02-regression-universal-model.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "c772a64f-823c-4973-bf41-3c9a239c601a",
   "metadata": {},
   "source": [
    "# Universal Regression Model"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6cc25fc3-e84a-4d87-ace3-e947218d58aa",
   "metadata": {},
   "source": [
    "### Importing libraries and define functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "5e31b3ce-df35-4b10-a99a-2f753f523fb9",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import scipy\n",
    "import math\n",
    "import os\n",
    "import sys"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "8375cb30",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Managing Warnings \n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "\n",
    "if not sys.warnoptions:\n",
    "    warnings.simplefilter(\"ignore\")\n",
    "    os.environ[\"PYTHONWARNINGS\"] = \"ignore\" # Also affect subprocesses"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "2d2c745b-9e8f-415c-a43c-f08fb5937b46",
   "metadata": {},
   "outputs": [],
   "source": [
    "import sklearn\n",
    "from sklearn.decomposition import PCA\n",
    "from sklearn import preprocessing\n",
    "from sklearn.impute import SimpleImputer\n",
    "from sklearn.model_selection import KFold\n",
    "from sklearn.model_selection import cross_val_score\n",
    "from sklearn.model_selection import cross_validate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "5b9378dc",
   "metadata": {},
   "outputs": [],
   "source": [
    "from yellowbrick.regressor import ResidualsPlot\n",
    "from yellowbrick.regressor import PredictionError\n",
    "from yellowbrick.model_selection import FeatureImportances\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "03a6e29f",
   "metadata": {},
   "outputs": [],
   "source": [
    "from pycaret.regression import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "8d5a3c8a",
   "metadata": {},
   "outputs": [],
   "source": [
    "import functions as f"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "d563849c",
   "metadata": {},
   "outputs": [],
   "source": [
    "from copy import deepcopy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "e1409fe1",
   "metadata": {},
   "outputs": [],
   "source": [
    "categorical_columns = ['NP_type', 'Coating','shape','Cell_name','Cell_species','cell_Organ','Cell_morphology',\n",
    "                       'Cell_age','cell_type','sex','Assay','Test_indicator']\n",
    "numerical_columns = ['core_size_nm','hydro_size_nm','Surf_charge_mV','Surface_area_m2_g','Dose_microg_mL','Duration_h',\n",
    "                    'Viability']\n",
    "\n",
    "numerical_columns_no_target = ['core_size_nm','hydro_size_nm','Surf_charge_mV','Surface_area_m2_g','Dose_microg_mL','Duration_h']"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bf6d7f7e",
   "metadata": {},
   "source": [
    "### Loading the train data into the data frame."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "da49b7e9",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_train_df = pd.read_csv(\"output/datasets/universal/data_regression_train_df.csv\", encoding=\"utf-8\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "f38a5c13",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "NP_type                object\n",
       "Coating                object\n",
       "core_size_nm          float64\n",
       "hydro_size_nm         float64\n",
       "Surf_charge_mV        float64\n",
       "Surface_area_m2_g     float64\n",
       "shape                  object\n",
       "Dose_microg_mL        float64\n",
       "Duration_h            float64\n",
       "Cell_name              object\n",
       "Cell_species           object\n",
       "cell_Organ             object\n",
       "Cell_morphology        object\n",
       "Cell_age               object\n",
       "cell_type              object\n",
       "sex                    object\n",
       "Assay                  object\n",
       "Test_indicator         object\n",
       "nanomaterial_group     object\n",
       "Viability             float64\n",
       "dtype: object"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_train_df.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "2d093a1e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(3724, 20)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_train_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "ebeeb573",
   "metadata": {},
   "outputs": [
    {
     "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>NP_type</th>\n",
       "      <th>Coating</th>\n",
       "      <th>core_size_nm</th>\n",
       "      <th>hydro_size_nm</th>\n",
       "      <th>Surf_charge_mV</th>\n",
       "      <th>Surface_area_m2_g</th>\n",
       "      <th>shape</th>\n",
       "      <th>Dose_microg_mL</th>\n",
       "      <th>Duration_h</th>\n",
       "      <th>Cell_name</th>\n",
       "      <th>Cell_species</th>\n",
       "      <th>cell_Organ</th>\n",
       "      <th>Cell_morphology</th>\n",
       "      <th>Cell_age</th>\n",
       "      <th>cell_type</th>\n",
       "      <th>sex</th>\n",
       "      <th>Assay</th>\n",
       "      <th>Test_indicator</th>\n",
       "      <th>nanomaterial_group</th>\n",
       "      <th>Viability</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>CuO</td>\n",
       "      <td>None</td>\n",
       "      <td>12.80</td>\n",
       "      <td>305.3</td>\n",
       "      <td>7.60</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>6.3</td>\n",
       "      <td>NaN</td>\n",
       "      <td>BEAS_2B</td>\n",
       "      <td>Human</td>\n",
       "      <td>Lung</td>\n",
       "      <td>Epithelial</td>\n",
       "      <td>Adult</td>\n",
       "      <td>Normal</td>\n",
       "      <td>NaN</td>\n",
       "      <td>MTS</td>\n",
       "      <td>TetrazoliumSalt</td>\n",
       "      <td>meta_oxide</td>\n",
       "      <td>82.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>PLGA</td>\n",
       "      <td>PVA</td>\n",
       "      <td>182.00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-3.50</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Sphere</td>\n",
       "      <td>500.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>Calu-3</td>\n",
       "      <td>Human</td>\n",
       "      <td>Lung</td>\n",
       "      <td>Epithelial</td>\n",
       "      <td>Adult</td>\n",
       "      <td>cancer</td>\n",
       "      <td>male</td>\n",
       "      <td>MTT</td>\n",
       "      <td>TetrazoliumSalt</td>\n",
       "      <td>other</td>\n",
       "      <td>94.29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>CuO</td>\n",
       "      <td>None</td>\n",
       "      <td>46.30</td>\n",
       "      <td>239.0</td>\n",
       "      <td>42.80</td>\n",
       "      <td>24.1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>HCMEC</td>\n",
       "      <td>Human</td>\n",
       "      <td>Heart</td>\n",
       "      <td>Endothelial</td>\n",
       "      <td>Adult</td>\n",
       "      <td>Normal</td>\n",
       "      <td>NaN</td>\n",
       "      <td>MTT</td>\n",
       "      <td>TetrazoliumSalt</td>\n",
       "      <td>meta_oxide</td>\n",
       "      <td>96.13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Hydroxyapatite</td>\n",
       "      <td>None</td>\n",
       "      <td>20.00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-5.30</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Rod</td>\n",
       "      <td>10.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>RAW 264.7</td>\n",
       "      <td>Mouse</td>\n",
       "      <td>Blood</td>\n",
       "      <td>Macrophage</td>\n",
       "      <td>Adult</td>\n",
       "      <td>cancer</td>\n",
       "      <td>male</td>\n",
       "      <td>CCK_8</td>\n",
       "      <td>TetrazoliumSalt</td>\n",
       "      <td>other</td>\n",
       "      <td>104.89</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Au</td>\n",
       "      <td>Citrate_based</td>\n",
       "      <td>61.06</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-30.55</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Sphere</td>\n",
       "      <td>1.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>CCL-110</td>\n",
       "      <td>Human</td>\n",
       "      <td>Skin</td>\n",
       "      <td>Fibroblast</td>\n",
       "      <td>fetus</td>\n",
       "      <td>Normal</td>\n",
       "      <td>female</td>\n",
       "      <td>MTS</td>\n",
       "      <td>TetrazoliumSalt</td>\n",
       "      <td>other</td>\n",
       "      <td>94.17</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          NP_type        Coating  core_size_nm  hydro_size_nm  Surf_charge_mV  \\\n",
       "0             CuO           None         12.80          305.3            7.60   \n",
       "1            PLGA            PVA        182.00            NaN           -3.50   \n",
       "2             CuO           None         46.30          239.0           42.80   \n",
       "3  Hydroxyapatite           None         20.00            NaN           -5.30   \n",
       "4              Au  Citrate_based         61.06            NaN          -30.55   \n",
       "\n",
       "   Surface_area_m2_g   shape  Dose_microg_mL  Duration_h  Cell_name  \\\n",
       "0                NaN     NaN             6.3         NaN    BEAS_2B   \n",
       "1                NaN  Sphere           500.0         4.0     Calu-3   \n",
       "2               24.1     NaN             0.0        24.0      HCMEC   \n",
       "3                NaN     Rod            10.0        24.0  RAW 264.7   \n",
       "4                NaN  Sphere             1.0        24.0    CCL-110   \n",
       "\n",
       "  Cell_species cell_Organ Cell_morphology Cell_age cell_type     sex  Assay  \\\n",
       "0        Human       Lung      Epithelial    Adult    Normal     NaN    MTS   \n",
       "1        Human       Lung      Epithelial    Adult    cancer    male    MTT   \n",
       "2        Human      Heart     Endothelial    Adult    Normal     NaN    MTT   \n",
       "3        Mouse      Blood      Macrophage    Adult    cancer    male  CCK_8   \n",
       "4        Human       Skin      Fibroblast    fetus    Normal  female    MTS   \n",
       "\n",
       "    Test_indicator nanomaterial_group  Viability  \n",
       "0  TetrazoliumSalt         meta_oxide      82.00  \n",
       "1  TetrazoliumSalt              other      94.29  \n",
       "2  TetrazoliumSalt         meta_oxide      96.13  \n",
       "3  TetrazoliumSalt              other     104.89  \n",
       "4  TetrazoliumSalt              other      94.17  "
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_train_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "19b3d57d",
   "metadata": {},
   "outputs": [],
   "source": [
    "#data_train_df_clean = data_train_df.dropna(thresh=data_train_df.shape[1]-3, axis=0)\n",
    "#print(data_train_df_clean.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4e176266",
   "metadata": {},
   "source": [
    "### Loading the test data into the data frame."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "3f3fa90b",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_test_df = pd.read_csv(\"output/datasets/universal/data_regression_test_df.csv\", encoding=\"utf-8\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "52dadafb",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "NP_type                object\n",
       "Coating                object\n",
       "core_size_nm          float64\n",
       "hydro_size_nm         float64\n",
       "Surf_charge_mV        float64\n",
       "Surface_area_m2_g     float64\n",
       "shape                  object\n",
       "Dose_microg_mL        float64\n",
       "Duration_h            float64\n",
       "Cell_name              object\n",
       "Cell_species           object\n",
       "cell_Organ             object\n",
       "Cell_morphology        object\n",
       "Cell_age               object\n",
       "cell_type              object\n",
       "sex                    object\n",
       "Assay                  object\n",
       "Test_indicator         object\n",
       "nanomaterial_group     object\n",
       "Viability             float64\n",
       "dtype: object"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_test_df.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "d29ca572",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(932, 20)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_test_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "fd043ed7",
   "metadata": {},
   "outputs": [
    {
     "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>NP_type</th>\n",
       "      <th>Coating</th>\n",
       "      <th>core_size_nm</th>\n",
       "      <th>hydro_size_nm</th>\n",
       "      <th>Surf_charge_mV</th>\n",
       "      <th>Surface_area_m2_g</th>\n",
       "      <th>shape</th>\n",
       "      <th>Dose_microg_mL</th>\n",
       "      <th>Duration_h</th>\n",
       "      <th>Cell_name</th>\n",
       "      <th>Cell_species</th>\n",
       "      <th>cell_Organ</th>\n",
       "      <th>Cell_morphology</th>\n",
       "      <th>Cell_age</th>\n",
       "      <th>cell_type</th>\n",
       "      <th>sex</th>\n",
       "      <th>Assay</th>\n",
       "      <th>Test_indicator</th>\n",
       "      <th>nanomaterial_group</th>\n",
       "      <th>Viability</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>IronOxide</td>\n",
       "      <td>OleicAcid</td>\n",
       "      <td>10.9</td>\n",
       "      <td>183.60</td>\n",
       "      <td>-28.50</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Sphere</td>\n",
       "      <td>25.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>SH­SY5Y</td>\n",
       "      <td>Human</td>\n",
       "      <td>BoneMarrow</td>\n",
       "      <td>Epithelial</td>\n",
       "      <td>Young</td>\n",
       "      <td>cancer</td>\n",
       "      <td>female</td>\n",
       "      <td>NR</td>\n",
       "      <td>toluylene red</td>\n",
       "      <td>meta_oxide</td>\n",
       "      <td>104.59</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Al2O3</td>\n",
       "      <td>None</td>\n",
       "      <td>39.7</td>\n",
       "      <td>NaN</td>\n",
       "      <td>36.30</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Sphere</td>\n",
       "      <td>0.1</td>\n",
       "      <td>24.0</td>\n",
       "      <td>HCMEC</td>\n",
       "      <td>Human</td>\n",
       "      <td>Heart</td>\n",
       "      <td>Endothelial</td>\n",
       "      <td>Adult</td>\n",
       "      <td>Normal</td>\n",
       "      <td>NaN</td>\n",
       "      <td>MTT</td>\n",
       "      <td>TetrazoliumSalt</td>\n",
       "      <td>meta_oxide</td>\n",
       "      <td>93.77</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>IronOxide</td>\n",
       "      <td>Folic_acid_based</td>\n",
       "      <td>80.0</td>\n",
       "      <td>420.55</td>\n",
       "      <td>-27.70</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>SKOV­3</td>\n",
       "      <td>Human</td>\n",
       "      <td>Ovary</td>\n",
       "      <td>Epithelial</td>\n",
       "      <td>Young</td>\n",
       "      <td>cancer</td>\n",
       "      <td>female</td>\n",
       "      <td>MTT</td>\n",
       "      <td>TetrazoliumSalt</td>\n",
       "      <td>meta_oxide</td>\n",
       "      <td>91.92</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>ZnO</td>\n",
       "      <td>None</td>\n",
       "      <td>100.0</td>\n",
       "      <td>258.55</td>\n",
       "      <td>-9.96</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Sphere</td>\n",
       "      <td>80.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>SH­SY5Y</td>\n",
       "      <td>Human</td>\n",
       "      <td>BoneMarrow</td>\n",
       "      <td>Epithelial</td>\n",
       "      <td>Young</td>\n",
       "      <td>cancer</td>\n",
       "      <td>female</td>\n",
       "      <td>NR</td>\n",
       "      <td>toluylene red</td>\n",
       "      <td>meta_oxide</td>\n",
       "      <td>49.35</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>MgO</td>\n",
       "      <td>None</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-7.21</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Hexagon</td>\n",
       "      <td>400.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>VERO</td>\n",
       "      <td>Monkey</td>\n",
       "      <td>Kidney</td>\n",
       "      <td>Epithelial</td>\n",
       "      <td>Adult</td>\n",
       "      <td>Normal</td>\n",
       "      <td>female</td>\n",
       "      <td>MTT</td>\n",
       "      <td>TetrazoliumSalt</td>\n",
       "      <td>meta_oxide</td>\n",
       "      <td>70.96</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     NP_type           Coating  core_size_nm  hydro_size_nm  Surf_charge_mV  \\\n",
       "0  IronOxide         OleicAcid          10.9         183.60          -28.50   \n",
       "1      Al2O3              None          39.7            NaN           36.30   \n",
       "2  IronOxide  Folic_acid_based          80.0         420.55          -27.70   \n",
       "3        ZnO              None         100.0         258.55           -9.96   \n",
       "4        MgO              None           NaN            NaN           -7.21   \n",
       "\n",
       "   Surface_area_m2_g    shape  Dose_microg_mL  Duration_h Cell_name  \\\n",
       "0                NaN   Sphere            25.0         6.0   SH­SY5Y   \n",
       "1                NaN   Sphere             0.1        24.0     HCMEC   \n",
       "2                NaN      NaN             0.0        24.0    SKOV­3   \n",
       "3                NaN   Sphere            80.0        24.0   SH­SY5Y   \n",
       "4                NaN  Hexagon           400.0        24.0      VERO   \n",
       "\n",
       "  Cell_species  cell_Organ Cell_morphology Cell_age cell_type     sex Assay  \\\n",
       "0        Human  BoneMarrow      Epithelial    Young    cancer  female    NR   \n",
       "1        Human       Heart     Endothelial    Adult    Normal     NaN   MTT   \n",
       "2        Human       Ovary      Epithelial    Young    cancer  female   MTT   \n",
       "3        Human  BoneMarrow      Epithelial    Young    cancer  female    NR   \n",
       "4       Monkey      Kidney      Epithelial    Adult    Normal  female   MTT   \n",
       "\n",
       "    Test_indicator nanomaterial_group  Viability  \n",
       "0    toluylene red         meta_oxide     104.59  \n",
       "1  TetrazoliumSalt         meta_oxide      93.77  \n",
       "2  TetrazoliumSalt         meta_oxide      91.92  \n",
       "3    toluylene red         meta_oxide      49.35  \n",
       "4  TetrazoliumSalt         meta_oxide      70.96  "
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_test_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "c9e19bb8",
   "metadata": {},
   "outputs": [],
   "source": [
    "#data_test_df_clean = data_test_df.dropna(thresh=data_test_df.shape[1]-3, axis=0)\n",
    "#print(data_test_df_clean.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cffe023c-bf03-4ba8-a278-8b1261da6572",
   "metadata": {},
   "source": [
    "### Use the best scenario identified in step 01"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "2cf8d7ef",
   "metadata": {},
   "outputs": [],
   "source": [
    "sc_ignore_features = None\n",
    "sc_high_cardinality_features = None\n",
    "\n",
    "sc_imputation_type = 'simple'\n",
    "sc_numeric_imputation = 'mean'\n",
    "\n",
    "sc_normalize = True\n",
    "sc_remove_outliers = False"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "eb0fb96c",
   "metadata": {},
   "source": [
    "### Define seeds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "7db1802e",
   "metadata": {},
   "outputs": [],
   "source": [
    "cv_inner_seed = 123456\n",
    "cv_outer_seed = 123"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b866744f",
   "metadata": {},
   "source": [
    "### The Universal Regression Model: Data Preprcoessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "cb2a02f9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style type=\"text/css\">\n",
       "#T_6ba07_row3_col1, #T_6ba07_row27_col1, #T_6ba07_row34_col1, #T_6ba07_row42_col1 {\n",
       "  background-color: lightgreen;\n",
       "}\n",
       "</style>\n",
       "<table id=\"T_6ba07\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th class=\"blank level0\" >&nbsp;</th>\n",
       "      <th id=\"T_6ba07_level0_col0\" class=\"col_heading level0 col0\" >Description</th>\n",
       "      <th id=\"T_6ba07_level0_col1\" class=\"col_heading level0 col1\" >Value</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row0\" class=\"row_heading level0 row0\" >0</th>\n",
       "      <td id=\"T_6ba07_row0_col0\" class=\"data row0 col0\" >session_id</td>\n",
       "      <td id=\"T_6ba07_row0_col1\" class=\"data row0 col1\" >123456</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row1\" class=\"row_heading level0 row1\" >1</th>\n",
       "      <td id=\"T_6ba07_row1_col0\" class=\"data row1 col0\" >Target</td>\n",
       "      <td id=\"T_6ba07_row1_col1\" class=\"data row1 col1\" >Viability</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row2\" class=\"row_heading level0 row2\" >2</th>\n",
       "      <td id=\"T_6ba07_row2_col0\" class=\"data row2 col0\" >Original Data</td>\n",
       "      <td id=\"T_6ba07_row2_col1\" class=\"data row2 col1\" >(3724, 20)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row3\" class=\"row_heading level0 row3\" >3</th>\n",
       "      <td id=\"T_6ba07_row3_col0\" class=\"data row3 col0\" >Missing Values</td>\n",
       "      <td id=\"T_6ba07_row3_col1\" class=\"data row3 col1\" >True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row4\" class=\"row_heading level0 row4\" >4</th>\n",
       "      <td id=\"T_6ba07_row4_col0\" class=\"data row4 col0\" >Numeric Features</td>\n",
       "      <td id=\"T_6ba07_row4_col1\" class=\"data row4 col1\" >6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row5\" class=\"row_heading level0 row5\" >5</th>\n",
       "      <td id=\"T_6ba07_row5_col0\" class=\"data row5 col0\" >Categorical Features</td>\n",
       "      <td id=\"T_6ba07_row5_col1\" class=\"data row5 col1\" >13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row6\" class=\"row_heading level0 row6\" >6</th>\n",
       "      <td id=\"T_6ba07_row6_col0\" class=\"data row6 col0\" >Ordinal Features</td>\n",
       "      <td id=\"T_6ba07_row6_col1\" class=\"data row6 col1\" >False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row7\" class=\"row_heading level0 row7\" >7</th>\n",
       "      <td id=\"T_6ba07_row7_col0\" class=\"data row7 col0\" >High Cardinality Features</td>\n",
       "      <td id=\"T_6ba07_row7_col1\" class=\"data row7 col1\" >False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row8\" class=\"row_heading level0 row8\" >8</th>\n",
       "      <td id=\"T_6ba07_row8_col0\" class=\"data row8 col0\" >High Cardinality Method</td>\n",
       "      <td id=\"T_6ba07_row8_col1\" class=\"data row8 col1\" >None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row9\" class=\"row_heading level0 row9\" >9</th>\n",
       "      <td id=\"T_6ba07_row9_col0\" class=\"data row9 col0\" >Transformed Train Set</td>\n",
       "      <td id=\"T_6ba07_row9_col1\" class=\"data row9 col1\" >(3724, 249)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row10\" class=\"row_heading level0 row10\" >10</th>\n",
       "      <td id=\"T_6ba07_row10_col0\" class=\"data row10 col0\" >Transformed Test Set</td>\n",
       "      <td id=\"T_6ba07_row10_col1\" class=\"data row10 col1\" >(932, 249)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row11\" class=\"row_heading level0 row11\" >11</th>\n",
       "      <td id=\"T_6ba07_row11_col0\" class=\"data row11 col0\" >Shuffle Train-Test</td>\n",
       "      <td id=\"T_6ba07_row11_col1\" class=\"data row11 col1\" >True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row12\" class=\"row_heading level0 row12\" >12</th>\n",
       "      <td id=\"T_6ba07_row12_col0\" class=\"data row12 col0\" >Stratify Train-Test</td>\n",
       "      <td id=\"T_6ba07_row12_col1\" class=\"data row12 col1\" >False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row13\" class=\"row_heading level0 row13\" >13</th>\n",
       "      <td id=\"T_6ba07_row13_col0\" class=\"data row13 col0\" >Fold Generator</td>\n",
       "      <td id=\"T_6ba07_row13_col1\" class=\"data row13 col1\" >KFold</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row14\" class=\"row_heading level0 row14\" >14</th>\n",
       "      <td id=\"T_6ba07_row14_col0\" class=\"data row14 col0\" >Fold Number</td>\n",
       "      <td id=\"T_6ba07_row14_col1\" class=\"data row14 col1\" >5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row15\" class=\"row_heading level0 row15\" >15</th>\n",
       "      <td id=\"T_6ba07_row15_col0\" class=\"data row15 col0\" >CPU Jobs</td>\n",
       "      <td id=\"T_6ba07_row15_col1\" class=\"data row15 col1\" >-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row16\" class=\"row_heading level0 row16\" >16</th>\n",
       "      <td id=\"T_6ba07_row16_col0\" class=\"data row16 col0\" >Use GPU</td>\n",
       "      <td id=\"T_6ba07_row16_col1\" class=\"data row16 col1\" >False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row17\" class=\"row_heading level0 row17\" >17</th>\n",
       "      <td id=\"T_6ba07_row17_col0\" class=\"data row17 col0\" >Log Experiment</td>\n",
       "      <td id=\"T_6ba07_row17_col1\" class=\"data row17 col1\" >False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row18\" class=\"row_heading level0 row18\" >18</th>\n",
       "      <td id=\"T_6ba07_row18_col0\" class=\"data row18 col0\" >Experiment Name</td>\n",
       "      <td id=\"T_6ba07_row18_col1\" class=\"data row18 col1\" >universal</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row19\" class=\"row_heading level0 row19\" >19</th>\n",
       "      <td id=\"T_6ba07_row19_col0\" class=\"data row19 col0\" >USI</td>\n",
       "      <td id=\"T_6ba07_row19_col1\" class=\"data row19 col1\" >0ce4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row20\" class=\"row_heading level0 row20\" >20</th>\n",
       "      <td id=\"T_6ba07_row20_col0\" class=\"data row20 col0\" >Imputation Type</td>\n",
       "      <td id=\"T_6ba07_row20_col1\" class=\"data row20 col1\" >simple</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row21\" class=\"row_heading level0 row21\" >21</th>\n",
       "      <td id=\"T_6ba07_row21_col0\" class=\"data row21 col0\" >Iterative Imputation Iteration</td>\n",
       "      <td id=\"T_6ba07_row21_col1\" class=\"data row21 col1\" >None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row22\" class=\"row_heading level0 row22\" >22</th>\n",
       "      <td id=\"T_6ba07_row22_col0\" class=\"data row22 col0\" >Numeric Imputer</td>\n",
       "      <td id=\"T_6ba07_row22_col1\" class=\"data row22 col1\" >mean</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row23\" class=\"row_heading level0 row23\" >23</th>\n",
       "      <td id=\"T_6ba07_row23_col0\" class=\"data row23 col0\" >Iterative Imputation Numeric Model</td>\n",
       "      <td id=\"T_6ba07_row23_col1\" class=\"data row23 col1\" >None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row24\" class=\"row_heading level0 row24\" >24</th>\n",
       "      <td id=\"T_6ba07_row24_col0\" class=\"data row24 col0\" >Categorical Imputer</td>\n",
       "      <td id=\"T_6ba07_row24_col1\" class=\"data row24 col1\" >constant</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row25\" class=\"row_heading level0 row25\" >25</th>\n",
       "      <td id=\"T_6ba07_row25_col0\" class=\"data row25 col0\" >Iterative Imputation Categorical Model</td>\n",
       "      <td id=\"T_6ba07_row25_col1\" class=\"data row25 col1\" >None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row26\" class=\"row_heading level0 row26\" >26</th>\n",
       "      <td id=\"T_6ba07_row26_col0\" class=\"data row26 col0\" >Unknown Categoricals Handling</td>\n",
       "      <td id=\"T_6ba07_row26_col1\" class=\"data row26 col1\" >least_frequent</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row27\" class=\"row_heading level0 row27\" >27</th>\n",
       "      <td id=\"T_6ba07_row27_col0\" class=\"data row27 col0\" >Normalize</td>\n",
       "      <td id=\"T_6ba07_row27_col1\" class=\"data row27 col1\" >True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row28\" class=\"row_heading level0 row28\" >28</th>\n",
       "      <td id=\"T_6ba07_row28_col0\" class=\"data row28 col0\" >Normalize Method</td>\n",
       "      <td id=\"T_6ba07_row28_col1\" class=\"data row28 col1\" >zscore</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row29\" class=\"row_heading level0 row29\" >29</th>\n",
       "      <td id=\"T_6ba07_row29_col0\" class=\"data row29 col0\" >Transformation</td>\n",
       "      <td id=\"T_6ba07_row29_col1\" class=\"data row29 col1\" >False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row30\" class=\"row_heading level0 row30\" >30</th>\n",
       "      <td id=\"T_6ba07_row30_col0\" class=\"data row30 col0\" >Transformation Method</td>\n",
       "      <td id=\"T_6ba07_row30_col1\" class=\"data row30 col1\" >None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row31\" class=\"row_heading level0 row31\" >31</th>\n",
       "      <td id=\"T_6ba07_row31_col0\" class=\"data row31 col0\" >PCA</td>\n",
       "      <td id=\"T_6ba07_row31_col1\" class=\"data row31 col1\" >False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row32\" class=\"row_heading level0 row32\" >32</th>\n",
       "      <td id=\"T_6ba07_row32_col0\" class=\"data row32 col0\" >PCA Method</td>\n",
       "      <td id=\"T_6ba07_row32_col1\" class=\"data row32 col1\" >None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row33\" class=\"row_heading level0 row33\" >33</th>\n",
       "      <td id=\"T_6ba07_row33_col0\" class=\"data row33 col0\" >PCA Components</td>\n",
       "      <td id=\"T_6ba07_row33_col1\" class=\"data row33 col1\" >None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row34\" class=\"row_heading level0 row34\" >34</th>\n",
       "      <td id=\"T_6ba07_row34_col0\" class=\"data row34 col0\" >Ignore Low Variance</td>\n",
       "      <td id=\"T_6ba07_row34_col1\" class=\"data row34 col1\" >True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row35\" class=\"row_heading level0 row35\" >35</th>\n",
       "      <td id=\"T_6ba07_row35_col0\" class=\"data row35 col0\" >Combine Rare Levels</td>\n",
       "      <td id=\"T_6ba07_row35_col1\" class=\"data row35 col1\" >False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row36\" class=\"row_heading level0 row36\" >36</th>\n",
       "      <td id=\"T_6ba07_row36_col0\" class=\"data row36 col0\" >Rare Level Threshold</td>\n",
       "      <td id=\"T_6ba07_row36_col1\" class=\"data row36 col1\" >None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row37\" class=\"row_heading level0 row37\" >37</th>\n",
       "      <td id=\"T_6ba07_row37_col0\" class=\"data row37 col0\" >Numeric Binning</td>\n",
       "      <td id=\"T_6ba07_row37_col1\" class=\"data row37 col1\" >False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row38\" class=\"row_heading level0 row38\" >38</th>\n",
       "      <td id=\"T_6ba07_row38_col0\" class=\"data row38 col0\" >Remove Outliers</td>\n",
       "      <td id=\"T_6ba07_row38_col1\" class=\"data row38 col1\" >False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row39\" class=\"row_heading level0 row39\" >39</th>\n",
       "      <td id=\"T_6ba07_row39_col0\" class=\"data row39 col0\" >Outliers Threshold</td>\n",
       "      <td id=\"T_6ba07_row39_col1\" class=\"data row39 col1\" >None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row40\" class=\"row_heading level0 row40\" >40</th>\n",
       "      <td id=\"T_6ba07_row40_col0\" class=\"data row40 col0\" >Remove Multicollinearity</td>\n",
       "      <td id=\"T_6ba07_row40_col1\" class=\"data row40 col1\" >False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row41\" class=\"row_heading level0 row41\" >41</th>\n",
       "      <td id=\"T_6ba07_row41_col0\" class=\"data row41 col0\" >Multicollinearity Threshold</td>\n",
       "      <td id=\"T_6ba07_row41_col1\" class=\"data row41 col1\" >None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row42\" class=\"row_heading level0 row42\" >42</th>\n",
       "      <td id=\"T_6ba07_row42_col0\" class=\"data row42 col0\" >Remove Perfect Collinearity</td>\n",
       "      <td id=\"T_6ba07_row42_col1\" class=\"data row42 col1\" >True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row43\" class=\"row_heading level0 row43\" >43</th>\n",
       "      <td id=\"T_6ba07_row43_col0\" class=\"data row43 col0\" >Clustering</td>\n",
       "      <td id=\"T_6ba07_row43_col1\" class=\"data row43 col1\" >False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row44\" class=\"row_heading level0 row44\" >44</th>\n",
       "      <td id=\"T_6ba07_row44_col0\" class=\"data row44 col0\" >Clustering Iteration</td>\n",
       "      <td id=\"T_6ba07_row44_col1\" class=\"data row44 col1\" >None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row45\" class=\"row_heading level0 row45\" >45</th>\n",
       "      <td id=\"T_6ba07_row45_col0\" class=\"data row45 col0\" >Polynomial Features</td>\n",
       "      <td id=\"T_6ba07_row45_col1\" class=\"data row45 col1\" >False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row46\" class=\"row_heading level0 row46\" >46</th>\n",
       "      <td id=\"T_6ba07_row46_col0\" class=\"data row46 col0\" >Polynomial Degree</td>\n",
       "      <td id=\"T_6ba07_row46_col1\" class=\"data row46 col1\" >None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row47\" class=\"row_heading level0 row47\" >47</th>\n",
       "      <td id=\"T_6ba07_row47_col0\" class=\"data row47 col0\" >Trignometry Features</td>\n",
       "      <td id=\"T_6ba07_row47_col1\" class=\"data row47 col1\" >False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row48\" class=\"row_heading level0 row48\" >48</th>\n",
       "      <td id=\"T_6ba07_row48_col0\" class=\"data row48 col0\" >Polynomial Threshold</td>\n",
       "      <td id=\"T_6ba07_row48_col1\" class=\"data row48 col1\" >None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row49\" class=\"row_heading level0 row49\" >49</th>\n",
       "      <td id=\"T_6ba07_row49_col0\" class=\"data row49 col0\" >Group Features</td>\n",
       "      <td id=\"T_6ba07_row49_col1\" class=\"data row49 col1\" >False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row50\" class=\"row_heading level0 row50\" >50</th>\n",
       "      <td id=\"T_6ba07_row50_col0\" class=\"data row50 col0\" >Feature Selection</td>\n",
       "      <td id=\"T_6ba07_row50_col1\" class=\"data row50 col1\" >False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row51\" class=\"row_heading level0 row51\" >51</th>\n",
       "      <td id=\"T_6ba07_row51_col0\" class=\"data row51 col0\" >Feature Selection Method</td>\n",
       "      <td id=\"T_6ba07_row51_col1\" class=\"data row51 col1\" >classic</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row52\" class=\"row_heading level0 row52\" >52</th>\n",
       "      <td id=\"T_6ba07_row52_col0\" class=\"data row52 col0\" >Features Selection Threshold</td>\n",
       "      <td id=\"T_6ba07_row52_col1\" class=\"data row52 col1\" >None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row53\" class=\"row_heading level0 row53\" >53</th>\n",
       "      <td id=\"T_6ba07_row53_col0\" class=\"data row53 col0\" >Feature Interaction</td>\n",
       "      <td id=\"T_6ba07_row53_col1\" class=\"data row53 col1\" >False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row54\" class=\"row_heading level0 row54\" >54</th>\n",
       "      <td id=\"T_6ba07_row54_col0\" class=\"data row54 col0\" >Feature Ratio</td>\n",
       "      <td id=\"T_6ba07_row54_col1\" class=\"data row54 col1\" >False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row55\" class=\"row_heading level0 row55\" >55</th>\n",
       "      <td id=\"T_6ba07_row55_col0\" class=\"data row55 col0\" >Interaction Threshold</td>\n",
       "      <td id=\"T_6ba07_row55_col1\" class=\"data row55 col1\" >None</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row56\" class=\"row_heading level0 row56\" >56</th>\n",
       "      <td id=\"T_6ba07_row56_col0\" class=\"data row56 col0\" >Transform Target</td>\n",
       "      <td id=\"T_6ba07_row56_col1\" class=\"data row56 col1\" >False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_6ba07_level0_row57\" class=\"row_heading level0 row57\" >57</th>\n",
       "      <td id=\"T_6ba07_row57_col0\" class=\"data row57 col0\" >Transform Target Method</td>\n",
       "      <td id=\"T_6ba07_row57_col1\" class=\"data row57 col1\" >box-cox</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<pandas.io.formats.style.Styler at 0x7f77e87e0520>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "stup = setup(data = data_train_df, target = 'Viability', \n",
    "              \n",
    "        test_data = data_test_df,\n",
    "             \n",
    "        categorical_features=categorical_columns,\n",
    "        numeric_features=numerical_columns_no_target,\n",
    "\n",
    "        imputation_type = sc_imputation_type, \n",
    "        numeric_imputation = sc_numeric_imputation, #default: mean\n",
    "\n",
    "        ignore_features = sc_ignore_features,\n",
    "        bin_numeric_features = None,\n",
    "        high_cardinality_features = sc_high_cardinality_features,\n",
    "\n",
    "        normalize = sc_normalize, #default: False\n",
    "        remove_outliers= sc_remove_outliers, #default: False\n",
    "\n",
    "        transformation= False, #default\n",
    "        transformation_method = 'yeo-johnson', #default\n",
    "        transform_target = False, \n",
    "        combine_rare_levels = False,\n",
    "        rare_level_threshold = 0.1, #default: 0.1\n",
    "\n",
    "        remove_multicollinearity = False, #default \n",
    "        multicollinearity_threshold = 0.9, #default\n",
    "\n",
    "        ignore_low_variance = True, \n",
    "\n",
    "        feature_interaction = False, \n",
    "        feature_ratio = False,\n",
    "        polynomial_features = False, \n",
    "        trigonometry_features = False,\n",
    "          \n",
    "        fold = 5,\n",
    "\n",
    "        # don't change the following\n",
    "        session_id = cv_inner_seed,\n",
    "        use_gpu=False,\n",
    "        log_experiment = False,\n",
    "        silent=True,\n",
    "        experiment_name = 'universal')\n",
    "\n",
    "set_config('seed', cv_inner_seed)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "ebe27ee8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(3724, 249)\n",
      "(3724,)\n"
     ]
    }
   ],
   "source": [
    "print(get_config(variable=\"X_train\").shape)\n",
    "print(get_config(variable=\"y_train\").shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "34f3d922",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(3724, 249)\n",
      "(3724,)\n"
     ]
    }
   ],
   "source": [
    "print(get_config(variable=\"X\").shape)\n",
    "print(get_config(variable=\"y\").shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "f8e4ec4d",
   "metadata": {},
   "outputs": [],
   "source": [
    "#set_config('X_train', get_config(variable=\"X\").copy()) \n",
    "#set_config('y_train', get_config(variable=\"y\").copy()) \n",
    "#print(get_config(variable=\"X_train\").shape)\n",
    "#print(get_config(variable=\"y_train\").shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "1b417175",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(932, 249)\n",
      "(932,)\n"
     ]
    }
   ],
   "source": [
    "print(get_config(variable=\"X_test\").shape)\n",
    "print(get_config(variable=\"y_test\").shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "e4387c1e",
   "metadata": {},
   "outputs": [],
   "source": [
    "#load_config('output/models/universal_regression_config.cfg')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "6fff2182",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_train_transformed_df = get_config(variable=\"X\").copy()\n",
    "data_train_transformed_df['Viability'] = get_config(variable=\"y\").copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "cf253b28",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_test_transformed_df = get_config(variable=\"X_test\").copy()\n",
    "data_test_transformed_df['Viability'] = get_config(variable=\"y_test\").copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "43f293b7",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_train_transformed_df.to_csv('output/datasets/universal/transformed/data_regression_train_transformed_df.csv', encoding='utf-8', index=False)\n",
    "data_test_transformed_df.to_csv('output/datasets/universal/transformed/data_regression_test_transformed_df.csv', encoding='utf-8', index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fe54e2d4",
   "metadata": {},
   "source": [
    "## Double cross-validation"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9db53e62",
   "metadata": {},
   "source": [
    "### Inner CV: 5-fold cross-validation for model selection / hyperparameter tuning for 8 models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "8922c9c0",
   "metadata": {},
   "outputs": [],
   "source": [
    "model_names = ['lr', 'lasso', 'ridge', 'svm', 'rf', 'et', 'gbr', 'lightgbm']\n",
    "#model_names = ['lr']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "69c30960",
   "metadata": {},
   "outputs": [],
   "source": [
    "inner_results = dict()\n",
    "inner_results_tuned = dict()\n",
    "outer_results = dict()\n",
    "predicted_results = dict()\n",
    "tuned_models = dict()\n",
    "final_models = dict()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "c2a73417",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style type=\"text/css\">\n",
       "#T_5eb36_row5_col0, #T_5eb36_row5_col1, #T_5eb36_row5_col2, #T_5eb36_row5_col3, #T_5eb36_row5_col4, #T_5eb36_row5_col5 {\n",
       "  background: yellow;\n",
       "}\n",
       "</style>\n",
       "<table id=\"T_5eb36\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th class=\"blank level0\" >&nbsp;</th>\n",
       "      <th id=\"T_5eb36_level0_col0\" class=\"col_heading level0 col0\" >MAE</th>\n",
       "      <th id=\"T_5eb36_level0_col1\" class=\"col_heading level0 col1\" >MSE</th>\n",
       "      <th id=\"T_5eb36_level0_col2\" class=\"col_heading level0 col2\" >RMSE</th>\n",
       "      <th id=\"T_5eb36_level0_col3\" class=\"col_heading level0 col3\" >R2</th>\n",
       "      <th id=\"T_5eb36_level0_col4\" class=\"col_heading level0 col4\" >RMSLE</th>\n",
       "      <th id=\"T_5eb36_level0_col5\" class=\"col_heading level0 col5\" >MAPE</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th class=\"index_name level0\" >Fold</th>\n",
       "      <th class=\"blank col0\" >&nbsp;</th>\n",
       "      <th class=\"blank col1\" >&nbsp;</th>\n",
       "      <th class=\"blank col2\" >&nbsp;</th>\n",
       "      <th class=\"blank col3\" >&nbsp;</th>\n",
       "      <th class=\"blank col4\" >&nbsp;</th>\n",
       "      <th class=\"blank col5\" >&nbsp;</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th id=\"T_5eb36_level0_row0\" class=\"row_heading level0 row0\" >0</th>\n",
       "      <td id=\"T_5eb36_row0_col0\" class=\"data row0 col0\" >9.70</td>\n",
       "      <td id=\"T_5eb36_row0_col1\" class=\"data row0 col1\" >235.23</td>\n",
       "      <td id=\"T_5eb36_row0_col2\" class=\"data row0 col2\" >15.34</td>\n",
       "      <td id=\"T_5eb36_row0_col3\" class=\"data row0 col3\" >0.78</td>\n",
       "      <td id=\"T_5eb36_row0_col4\" class=\"data row0 col4\" >0.46</td>\n",
       "      <td id=\"T_5eb36_row0_col5\" class=\"data row0 col5\" >0.56</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5eb36_level0_row1\" class=\"row_heading level0 row1\" >1</th>\n",
       "      <td id=\"T_5eb36_row1_col0\" class=\"data row1 col0\" >9.00</td>\n",
       "      <td id=\"T_5eb36_row1_col1\" class=\"data row1 col1\" >172.09</td>\n",
       "      <td id=\"T_5eb36_row1_col2\" class=\"data row1 col2\" >13.12</td>\n",
       "      <td id=\"T_5eb36_row1_col3\" class=\"data row1 col3\" >0.81</td>\n",
       "      <td id=\"T_5eb36_row1_col4\" class=\"data row1 col4\" >0.48</td>\n",
       "      <td id=\"T_5eb36_row1_col5\" class=\"data row1 col5\" >1.37</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5eb36_level0_row2\" class=\"row_heading level0 row2\" >2</th>\n",
       "      <td id=\"T_5eb36_row2_col0\" class=\"data row2 col0\" >9.76</td>\n",
       "      <td id=\"T_5eb36_row2_col1\" class=\"data row2 col1\" >211.11</td>\n",
       "      <td id=\"T_5eb36_row2_col2\" class=\"data row2 col2\" >14.53</td>\n",
       "      <td id=\"T_5eb36_row2_col3\" class=\"data row2 col3\" >0.78</td>\n",
       "      <td id=\"T_5eb36_row2_col4\" class=\"data row2 col4\" >0.37</td>\n",
       "      <td id=\"T_5eb36_row2_col5\" class=\"data row2 col5\" >0.34</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5eb36_level0_row3\" class=\"row_heading level0 row3\" >3</th>\n",
       "      <td id=\"T_5eb36_row3_col0\" class=\"data row3 col0\" >8.95</td>\n",
       "      <td id=\"T_5eb36_row3_col1\" class=\"data row3 col1\" >187.55</td>\n",
       "      <td id=\"T_5eb36_row3_col2\" class=\"data row3 col2\" >13.69</td>\n",
       "      <td id=\"T_5eb36_row3_col3\" class=\"data row3 col3\" >0.82</td>\n",
       "      <td id=\"T_5eb36_row3_col4\" class=\"data row3 col4\" >0.39</td>\n",
       "      <td id=\"T_5eb36_row3_col5\" class=\"data row3 col5\" >0.47</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5eb36_level0_row4\" class=\"row_heading level0 row4\" >4</th>\n",
       "      <td id=\"T_5eb36_row4_col0\" class=\"data row4 col0\" >9.81</td>\n",
       "      <td id=\"T_5eb36_row4_col1\" class=\"data row4 col1\" >221.14</td>\n",
       "      <td id=\"T_5eb36_row4_col2\" class=\"data row4 col2\" >14.87</td>\n",
       "      <td id=\"T_5eb36_row4_col3\" class=\"data row4 col3\" >0.78</td>\n",
       "      <td id=\"T_5eb36_row4_col4\" class=\"data row4 col4\" >0.42</td>\n",
       "      <td id=\"T_5eb36_row4_col5\" class=\"data row4 col5\" >0.47</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5eb36_level0_row5\" class=\"row_heading level0 row5\" >Mean</th>\n",
       "      <td id=\"T_5eb36_row5_col0\" class=\"data row5 col0\" >9.44</td>\n",
       "      <td id=\"T_5eb36_row5_col1\" class=\"data row5 col1\" >205.43</td>\n",
       "      <td id=\"T_5eb36_row5_col2\" class=\"data row5 col2\" >14.31</td>\n",
       "      <td id=\"T_5eb36_row5_col3\" class=\"data row5 col3\" >0.80</td>\n",
       "      <td id=\"T_5eb36_row5_col4\" class=\"data row5 col4\" >0.42</td>\n",
       "      <td id=\"T_5eb36_row5_col5\" class=\"data row5 col5\" >0.64</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5eb36_level0_row6\" class=\"row_heading level0 row6\" >Std</th>\n",
       "      <td id=\"T_5eb36_row6_col0\" class=\"data row6 col0\" >0.39</td>\n",
       "      <td id=\"T_5eb36_row6_col1\" class=\"data row6 col1\" >22.80</td>\n",
       "      <td id=\"T_5eb36_row6_col2\" class=\"data row6 col2\" >0.80</td>\n",
       "      <td id=\"T_5eb36_row6_col3\" class=\"data row6 col3\" >0.02</td>\n",
       "      <td id=\"T_5eb36_row6_col4\" class=\"data row6 col4\" >0.04</td>\n",
       "      <td id=\"T_5eb36_row6_col5\" class=\"data row6 col5\" >0.37</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<pandas.io.formats.style.Styler at 0x7f77e806ec40>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for model_name in model_names:\n",
    "\n",
    "    default_model = create_model(model_name, round=2)\n",
    "\n",
    "    results_df = pull()\n",
    "    results_df = results_df.reset_index(level=0)\n",
    "    \n",
    "    inner_results[model_name] = results_df.copy()\n",
    "    results_df.to_csv(\"output/models_results/universal/\" + \"regression_model_\"+model_name+\"_inner_cross_validation_results.tsv\", \n",
    "                    index=False, encoding=\"utf-8\", sep=\"\\t\")\n",
    "\n",
    "\n",
    "    tuned_model = tune_model(default_model, choose_better = True, search_library = 'optuna', n_iter = 50, round=2)\n",
    "\n",
    "    results_df = pull()\n",
    "    results_df = results_df.reset_index(level=0)\n",
    "    \n",
    "    inner_results_tuned[model_name] = results_df.copy()\n",
    "    results_df.to_csv(\"output/models_results/universal/\" + \"regression_model_\"+model_name+\"_tuned_inner_cross_validation_results.tsv\", \n",
    "                    index=False, encoding=\"utf-8\", sep=\"\\t\")\n",
    "    \n",
    "    tuned_models[model_name] = deepcopy(tuned_model)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2543bed8",
   "metadata": {},
   "source": [
    "### Outer CV: 10-fold cross-validation for the tuned model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "7d9c9302",
   "metadata": {},
   "outputs": [],
   "source": [
    "X = get_config(variable=\"X_train\").copy()\n",
    "y = get_config(variable=\"y_train\").copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "3c21760e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(3724, 249)"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "5a9c93d9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(3724,)"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "f052eebd",
   "metadata": {},
   "outputs": [],
   "source": [
    "cv_outer = KFold(n_splits=10, shuffle=True, random_state=cv_outer_seed)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "564a8934",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'\\nfor model_name in model_names:\\n    scores = cross_val_score(tuned_models[model_name], X, y, cv=cv_outer)\\n    scores = np.round(scores, 2);\\n    outer_results[model_name] = scores.copy()\\n    \\n    with open(\"output/models_results/universal/\" + \"regression_model_\"+model_name+\"_tuned_outer_cross_validation_results.tsv\", \\'w\\', newline=\\'\\') as file:\\n        for score in scores:\\n            file.write(str(score) + \"\\n\")\\n        \\n    #with open(the_filename, \\'r\\') as f:\\n    #    my_list = [line.rstrip(\\'\\n\\') for line in f]\\n'"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "'''\n",
    "for model_name in model_names:\n",
    "    scores = cross_val_score(tuned_models[model_name], X, y, cv=cv_outer)\n",
    "    scores = np.round(scores, 2);\n",
    "    outer_results[model_name] = scores.copy()\n",
    "    \n",
    "    with open(\"output/models_results/universal/\" + \"regression_model_\"+model_name+\"_tuned_outer_cross_validation_results.tsv\", 'w', newline='') as file:\n",
    "        for score in scores:\n",
    "            file.write(str(score) + \"\\n\")\n",
    "        \n",
    "    #with open(the_filename, 'r') as f:\n",
    "    #    my_list = [line.rstrip('\\n') for line in f]\n",
    "'''"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "d74323f9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "lr scores: -72.81, 0.39, 0.41, 0.34, -10.15, 0.42, 0.41, -83.71, -59802.52, 0.41\n",
      "lr mean/std: -5996.68 / 17935.31\n",
      "lasso scores: 0.35, 0.28, 0.28, 0.31, 0.31, 0.29, 0.3, 0.31, 0.37, 0.3\n",
      "lasso mean/std: 0.31 / 0.03\n",
      "ridge scores: 0.52, 0.4, 0.41, 0.35, 0.45, 0.42, 0.41, 0.43, 0.45, 0.42\n",
      "ridge mean/std: 0.43 / 0.04\n",
      "svm scores: 0.53, 0.39, 0.42, 0.42, 0.49, 0.42, 0.45, 0.46, 0.58, 0.39\n",
      "svm mean/std: 0.45 / 0.06\n",
      "rf scores: 0.81, 0.77, 0.81, 0.78, 0.88, 0.82, 0.76, 0.83, 0.83, 0.79\n",
      "rf mean/std: 0.81 / 0.03\n",
      "et scores: 0.84, 0.82, 0.84, 0.82, 0.9, 0.85, 0.81, 0.87, 0.84, 0.82\n",
      "et mean/std: 0.84 / 0.03\n",
      "gbr scores: 0.82, 0.82, 0.85, 0.79, 0.88, 0.86, 0.78, 0.87, 0.84, 0.83\n",
      "gbr mean/std: 0.83 / 0.03\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.9208010711051967, subsample=1.0 will be ignored. Current value: bagging_fraction=0.9208010711051967\n",
      "[LightGBM] [Warning] feature_fraction is set=0.7811105331025159, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.7811105331025159\n",
      "[LightGBM] [Warning] bagging_freq is set=2, subsample_freq=0 will be ignored. Current value: bagging_freq=2\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.9208010711051967, subsample=1.0 will be ignored. Current value: bagging_fraction=0.9208010711051967\n",
      "[LightGBM] [Warning] feature_fraction is set=0.7811105331025159, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.7811105331025159\n",
      "[LightGBM] [Warning] bagging_freq is set=2, subsample_freq=0 will be ignored. Current value: bagging_freq=2\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.9208010711051967, subsample=1.0 will be ignored. Current value: bagging_fraction=0.9208010711051967\n",
      "[LightGBM] [Warning] feature_fraction is set=0.7811105331025159, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.7811105331025159\n",
      "[LightGBM] [Warning] bagging_freq is set=2, subsample_freq=0 will be ignored. Current value: bagging_freq=2\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.9208010711051967, subsample=1.0 will be ignored. Current value: bagging_fraction=0.9208010711051967\n",
      "[LightGBM] [Warning] feature_fraction is set=0.7811105331025159, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.7811105331025159\n",
      "[LightGBM] [Warning] bagging_freq is set=2, subsample_freq=0 will be ignored. Current value: bagging_freq=2\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.9208010711051967, subsample=1.0 will be ignored. Current value: bagging_fraction=0.9208010711051967\n",
      "[LightGBM] [Warning] feature_fraction is set=0.7811105331025159, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.7811105331025159\n",
      "[LightGBM] [Warning] bagging_freq is set=2, subsample_freq=0 will be ignored. Current value: bagging_freq=2\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.9208010711051967, subsample=1.0 will be ignored. Current value: bagging_fraction=0.9208010711051967\n",
      "[LightGBM] [Warning] feature_fraction is set=0.7811105331025159, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.7811105331025159\n",
      "[LightGBM] [Warning] bagging_freq is set=2, subsample_freq=0 will be ignored. Current value: bagging_freq=2\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.9208010711051967, subsample=1.0 will be ignored. Current value: bagging_fraction=0.9208010711051967\n",
      "[LightGBM] [Warning] feature_fraction is set=0.7811105331025159, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.7811105331025159\n",
      "[LightGBM] [Warning] bagging_freq is set=2, subsample_freq=0 will be ignored. Current value: bagging_freq=2\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.9208010711051967, subsample=1.0 will be ignored. Current value: bagging_fraction=0.9208010711051967\n",
      "[LightGBM] [Warning] feature_fraction is set=0.7811105331025159, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.7811105331025159\n",
      "[LightGBM] [Warning] bagging_freq is set=2, subsample_freq=0 will be ignored. Current value: bagging_freq=2\n",
      "[LightGBM] [Warning] bagging_fraction is set=0.9208010711051967, subsample=1.0 will be ignored. Current value: bagging_fraction=0.9208010711051967\n",
      "[LightGBM] [Warning] feature_fraction is set=0.7811105331025159, colsample_bytree=1.0 will be ignored. Current value: feature_fraction=0.7811105331025159\n",
      "[LightGBM] [Warning] bagging_freq is set=2, subsample_freq=0 will be ignored. Current value: bagging_freq=2\n",
      "lightgbm scores: 0.81, 0.77, 0.82, 0.79, 0.86, 0.84, 0.76, 0.85, 0.84, 0.77\n",
      "lightgbm mean/std: 0.81 / 0.03\n"
     ]
    }
   ],
   "source": [
    "for model_name in model_names:\n",
    "    scores = cross_validate(tuned_models[model_name], X, y, cv=cv_outer, scoring=('r2','neg_root_mean_squared_error','neg_mean_absolute_error'))\n",
    "    scores_r2 = np.round(scores['test_r2'], 2);\n",
    "    scores_rmse = np.round(scores['test_neg_root_mean_squared_error'], 2);\n",
    "    scores_mae = np.round(scores['test_neg_mean_absolute_error'], 2);\n",
    "    outer_results[model_name] = scores.copy()\n",
    "    \n",
    "    with open(\"output/models_results/universal/\" + \"regression_model_\"+model_name+\"_tuned_outer_cross_validation_results.tsv\", 'w', newline='') as file:\n",
    "        file.write('R2\\tRMSE\\tMAE' + \"\\n\")\n",
    "        for idx, x in enumerate(scores_r2):\n",
    "            file.write(str(scores_r2[idx]) + '\\t' + str(scores_rmse[idx]) + '\\t' + str(scores_mae[idx]) + \"\\n\")\n",
    "\n",
    "    print(model_name + \" scores: \" + ', '.join(scores_r2.astype('str')))\n",
    "    print(model_name + \" mean/std: \" + str(np.round(scores_r2.mean(),2)) + \" / \" + str(np.round(scores_r2.std(), 2)))\n",
    "    \n",
    "    #with open(the_filename, 'r') as f:\n",
    "    #    my_list = [line.rstrip('\\n') for line in f]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "567af0bf",
   "metadata": {},
   "source": [
    "### Finalize the models and evaluate on the test set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "bfae263b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style type=\"text/css\">\n",
       "</style>\n",
       "<table id=\"T_0b1f1\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th class=\"blank level0\" >&nbsp;</th>\n",
       "      <th id=\"T_0b1f1_level0_col0\" class=\"col_heading level0 col0\" >Model</th>\n",
       "      <th id=\"T_0b1f1_level0_col1\" class=\"col_heading level0 col1\" >MAE</th>\n",
       "      <th id=\"T_0b1f1_level0_col2\" class=\"col_heading level0 col2\" >MSE</th>\n",
       "      <th id=\"T_0b1f1_level0_col3\" class=\"col_heading level0 col3\" >RMSE</th>\n",
       "      <th id=\"T_0b1f1_level0_col4\" class=\"col_heading level0 col4\" >R2</th>\n",
       "      <th id=\"T_0b1f1_level0_col5\" class=\"col_heading level0 col5\" >RMSLE</th>\n",
       "      <th id=\"T_0b1f1_level0_col6\" class=\"col_heading level0 col6\" >MAPE</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th id=\"T_0b1f1_level0_row0\" class=\"row_heading level0 row0\" >0</th>\n",
       "      <td id=\"T_0b1f1_row0_col0\" class=\"data row0 col0\" >Linear Regression</td>\n",
       "      <td id=\"T_0b1f1_row0_col1\" class=\"data row0 col1\" >76.94</td>\n",
       "      <td id=\"T_0b1f1_row0_col2\" class=\"data row0 col2\" >3206489.18</td>\n",
       "      <td id=\"T_0b1f1_row0_col3\" class=\"data row0 col3\" >1790.67</td>\n",
       "      <td id=\"T_0b1f1_row0_col4\" class=\"data row0 col4\" >-3069.49</td>\n",
       "      <td id=\"T_0b1f1_row0_col5\" class=\"data row0 col5\" >0.76</td>\n",
       "      <td id=\"T_0b1f1_row0_col6\" class=\"data row0 col6\" >2.65</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<pandas.io.formats.style.Styler at 0x7f7816b8c190>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<style type=\"text/css\">\n",
       "</style>\n",
       "<table id=\"T_5ff5a\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th class=\"blank level0\" >&nbsp;</th>\n",
       "      <th id=\"T_5ff5a_level0_col0\" class=\"col_heading level0 col0\" >Model</th>\n",
       "      <th id=\"T_5ff5a_level0_col1\" class=\"col_heading level0 col1\" >MAE</th>\n",
       "      <th id=\"T_5ff5a_level0_col2\" class=\"col_heading level0 col2\" >MSE</th>\n",
       "      <th id=\"T_5ff5a_level0_col3\" class=\"col_heading level0 col3\" >RMSE</th>\n",
       "      <th id=\"T_5ff5a_level0_col4\" class=\"col_heading level0 col4\" >R2</th>\n",
       "      <th id=\"T_5ff5a_level0_col5\" class=\"col_heading level0 col5\" >RMSLE</th>\n",
       "      <th id=\"T_5ff5a_level0_col6\" class=\"col_heading level0 col6\" >MAPE</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th id=\"T_5ff5a_level0_row0\" class=\"row_heading level0 row0\" >0</th>\n",
       "      <td id=\"T_5ff5a_row0_col0\" class=\"data row0 col0\" >Lasso Regression</td>\n",
       "      <td id=\"T_5ff5a_row0_col1\" class=\"data row0 col1\" >20.17</td>\n",
       "      <td id=\"T_5ff5a_row0_col2\" class=\"data row0 col2\" >701.20</td>\n",
       "      <td id=\"T_5ff5a_row0_col3\" class=\"data row0 col3\" >26.48</td>\n",
       "      <td id=\"T_5ff5a_row0_col4\" class=\"data row0 col4\" >0.33</td>\n",
       "      <td id=\"T_5ff5a_row0_col5\" class=\"data row0 col5\" >0.79</td>\n",
       "      <td id=\"T_5ff5a_row0_col6\" class=\"data row0 col6\" >2.29</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<pandas.io.formats.style.Styler at 0x7f78169c2d30>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<style type=\"text/css\">\n",
       "</style>\n",
       "<table id=\"T_c89fc\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th class=\"blank level0\" >&nbsp;</th>\n",
       "      <th id=\"T_c89fc_level0_col0\" class=\"col_heading level0 col0\" >Model</th>\n",
       "      <th id=\"T_c89fc_level0_col1\" class=\"col_heading level0 col1\" >MAE</th>\n",
       "      <th id=\"T_c89fc_level0_col2\" class=\"col_heading level0 col2\" >MSE</th>\n",
       "      <th id=\"T_c89fc_level0_col3\" class=\"col_heading level0 col3\" >RMSE</th>\n",
       "      <th id=\"T_c89fc_level0_col4\" class=\"col_heading level0 col4\" >R2</th>\n",
       "      <th id=\"T_c89fc_level0_col5\" class=\"col_heading level0 col5\" >RMSLE</th>\n",
       "      <th id=\"T_c89fc_level0_col6\" class=\"col_heading level0 col6\" >MAPE</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th id=\"T_c89fc_level0_row0\" class=\"row_heading level0 row0\" >0</th>\n",
       "      <td id=\"T_c89fc_row0_col0\" class=\"data row0 col0\" >Ridge Regression</td>\n",
       "      <td id=\"T_c89fc_row0_col1\" class=\"data row0 col1\" >18.21</td>\n",
       "      <td id=\"T_c89fc_row0_col2\" class=\"data row0 col2\" >582.38</td>\n",
       "      <td id=\"T_c89fc_row0_col3\" class=\"data row0 col3\" >24.13</td>\n",
       "      <td id=\"T_c89fc_row0_col4\" class=\"data row0 col4\" >0.44</td>\n",
       "      <td id=\"T_c89fc_row0_col5\" class=\"data row0 col5\" >0.73</td>\n",
       "      <td id=\"T_c89fc_row0_col6\" class=\"data row0 col6\" >1.97</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<pandas.io.formats.style.Styler at 0x7f7816b8cbe0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<style type=\"text/css\">\n",
       "</style>\n",
       "<table id=\"T_5238c\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th class=\"blank level0\" >&nbsp;</th>\n",
       "      <th id=\"T_5238c_level0_col0\" class=\"col_heading level0 col0\" >Model</th>\n",
       "      <th id=\"T_5238c_level0_col1\" class=\"col_heading level0 col1\" >MAE</th>\n",
       "      <th id=\"T_5238c_level0_col2\" class=\"col_heading level0 col2\" >MSE</th>\n",
       "      <th id=\"T_5238c_level0_col3\" class=\"col_heading level0 col3\" >RMSE</th>\n",
       "      <th id=\"T_5238c_level0_col4\" class=\"col_heading level0 col4\" >R2</th>\n",
       "      <th id=\"T_5238c_level0_col5\" class=\"col_heading level0 col5\" >RMSLE</th>\n",
       "      <th id=\"T_5238c_level0_col6\" class=\"col_heading level0 col6\" >MAPE</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th id=\"T_5238c_level0_row0\" class=\"row_heading level0 row0\" >0</th>\n",
       "      <td id=\"T_5238c_row0_col0\" class=\"data row0 col0\" >Support Vector Regression</td>\n",
       "      <td id=\"T_5238c_row0_col1\" class=\"data row0 col1\" >15.43</td>\n",
       "      <td id=\"T_5238c_row0_col2\" class=\"data row0 col2\" >547.55</td>\n",
       "      <td id=\"T_5238c_row0_col3\" class=\"data row0 col3\" >23.40</td>\n",
       "      <td id=\"T_5238c_row0_col4\" class=\"data row0 col4\" >0.48</td>\n",
       "      <td id=\"T_5238c_row0_col5\" class=\"data row0 col5\" >0.73</td>\n",
       "      <td id=\"T_5238c_row0_col6\" class=\"data row0 col6\" >2.14</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<pandas.io.formats.style.Styler at 0x7f781745d340>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<style type=\"text/css\">\n",
       "</style>\n",
       "<table id=\"T_0f0fd\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th class=\"blank level0\" >&nbsp;</th>\n",
       "      <th id=\"T_0f0fd_level0_col0\" class=\"col_heading level0 col0\" >Model</th>\n",
       "      <th id=\"T_0f0fd_level0_col1\" class=\"col_heading level0 col1\" >MAE</th>\n",
       "      <th id=\"T_0f0fd_level0_col2\" class=\"col_heading level0 col2\" >MSE</th>\n",
       "      <th id=\"T_0f0fd_level0_col3\" class=\"col_heading level0 col3\" >RMSE</th>\n",
       "      <th id=\"T_0f0fd_level0_col4\" class=\"col_heading level0 col4\" >R2</th>\n",
       "      <th id=\"T_0f0fd_level0_col5\" class=\"col_heading level0 col5\" >RMSLE</th>\n",
       "      <th id=\"T_0f0fd_level0_col6\" class=\"col_heading level0 col6\" >MAPE</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th id=\"T_0f0fd_level0_row0\" class=\"row_heading level0 row0\" >0</th>\n",
       "      <td id=\"T_0f0fd_row0_col0\" class=\"data row0 col0\" >Random Forest Regressor</td>\n",
       "      <td id=\"T_0f0fd_row0_col1\" class=\"data row0 col1\" >8.30</td>\n",
       "      <td id=\"T_0f0fd_row0_col2\" class=\"data row0 col2\" >169.15</td>\n",
       "      <td id=\"T_0f0fd_row0_col3\" class=\"data row0 col3\" >13.01</td>\n",
       "      <td id=\"T_0f0fd_row0_col4\" class=\"data row0 col4\" >0.84</td>\n",
       "      <td id=\"T_0f0fd_row0_col5\" class=\"data row0 col5\" >0.44</td>\n",
       "      <td id=\"T_0f0fd_row0_col6\" class=\"data row0 col6\" >0.84</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<pandas.io.formats.style.Styler at 0x7f77e87dd280>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<style type=\"text/css\">\n",
       "</style>\n",
       "<table id=\"T_34fdb\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th class=\"blank level0\" >&nbsp;</th>\n",
       "      <th id=\"T_34fdb_level0_col0\" class=\"col_heading level0 col0\" >Model</th>\n",
       "      <th id=\"T_34fdb_level0_col1\" class=\"col_heading level0 col1\" >MAE</th>\n",
       "      <th id=\"T_34fdb_level0_col2\" class=\"col_heading level0 col2\" >MSE</th>\n",
       "      <th id=\"T_34fdb_level0_col3\" class=\"col_heading level0 col3\" >RMSE</th>\n",
       "      <th id=\"T_34fdb_level0_col4\" class=\"col_heading level0 col4\" >R2</th>\n",
       "      <th id=\"T_34fdb_level0_col5\" class=\"col_heading level0 col5\" >RMSLE</th>\n",
       "      <th id=\"T_34fdb_level0_col6\" class=\"col_heading level0 col6\" >MAPE</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th id=\"T_34fdb_level0_row0\" class=\"row_heading level0 row0\" >0</th>\n",
       "      <td id=\"T_34fdb_row0_col0\" class=\"data row0 col0\" >Extra Trees Regressor</td>\n",
       "      <td id=\"T_34fdb_row0_col1\" class=\"data row0 col1\" >7.62</td>\n",
       "      <td id=\"T_34fdb_row0_col2\" class=\"data row0 col2\" >148.64</td>\n",
       "      <td id=\"T_34fdb_row0_col3\" class=\"data row0 col3\" >12.19</td>\n",
       "      <td id=\"T_34fdb_row0_col4\" class=\"data row0 col4\" >0.86</td>\n",
       "      <td id=\"T_34fdb_row0_col5\" class=\"data row0 col5\" >0.37</td>\n",
       "      <td id=\"T_34fdb_row0_col6\" class=\"data row0 col6\" >0.57</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<pandas.io.formats.style.Styler at 0x7f77e880bc40>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<style type=\"text/css\">\n",
       "</style>\n",
       "<table id=\"T_61341\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th class=\"blank level0\" >&nbsp;</th>\n",
       "      <th id=\"T_61341_level0_col0\" class=\"col_heading level0 col0\" >Model</th>\n",
       "      <th id=\"T_61341_level0_col1\" class=\"col_heading level0 col1\" >MAE</th>\n",
       "      <th id=\"T_61341_level0_col2\" class=\"col_heading level0 col2\" >MSE</th>\n",
       "      <th id=\"T_61341_level0_col3\" class=\"col_heading level0 col3\" >RMSE</th>\n",
       "      <th id=\"T_61341_level0_col4\" class=\"col_heading level0 col4\" >R2</th>\n",
       "      <th id=\"T_61341_level0_col5\" class=\"col_heading level0 col5\" >RMSLE</th>\n",
       "      <th id=\"T_61341_level0_col6\" class=\"col_heading level0 col6\" >MAPE</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th id=\"T_61341_level0_row0\" class=\"row_heading level0 row0\" >0</th>\n",
       "      <td id=\"T_61341_row0_col0\" class=\"data row0 col0\" >Gradient Boosting Regressor</td>\n",
       "      <td id=\"T_61341_row0_col1\" class=\"data row0 col1\" >8.14</td>\n",
       "      <td id=\"T_61341_row0_col2\" class=\"data row0 col2\" >143.36</td>\n",
       "      <td id=\"T_61341_row0_col3\" class=\"data row0 col3\" >11.97</td>\n",
       "      <td id=\"T_61341_row0_col4\" class=\"data row0 col4\" >0.86</td>\n",
       "      <td id=\"T_61341_row0_col5\" class=\"data row0 col5\" >0.41</td>\n",
       "      <td id=\"T_61341_row0_col6\" class=\"data row0 col6\" >0.75</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<pandas.io.formats.style.Styler at 0x7f77c7448b20>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<style type=\"text/css\">\n",
       "</style>\n",
       "<table id=\"T_2e884\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th class=\"blank level0\" >&nbsp;</th>\n",
       "      <th id=\"T_2e884_level0_col0\" class=\"col_heading level0 col0\" >Model</th>\n",
       "      <th id=\"T_2e884_level0_col1\" class=\"col_heading level0 col1\" >MAE</th>\n",
       "      <th id=\"T_2e884_level0_col2\" class=\"col_heading level0 col2\" >MSE</th>\n",
       "      <th id=\"T_2e884_level0_col3\" class=\"col_heading level0 col3\" >RMSE</th>\n",
       "      <th id=\"T_2e884_level0_col4\" class=\"col_heading level0 col4\" >R2</th>\n",
       "      <th id=\"T_2e884_level0_col5\" class=\"col_heading level0 col5\" >RMSLE</th>\n",
       "      <th id=\"T_2e884_level0_col6\" class=\"col_heading level0 col6\" >MAPE</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th id=\"T_2e884_level0_row0\" class=\"row_heading level0 row0\" >0</th>\n",
       "      <td id=\"T_2e884_row0_col0\" class=\"data row0 col0\" >Light Gradient Boosting Machine</td>\n",
       "      <td id=\"T_2e884_row0_col1\" class=\"data row0 col1\" >8.88</td>\n",
       "      <td id=\"T_2e884_row0_col2\" class=\"data row0 col2\" >171.98</td>\n",
       "      <td id=\"T_2e884_row0_col3\" class=\"data row0 col3\" >13.11</td>\n",
       "      <td id=\"T_2e884_row0_col4\" class=\"data row0 col4\" >0.84</td>\n",
       "      <td id=\"T_2e884_row0_col5\" class=\"data row0 col5\" >0.45</td>\n",
       "      <td id=\"T_2e884_row0_col6\" class=\"data row0 col6\" >0.88</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<pandas.io.formats.style.Styler at 0x7f7816ab6970>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for model_name in model_names:\n",
    "\n",
    "    final_model = finalize_model(tuned_models[model_name])\n",
    "    \n",
    "    final_models[model_name] = deepcopy(final_model)\n",
    "    \n",
    "    hyperparams = pd.DataFrame.from_dict(final_model.get_params(), orient='index', columns=['parameter_value'])\n",
    "    hyperparams = hyperparams.reset_index(level=0)\n",
    "    hyperparams.to_csv(\"output/models_parameters/universal/\" + \"regression_model_\"+model_name+\"_hyperparams.tsv\", \n",
    "                    index=False, encoding=\"utf-8\", sep=\"\\t\")\n",
    "    \n",
    "    model_predict = predict_model(final_model, data=data_test_df, round=2)\n",
    "    \n",
    "    results_df = pull()\n",
    "    results_df = results_df.reset_index(level=0)\n",
    "    \n",
    "    predicted_results[model_name] = results_df.copy()\n",
    "    results_df.to_csv(\"output/models_results/universal/\" + \"regression_model_\"+model_name+\"_test_set_validation_results.tsv\", \n",
    "                    index=False, encoding=\"utf-8\", sep=\"\\t\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7cef3c2f",
   "metadata": {},
   "source": [
    "### Save models and configs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "f7dbf7e3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Transformation Pipeline and Model Successfully Saved\n",
      "Transformation Pipeline and Model Successfully Saved\n",
      "Transformation Pipeline and Model Successfully Saved\n",
      "Transformation Pipeline and Model Successfully Saved\n",
      "Transformation Pipeline and Model Successfully Saved\n",
      "Transformation Pipeline and Model Successfully Saved\n",
      "Transformation Pipeline and Model Successfully Saved\n",
      "Transformation Pipeline and Model Successfully Saved\n",
      "Transformation Pipeline and Model Successfully Saved\n",
      "Transformation Pipeline and Model Successfully Saved\n",
      "Transformation Pipeline and Model Successfully Saved\n",
      "Transformation Pipeline and Model Successfully Saved\n",
      "Transformation Pipeline and Model Successfully Saved\n",
      "Transformation Pipeline and Model Successfully Saved\n",
      "Transformation Pipeline and Model Successfully Saved\n",
      "Transformation Pipeline and Model Successfully Saved\n"
     ]
    }
   ],
   "source": [
    "for model_name in model_names:\n",
    "    save_config('output/models/universal/regression_config.cfg')\n",
    "    save_model(tuned_models[model_name], 'output/models/universal/regression_model_'+model_name+'_tuned.mdl')\n",
    "    save_model(final_models[model_name], 'output/models/universal/regression_model_'+model_name+'_final.mdl')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5f8dee9d",
   "metadata": {},
   "source": [
    "### Create Plots"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "521ec3e3",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_test = get_config(variable=\"X_test\").copy()\n",
    "y_test = get_config(variable=\"y_test\").copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "44efd5ca",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 800x550 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for model_name in ['rf', 'et', 'gbr', 'lr', 'lasso', 'ridge']:\n",
    "    \n",
    "    #import matplotlib.pyplot as plt\n",
    "    #plt.rcParams[\"figure.figsize\"] = [16,9]\n",
    "\n",
    "    visualizer1 = FeatureImportances(final_models[model_name], size=(800,550), topn=10)\n",
    "    visualizer1.fit(X, y)\n",
    "    visualizer1.show(outpath=\"output/plots/universal/\" + \"regression_model_\"+model_name+\"_feature_importance_plot.png\",dpi = 300)\n",
    "\n",
    "    plt.clf() "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "b4489866",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 800x550 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for model_name in model_names:\n",
    "\n",
    "    visualizer2 = ResidualsPlot(final_models[model_name], hist=False, qqplot=True, size=(800,550))\n",
    "    visualizer2.fit(X, y)\n",
    "    visualizer2.score(X_test, y_test)\n",
    "    visualizer2.show(outpath=\"output/plots/universal/\" + \"regression_model_\"+model_name+\"_residuals_plot.png\",dpi = 300)\n",
    "    \n",
    "    plt.clf() "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "1a279e6d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 800x550 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for model_name in model_names:\n",
    "\n",
    "    visualizer3 = PredictionError(final_models[model_name], size=(800,550))\n",
    "    visualizer3.fit(X, y)\n",
    "    visualizer3.score(X_test, y_test)\n",
    "    visualizer3.show(outpath=\"output/plots/universal/\" + \"regression_model_\"+model_name+\"_error_plot.png\",dpi = 300)    \n",
    "    \n",
    "    plt.clf()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3a2eafd5",
   "metadata": {},
   "source": [
    "### Load model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "2406f629",
   "metadata": {},
   "outputs": [],
   "source": [
    "model_names = ['lr', 'lasso', 'ridge', 'svm', 'rf', 'et', 'gbr', 'lightgbm']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "6f5acd9b",
   "metadata": {},
   "outputs": [],
   "source": [
    "final_config = load_config('output/models/universal/regression_config.cfg')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "7503a083",
   "metadata": {},
   "outputs": [],
   "source": [
    "tuned_models_load = dict()\n",
    "final_models_load = dict()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "af29b475",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Transformation Pipeline and Model Successfully Loaded\n",
      "Transformation Pipeline and Model Successfully Loaded\n",
      "Transformation Pipeline and Model Successfully Loaded\n",
      "Transformation Pipeline and Model Successfully Loaded\n",
      "Transformation Pipeline and Model Successfully Loaded\n",
      "Transformation Pipeline and Model Successfully Loaded\n",
      "Transformation Pipeline and Model Successfully Loaded\n",
      "Transformation Pipeline and Model Successfully Loaded\n",
      "Transformation Pipeline and Model Successfully Loaded\n",
      "Transformation Pipeline and Model Successfully Loaded\n",
      "Transformation Pipeline and Model Successfully Loaded\n",
      "Transformation Pipeline and Model Successfully Loaded\n",
      "Transformation Pipeline and Model Successfully Loaded\n",
      "Transformation Pipeline and Model Successfully Loaded\n",
      "Transformation Pipeline and Model Successfully Loaded\n",
      "Transformation Pipeline and Model Successfully Loaded\n"
     ]
    }
   ],
   "source": [
    "for model_name in model_names:\n",
    "    tuned_models_load[model_name] = load_model('output/models/universal/regression_model_'+model_name+'_tuned.mdl')\n",
    "    final_models_load[model_name] = load_model('output/models/universal/regression_model_'+model_name+'_final.mdl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "acba0082",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'boosting_type': 'gbdt',\n",
       " 'objective': None,\n",
       " 'num_leaves': 225,\n",
       " 'max_depth': -1,\n",
       " 'learning_rate': 0.0492126940458128,\n",
       " 'n_estimators': 206,\n",
       " 'subsample_for_bin': 200000,\n",
       " 'min_split_gain': 0.4382254035210341,\n",
       " 'min_child_weight': 0.001,\n",
       " 'min_child_samples': 20,\n",
       " 'subsample': 1.0,\n",
       " 'subsample_freq': 0,\n",
       " 'colsample_bytree': 1.0,\n",
       " 'reg_alpha': 1.253532007236206e-06,\n",
       " 'reg_lambda': 0.000832480550309834,\n",
       " 'random_state': 123456,\n",
       " 'n_jobs': -1,\n",
       " 'silent': 'warn',\n",
       " 'importance_type': 'split',\n",
       " '_Booster': <lightgbm.basic.Booster at 0x7f84173f4c70>,\n",
       " '_evals_result': None,\n",
       " '_best_score': defaultdict(collections.OrderedDict, {}),\n",
       " '_best_iteration': None,\n",
       " '_other_params': {'feature_fraction': 0.7811105331025159,\n",
       "  'bagging_fraction': 0.9208010711051967,\n",
       "  'bagging_freq': 2},\n",
       " '_objective': 'regression',\n",
       " 'class_weight': None,\n",
       " '_class_weight': None,\n",
       " '_class_map': None,\n",
       " '_n_features': 249,\n",
       " '_n_features_in': 249,\n",
       " '_classes': None,\n",
       " '_n_classes': None,\n",
       " 'feature_fraction': 0.7811105331025159,\n",
       " 'bagging_fraction': 0.9208010711051967,\n",
       " 'bagging_freq': 2,\n",
       " '_fobj': None,\n",
       " 'fitted_': True}"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vars(final_models_load['lightgbm']._final_estimator)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "1cffe116",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 800x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_model(final_models_load['et']._final_estimator, 'feature')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "24442ff3",
   "metadata": {},
   "source": [
    "## Applicability Domain - Williams Plot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "06f039fa",
   "metadata": {},
   "outputs": [],
   "source": [
    "res_train = predict_model(final_models_load['et'], data=data_train_df, round=2)\n",
    "res_test = predict_model(final_models_load['et'], data=data_test_df, round=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "id": "f0171da6",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f840adc16a0>"
      ]
     },
     "execution_count": 93,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 800x550 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "train_plot = f.calculate_leverage(res_train)\n",
    "train_plot = f.calculate_residuals(train_plot)\n",
    "train_plot = f.calculate_standard_residuals(train_plot)\n",
    "\n",
    "test_plot = f.calculate_leverage(res_test)\n",
    "test_plot = f.calculate_residuals(test_plot)\n",
    "test_plot = f.calculate_standard_residuals(test_plot)\n",
    "\n",
    "hat = data_train_df.shape[1] * 3 / data_train_df.shape[0]\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "ax.scatter(train_plot['Leverage'], train_plot['Standard Residual'], marker='o', c='green', label = 'Train')\n",
    "ax.scatter(test_plot['Leverage'], test_plot['Standard Residual'], marker='o', c='orange', label = 'Test')\n",
    "ax.set_xlabel('Leverage')\n",
    "ax.set_ylabel('Standardized Residuals')\n",
    "ax.axhline(y=3.0, xmin=0, xmax=3.0, color='k')\n",
    "ax.axhline(y=-3.0, xmin=0.0, xmax=3.0, color='k')\n",
    "ax.axvline(x=hat, ymin=-10.0, ymax=10.0, color='k')\n",
    "ax.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "id": "462181cf",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Valid    1.0\n",
      "Name: Within Domain, dtype: float64\n",
      "Valid    932\n",
      "Name: Within Domain, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "print(test_plot['Within Domain'].value_counts(normalize=True))\n",
    "print(test_plot['Within Domain'].value_counts(normalize=False))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "id": "b638b8e8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Valid      0.979614\n",
      "Invalid    0.020386\n",
      "Name: Within SD, dtype: float64\n",
      "Valid      913\n",
      "Invalid     19\n",
      "Name: Within SD, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "print(test_plot['Within SD'].value_counts(normalize=True))\n",
    "print(test_plot['Within SD'].value_counts(normalize=False))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "id": "ff2337b3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Valid      0.979614\n",
      "Invalid    0.020386\n",
      "Name: Within Domain and SD, dtype: float64\n",
      "Valid      913\n",
      "Invalid     19\n",
      "Name: Within Domain and SD, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "print(test_plot['Within Domain and SD'].value_counts(normalize=True))\n",
    "print(test_plot['Within Domain and SD'].value_counts(normalize=False))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}