n",
              "      <td>-1.221062e+06</td>\n",
              "      <td>31</td>\n",
              "      <td>1105.016782</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>31</th>\n",
              "      <td>22</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 22, 'regress__regresso...</td>\n",
              "      <td>0.580211</td>\n",
              "      <td>32</td>\n",
              "      <td>-1.221132e+06</td>\n",
              "      <td>32</td>\n",
              "      <td>1105.048304</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>32</th>\n",
              "      <td>22</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 22, 'regress__regresso...</td>\n",
              "      <td>0.580211</td>\n",
              "      <td>33</td>\n",
              "      <td>-1.221132e+06</td>\n",
              "      <td>33</td>\n",
              "      <td>1105.048304</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>33</th>\n",
              "      <td>21</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 21, 'regress__regresso...</td>\n",
              "      <td>0.580177</td>\n",
              "      <td>34</td>\n",
              "      <td>-1.221246e+06</td>\n",
              "      <td>34</td>\n",
              "      <td>1105.099961</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>34</th>\n",
              "      <td>22</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 22, 'regress__regresso...</td>\n",
              "      <td>0.580146</td>\n",
              "      <td>35</td>\n",
              "      <td>-1.221322e+06</td>\n",
              "      <td>35</td>\n",
              "      <td>1105.134450</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>35</th>\n",
              "      <td>23</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 23, 'regress__regresso...</td>\n",
              "      <td>0.580132</td>\n",
              "      <td>36</td>\n",
              "      <td>-1.221376e+06</td>\n",
              "      <td>36</td>\n",
              "      <td>1105.158993</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>36</th>\n",
              "      <td>23</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 23, 'regress__regresso...</td>\n",
              "      <td>0.580132</td>\n",
              "      <td>37</td>\n",
              "      <td>-1.221376e+06</td>\n",
              "      <td>37</td>\n",
              "      <td>1105.158993</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>37</th>\n",
              "      <td>22</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 22, 'regress__regresso...</td>\n",
              "      <td>0.580102</td>\n",
              "      <td>38</td>\n",
              "      <td>-1.221458e+06</td>\n",
              "      <td>38</td>\n",
              "      <td>1105.195900</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>38</th>\n",
              "      <td>23</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 23, 'regress__regresso...</td>\n",
              "      <td>0.580067</td>\n",
              "      <td>39</td>\n",
              "      <td>-1.221570e+06</td>\n",
              "      <td>39</td>\n",
              "      <td>1105.246399</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>39</th>\n",
              "      <td>23</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 23, 'regress__regresso...</td>\n",
              "      <td>0.580011</td>\n",
              "      <td>40</td>\n",
              "      <td>-1.221739e+06</td>\n",
              "      <td>40</td>\n",
              "      <td>1105.322822</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>40</th>\n",
              "      <td>26</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 26, 'regress__regresso...</td>\n",
              "      <td>0.579977</td>\n",
              "      <td>41</td>\n",
              "      <td>-1.221836e+06</td>\n",
              "      <td>41</td>\n",
              "      <td>1105.366730</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>41</th>\n",
              "      <td>26</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 26, 'regress__regresso...</td>\n",
              "      <td>0.579977</td>\n",
              "      <td>42</td>\n",
              "      <td>-1.221836e+06</td>\n",
              "      <td>42</td>\n",
              "      <td>1105.366730</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>42</th>\n",
              "      <td>26</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 26, 'regress__regresso...</td>\n",
              "      <td>0.579923</td>\n",
              "      <td>43</td>\n",
              "      <td>-1.221995e+06</td>\n",
              "      <td>43</td>\n",
              "      <td>1105.438643</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>43</th>\n",
              "      <td>25</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 25, 'regress__regresso...</td>\n",
              "      <td>0.579903</td>\n",
              "      <td>44</td>\n",
              "      <td>-1.222062e+06</td>\n",
              "      <td>44</td>\n",
              "      <td>1105.469122</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>44</th>\n",
              "      <td>25</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 25, 'regress__regresso...</td>\n",
              "      <td>0.579903</td>\n",
              "      <td>45</td>\n",
              "      <td>-1.222062e+06</td>\n",
              "      <td>45</td>\n",
              "      <td>1105.469122</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>45</th>\n",
              "      <td>28</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 28, 'regress__regresso...</td>\n",
              "      <td>0.579895</td>\n",
              "      <td>46</td>\n",
              "      <td>-1.222140e+06</td>\n",
              "      <td>46</td>\n",
              "      <td>1105.504229</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>46</th>\n",
              "      <td>28</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 28, 'regress__regresso...</td>\n",
              "      <td>0.579895</td>\n",
              "      <td>47</td>\n",
              "      <td>-1.222140e+06</td>\n",
              "      <td>47</td>\n",
              "      <td>1105.504229</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>47</th>\n",
              "      <td>24</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>False</td>\n",
              "      <td>{'feature_selector__k': 24, 'regress__regresso...</td>\n",
              "      <td>0.579866</td>\n",
              "      <td>50</td>\n",
              "      <td>-1.222168e+06</td>\n",
              "      <td>48</td>\n",
              "      <td>1105.517130</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>48</th>\n",
              "      <td>24</td>\n",
              "      <td>LinearRegression</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 24, 'regress__regresso...</td>\n",
              "      <td>0.579866</td>\n",
              "      <td>51</td>\n",
              "      <td>-1.222168e+06</td>\n",
              "      <td>49</td>\n",
              "      <td>1105.517130</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>49</th>\n",
              "      <td>25</td>\n",
              "      <td>Ridge</td>\n",
              "      <td>Box-Cox</td>\n",
              "      <td>True</td>\n",
              "      <td>{'feature_selector__k': 25, 'regress__regresso...</td>\n",
              "      <td>0.579850</td>\n",
              "      <td>54</td>\n",
              "      <td>-1.222217e+06</td>\n",
              "      <td>50</td>\n",
              "      <td>1105.539260</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   num_features_in_model        model_type  ... rank_test_MSE         RMSE\n",
              "0                     13  LinearRegression  ...             1  1103.812926\n",
              "1                     14  LinearRegression  ...             2  1103.834566\n",
              "2                     15  LinearRegression  ...             3  1103.898306\n",
              "3                     13             Ridge  ...             4  1104.054906\n",
              "4                     12  LinearRegression  ...             5  1104.081241\n",
              "5                     14             Ridge  ...             6  1104.092393\n",
              "6                     15             Ridge  ...             7  1104.153533\n",
              "7                     11  LinearRegression  ...             8  1104.228762\n",
              "8                     12             Ridge  ...             9  1104.287437\n",
              "9                     19  LinearRegression  ...            10  1104.311278\n",
              "10                    19  LinearRegression  ...            11  1104.311278\n",
              "11                    18  LinearRegression  ...            12  1104.342736\n",
              "12                    18  LinearRegression  ...            13  1104.342736\n",
              "13                    16  LinearRegression  ...            14  1104.345085\n",
              "14                    17  LinearRegression  ...            15  1104.360811\n",
              "15                    17  LinearRegression  ...            16  1104.360811\n",
              "16                    19             Ridge  ...            17  1104.403766\n",
              "17                    18             Ridge  ...            18  1104.433598\n",
              "18                    17             Ridge  ...            19  1104.452859\n",
              "19                    20  LinearRegression  ...            20  1104.478868\n",
              "20                    20  LinearRegression  ...            21  1104.478868\n",
              "21                    19             Ridge  ...            22  1104.483971\n",
              "22                    18             Ridge  ...            23  1104.509144\n",
              "23                    20             Ridge  ...            24  1104.570177\n",
              "24                    17             Ridge  ...            25  1104.600197\n",
              "25                    16             Ridge  ...            26  1104.605507\n",
              "26                    11             Ridge  ...            27  1104.618990\n",
              "27                    20             Ridge  ...            28  1104.637842\n",
              "28                    21  LinearRegression  ...            29  1104.928963\n",
              "29                    21  LinearRegression  ...            30  1104.928963\n",
              "30                    21             Ridge  ...            31  1105.016782\n",
              "31                    22  LinearRegression  ...            32  1105.048304\n",
              "32                    22  LinearRegression  ...            33  1105.048304\n",
              "33                    21             Ridge  ...            34  1105.099961\n",
              "34                    22             Ridge  ...            35  1105.134450\n",
              "35                    23  LinearRegression  ...            36  1105.158993\n",
              "36                    23  LinearRegression  ...            37  1105.158993\n",
              "37                    22             Ridge  ...            38  1105.195900\n",
              "38                    23             Ridge  ...            39  1105.246399\n",
              "39                    23             Ridge  ...            40  1105.322822\n",
              "40                    26  LinearRegression  ...            41  1105.366730\n",
              "41                    26  LinearRegression  ...            42  1105.366730\n",
              "42                    26             Ridge  ...            43  1105.438643\n",
              "43                    25  LinearRegression  ...            44  1105.469122\n",
              "44                    25  LinearRegression  ...            45  1105.469122\n",
              "45                    28  LinearRegression  ...            46  1105.504229\n",
              "46                    28  LinearRegression  ...            47  1105.504229\n",
              "47                    24  LinearRegression  ...            48  1105.517130\n",
              "48                    24  LinearRegression  ...            49  1105.517130\n",
              "49                    25             Ridge  ...            50  1105.539260\n",
              "\n",
              "[50 rows x 10 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 52
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "id": "0BpBZRW2uiES",
        "outputId": "0af49040-d25e-4189-b649-8645a9103d86"
      },
      "source": [
        "# Get a boolean array that indicates which features SelectKBest chose when the best model was being fit.\r\n",
        "best_model_features_selected_alt = gs_results2.best_estimator_.named_steps['feature_selector'].get_support()\r\n",
        "\r\n",
        "# Get the column transformer instance used when creating the best model. \r\n",
        "best_model_ColumnTransformer_alt =  gs_results2.best_estimator_.named_steps['preprocessing']\r\n",
        "\r\n",
        "# Get the best models column names from the column transformer. \r\n",
        "best_model_column_names_alt = get_column_names_from_ColumnTransformer(best_model_ColumnTransformer_alt)\r\n",
        "\r\n",
        "# Convert the best models column names to a np.array so we can index the list using the boolean matrix\r\n",
        "# that indicates which features were used when fitting the best model. \r\n",
        "best_model_column_names_alt = np.array(best_model_column_names_alt)\r\n",
        "\r\n",
        "# Get the names of the features used in the best model.\r\n",
        "best_model_feature_names_alt = best_model_column_names_alt[best_model_features_selected_alt]\r\n",
        "\r\n",
        "# Convert the names of best features from a numpy array back to a python list.\r\n",
        "best_model_feature_names_alt = list(best_model_feature_names_alt)\r\n",
        "\r\n",
        "# Get the coefficients of the best model. \r\n",
        "best_model_coefs_alt = gs_results2.best_estimator_._final_estimator.regressor_.coef_\r\n",
        "\r\n",
        "# Convert the coefficients of the best model to a python list.\r\n",
        "best_model_coefs_alt = list(best_model_coefs_alt)\r\n",
        "\r\n",
        "# Build a string that represents the ridge regression equation. \r\n",
        "regression_equation_alt_encoding = \"Sales = \" \r\n",
        "for feat, coef in enumerate(zip(best_model_feature_names_alt, best_model_coefs_alt)): \r\n",
        "  term = str(coef[1])[:7] + \"*(\" + str(coef[0]) + \") + \"\r\n",
        "  regression_equation_alt_encoding = regression_equation_alt_encoding + term\r\n",
        "\r\n",
        "regression_equation_alt_encoding = regression_equation_alt_encoding[:-2]\r\n",
        "\r\n",
        "regression_equation_alt_encoding"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'Sales = 0.01270*(x0_Fruits and Vegetables) + 0.08816*(x1_OUT017) + -1.6986*(x1_OUT018) + -0.0570*(x1_OUT019) + -2.4226*(x1_OUT027) + 0.11462*(x1_OUT035) + 0.01475*(x1_OUT049) + 0.01272*(x2_Food) + 0.03666*(Outlet_Size) + -0.0383*(Outlet_Location_Type) + 1.56817*(Outlet_Type) + -0.0573*(2) + 0.00885*(4) '"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 49
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "f89U9Y4z9DfC"
      },
      "source": [
        "## Equation 1\r\n",
        "\r\n",
        "Calculated using the following encoding: \r\n",
        "\r\n",
        "Nominal categorical features:  ['Item_Type', 'Outlet_Identifier', 'Outlet_Location_Type', 'Outlet_Type', 'Item_Id_Binned']\r\n",
        "\r\n",
        "Ordinal categorical features: ['Item_Fat_Content', 'Outlet_Size', 'Years_of_operation']\r\n",
        "\r\n",
        "numeric features:  ['Item_Weight', 'Item_Visibility', 'Item_MRP']"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "id": "5gSW4Zup5j3F",
        "outputId": "9c2748bd-e457-46e3-e8ac-fd570b64d018"
      },
      "source": [
        "regression_equation"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'Sales = 0.01270*(x0_Fruits and Vegetables) +  0.08816*(x1_OUT017) +  0.56196*(x1_OUT018) +  -0.6845*(x1_OUT019) +  0.98404*(x1_OUT027) +  0.11462*(x1_OUT035) +  -0.2990*(x1_OUT049) +  -0.0383*(x2_Tier 2) +  -0.7043*(x2_Tier 3) +  0.94062*(x3_Supermarket Type1) +  0.56196*(x3_Supermarket Type2) +  0.98404*(x3_Supermarket Type3) +  0.01272*(x4_Food) +  0.35044*(Outlet_Size) +  -0.0573*(2) +  0.00885*(4) '"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 50
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "w0BiK3-G9gRm"
      },
      "source": [
        "## Equation 2\r\n",
        "\r\n",
        "Calculated using the following encoding: \r\n",
        "\r\n",
        "Nominal categorical features:  ['Item_Type', 'Outlet_Identifier', 'Item_Id_Binned']\r\n",
        "\r\n",
        "Ordinal categorical features: ['Item_Fat_Content', 'Outlet_Size', 'Outlet_Location_Type', 'Outlet_Type', 'Years_of_operation']\r\n",
        "\r\n",
        "numeric features:  ['Item_Weight', 'Item_Visibility', 'Item_MRP']\r\n",
        "\r\n",
        "\r\n",
        "Note: the only difference between this set of models and the one above is that 'Outlet type' and 'Outlet location type' moved from being nominal categorical features to ordinal categorical features. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        },
        "id": "_bsTN5ab-ETu",
        "outputId": "0d02658e-adcd-484b-c03a-67653d2e7c86"
      },
      "source": [
        "regression_equation_alt_encoding"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'Sales = 0.01270*(x0_Fruits and Vegetables) + 0.08816*(x1_OUT017) + -1.6986*(x1_OUT018) + -0.0570*(x1_OUT019) + -2.4226*(x1_OUT027) + 0.11462*(x1_OUT035) + 0.01475*(x1_OUT049) + 0.01272*(x2_Food) + 0.03666*(Outlet_Size) + -0.0383*(Outlet_Location_Type) + 1.56817*(Outlet_Type) + -0.0573*(2) + 0.00885*(4) '"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 38
        }
      ]
    }
  ]
}