IU-ML-Unsupervised-Unit5.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "6eb2a04f",
   "metadata": {},
   "source": [
    "# Machine Learning - Unsupervised Learning and Feature Engineering\n",
    "## Unit 5. Feature Selection\n",
    "### Lecturer: Tai Le Quy\n",
    "#### 30.8.2023\n",
    "Thanks Prof. Dr. Christian Müller-Kett for his code."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ec3ed8bc",
   "metadata": {},
   "source": [
    "## 1 Feature Importance"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a2214ce2",
   "metadata": {},
   "source": [
    "### 1.1 ANOVA test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "a61673c1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Feature Importance\n",
    "# ANOVA test\n",
    "\n",
    "#SelectKBest is a function that takes two arrays as input: the data points X and target \n",
    "#or response variable or column y, and\n",
    "#returns a pair of arrays (features scores and p-values) or just the features’ scores array.\n",
    "#It return f_classif is a function that performs ANOVA test by assigning it a p-value for\n",
    "#a classification task. Then, SelectKBest will rank the different features according to their\n",
    "#assigned p-value (the lower the better) and will retain the best k features. Hence, k\n",
    "#must be predefined by the user.\n",
    "\n",
    "#%% import libraries\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.datasets import load_iris\n",
    "from sklearn.feature_selection import SelectKBest   \n",
    "from sklearn.feature_selection import f_classif"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "797f7dc8",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% load sample data\n",
    "iris = load_iris()\n",
    "feature_names = load_iris().feature_names\n",
    "X = pd.DataFrame(iris.data, columns=iris.feature_names)\n",
    "y = iris.target"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "f0d2f2b8",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%%\n",
    "# create feature selector which uses F-values\n",
    "# in an ANOVA-test between test samples and labels\n",
    "# to select 2 features\n",
    "selector = SelectKBest(score_func=f_classif, k=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "2f72cd40",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% fit the selector\n",
    "X_new = selector.fit_transform(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "f4579ba1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[1.4, 0.2],\n",
       "       [1.4, 0.2],\n",
       "       [1.3, 0.2],\n",
       "       [1.5, 0.2],\n",
       "       [1.4, 0.2],\n",
       "       [1.7, 0.4],\n",
       "       [1.4, 0.3],\n",
       "       [1.5, 0.2],\n",
       "       [1.4, 0.2],\n",
       "       [1.5, 0.1],\n",
       "       [1.5, 0.2],\n",
       "       [1.6, 0.2],\n",
       "       [1.4, 0.1],\n",
       "       [1.1, 0.1],\n",
       "       [1.2, 0.2],\n",
       "       [1.5, 0.4],\n",
       "       [1.3, 0.4],\n",
       "       [1.4, 0.3],\n",
       "       [1.7, 0.3],\n",
       "       [1.5, 0.3],\n",
       "       [1.7, 0.2],\n",
       "       [1.5, 0.4],\n",
       "       [1. , 0.2],\n",
       "       [1.7, 0.5],\n",
       "       [1.9, 0.2],\n",
       "       [1.6, 0.2],\n",
       "       [1.6, 0.4],\n",
       "       [1.5, 0.2],\n",
       "       [1.4, 0.2],\n",
       "       [1.6, 0.2],\n",
       "       [1.6, 0.2],\n",
       "       [1.5, 0.4],\n",
       "       [1.5, 0.1],\n",
       "       [1.4, 0.2],\n",
       "       [1.5, 0.2],\n",
       "       [1.2, 0.2],\n",
       "       [1.3, 0.2],\n",
       "       [1.4, 0.1],\n",
       "       [1.3, 0.2],\n",
       "       [1.5, 0.2],\n",
       "       [1.3, 0.3],\n",
       "       [1.3, 0.3],\n",
       "       [1.3, 0.2],\n",
       "       [1.6, 0.6],\n",
       "       [1.9, 0.4],\n",
       "       [1.4, 0.3],\n",
       "       [1.6, 0.2],\n",
       "       [1.4, 0.2],\n",
       "       [1.5, 0.2],\n",
       "       [1.4, 0.2],\n",
       "       [4.7, 1.4],\n",
       "       [4.5, 1.5],\n",
       "       [4.9, 1.5],\n",
       "       [4. , 1.3],\n",
       "       [4.6, 1.5],\n",
       "       [4.5, 1.3],\n",
       "       [4.7, 1.6],\n",
       "       [3.3, 1. ],\n",
       "       [4.6, 1.3],\n",
       "       [3.9, 1.4],\n",
       "       [3.5, 1. ],\n",
       "       [4.2, 1.5],\n",
       "       [4. , 1. ],\n",
       "       [4.7, 1.4],\n",
       "       [3.6, 1.3],\n",
       "       [4.4, 1.4],\n",
       "       [4.5, 1.5],\n",
       "       [4.1, 1. ],\n",
       "       [4.5, 1.5],\n",
       "       [3.9, 1.1],\n",
       "       [4.8, 1.8],\n",
       "       [4. , 1.3],\n",
       "       [4.9, 1.5],\n",
       "       [4.7, 1.2],\n",
       "       [4.3, 1.3],\n",
       "       [4.4, 1.4],\n",
       "       [4.8, 1.4],\n",
       "       [5. , 1.7],\n",
       "       [4.5, 1.5],\n",
       "       [3.5, 1. ],\n",
       "       [3.8, 1.1],\n",
       "       [3.7, 1. ],\n",
       "       [3.9, 1.2],\n",
       "       [5.1, 1.6],\n",
       "       [4.5, 1.5],\n",
       "       [4.5, 1.6],\n",
       "       [4.7, 1.5],\n",
       "       [4.4, 1.3],\n",
       "       [4.1, 1.3],\n",
       "       [4. , 1.3],\n",
       "       [4.4, 1.2],\n",
       "       [4.6, 1.4],\n",
       "       [4. , 1.2],\n",
       "       [3.3, 1. ],\n",
       "       [4.2, 1.3],\n",
       "       [4.2, 1.2],\n",
       "       [4.2, 1.3],\n",
       "       [4.3, 1.3],\n",
       "       [3. , 1.1],\n",
       "       [4.1, 1.3],\n",
       "       [6. , 2.5],\n",
       "       [5.1, 1.9],\n",
       "       [5.9, 2.1],\n",
       "       [5.6, 1.8],\n",
       "       [5.8, 2.2],\n",
       "       [6.6, 2.1],\n",
       "       [4.5, 1.7],\n",
       "       [6.3, 1.8],\n",
       "       [5.8, 1.8],\n",
       "       [6.1, 2.5],\n",
       "       [5.1, 2. ],\n",
       "       [5.3, 1.9],\n",
       "       [5.5, 2.1],\n",
       "       [5. , 2. ],\n",
       "       [5.1, 2.4],\n",
       "       [5.3, 2.3],\n",
       "       [5.5, 1.8],\n",
       "       [6.7, 2.2],\n",
       "       [6.9, 2.3],\n",
       "       [5. , 1.5],\n",
       "       [5.7, 2.3],\n",
       "       [4.9, 2. ],\n",
       "       [6.7, 2. ],\n",
       "       [4.9, 1.8],\n",
       "       [5.7, 2.1],\n",
       "       [6. , 1.8],\n",
       "       [4.8, 1.8],\n",
       "       [4.9, 1.8],\n",
       "       [5.6, 2.1],\n",
       "       [5.8, 1.6],\n",
       "       [6.1, 1.9],\n",
       "       [6.4, 2. ],\n",
       "       [5.6, 2.2],\n",
       "       [5.1, 1.5],\n",
       "       [5.6, 1.4],\n",
       "       [6.1, 2.3],\n",
       "       [5.6, 2.4],\n",
       "       [5.5, 1.8],\n",
       "       [4.8, 1.8],\n",
       "       [5.4, 2.1],\n",
       "       [5.6, 2.4],\n",
       "       [5.1, 2.3],\n",
       "       [5.1, 1.9],\n",
       "       [5.9, 2.3],\n",
       "       [5.7, 2.5],\n",
       "       [5.2, 2.3],\n",
       "       [5. , 1.9],\n",
       "       [5.2, 2. ],\n",
       "       [5.4, 2.3],\n",
       "       [5.1, 1.8]])"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_new"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "9503493a",
   "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>features</th>\n",
       "      <th>Scores</th>\n",
       "      <th>p-values</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>sepal length (cm)</td>\n",
       "      <td>119.264502</td>\n",
       "      <td>1.669669e-31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>sepal width (cm)</td>\n",
       "      <td>49.160040</td>\n",
       "      <td>4.492017e-17</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>petal length (cm)</td>\n",
       "      <td>1180.161182</td>\n",
       "      <td>2.856777e-91</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>petal width (cm)</td>\n",
       "      <td>960.007147</td>\n",
       "      <td>4.169446e-85</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            features       Scores      p-values\n",
       "0  sepal length (cm)   119.264502  1.669669e-31\n",
       "1   sepal width (cm)    49.160040  4.492017e-17\n",
       "2  petal length (cm)  1180.161182  2.856777e-91\n",
       "3   petal width (cm)   960.007147  4.169446e-85"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#%% print F- and p-values per feature\n",
    "pd.DataFrame({'features': feature_names, \n",
    "    'Scores': selector.scores_, \n",
    "    'p-values': selector.pvalues_})"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "35151e20",
   "metadata": {},
   "source": [
    "### 1.2 Chi-square test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "f605adb5",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% import libraries\n",
    "import pandas as pd\n",
    "from sklearn.feature_selection import SelectKBest, chi2\n",
    "from sklearn.preprocessing import LabelEncoder"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "d2061432",
   "metadata": {},
   "outputs": [],
   "source": [
    "#Load the data\n",
    "#This dataset is used to predict the bank churn rate: left a bank or decided to stay\n",
    "#https://www.kaggle.com/datasets/sonalidasgupta95/churn-prediction-of-bank-customers?resource=download\n",
    "churn_df = pd.read_csv('Churn_Modelling.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "830a2c85",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% prepare the data\n",
    "y = churn_df['Exited']\n",
    "X = churn_df[['Gender', 'HasCrCard', 'IsActiveMember']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "416c3673",
   "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>Gender</th>\n",
       "      <th>HasCrCard</th>\n",
       "      <th>IsActiveMember</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Female</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Female</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Female</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Female</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Female</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9995</th>\n",
       "      <td>Male</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9996</th>\n",
       "      <td>Male</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9997</th>\n",
       "      <td>Female</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9998</th>\n",
       "      <td>Male</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9999</th>\n",
       "      <td>Female</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>10000 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      Gender  HasCrCard  IsActiveMember\n",
       "0     Female          1               1\n",
       "1     Female          0               1\n",
       "2     Female          1               0\n",
       "3     Female          0               0\n",
       "4     Female          1               1\n",
       "...      ...        ...             ...\n",
       "9995    Male          1               0\n",
       "9996    Male          1               1\n",
       "9997  Female          0               1\n",
       "9998    Male          1               0\n",
       "9999  Female          1               0\n",
       "\n",
       "[10000 rows x 3 columns]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "9540c17c",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\ipykernel_launcher.py:2: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  \n"
     ]
    }
   ],
   "source": [
    "#%% apply label encoder\n",
    "X['Gender'] = LabelEncoder().fit_transform(X['Gender'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "869f3a72",
   "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>Gender</th>\n",
       "      <th>HasCrCard</th>\n",
       "      <th>IsActiveMember</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9995</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9996</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9997</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9998</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9999</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>10000 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      Gender  HasCrCard  IsActiveMember\n",
       "0          0          1               1\n",
       "1          0          0               1\n",
       "2          0          1               0\n",
       "3          0          0               0\n",
       "4          0          1               1\n",
       "...      ...        ...             ...\n",
       "9995       1          1               0\n",
       "9996       1          1               1\n",
       "9997       0          0               1\n",
       "9998       1          1               0\n",
       "9999       0          1               0\n",
       "\n",
       "[10000 rows x 3 columns]"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "7cd46cce",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "SelectKBest(k=2, score_func=<function chi2 at 0x0000027715F4B8B8>)"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#%% create and fit feature selector\n",
    "selector = SelectKBest(chi2, k=2)\n",
    "selector.fit(X,y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "d960ae2c",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% apply feature selector to the data\n",
    "X_new = selector.transform(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "4154921b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0, 1],\n",
       "       [0, 1],\n",
       "       [0, 0],\n",
       "       ...,\n",
       "       [0, 1],\n",
       "       [1, 0],\n",
       "       [0, 0]], dtype=int64)"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_new"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "275d7a9d",
   "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>features</th>\n",
       "      <th>Scores</th>\n",
       "      <th>p-values</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Gender</td>\n",
       "      <td>51.539926</td>\n",
       "      <td>7.015575e-13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>HasCrCard</td>\n",
       "      <td>0.150041</td>\n",
       "      <td>6.984962e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>IsActiveMember</td>\n",
       "      <td>118.199414</td>\n",
       "      <td>1.568036e-27</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         features      Scores      p-values\n",
       "0          Gender   51.539926  7.015575e-13\n",
       "1       HasCrCard    0.150041  6.984962e-01\n",
       "2  IsActiveMember  118.199414  1.568036e-27"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#%% print Chi²-statistics- and p-values per feature\n",
    "pd.DataFrame({'features': X.columns.values, 'Scores': selector.scores_, 'p-values': selector.pvalues_})\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7e67bdd7",
   "metadata": {},
   "source": [
    "### 1.3 Mutual information"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "9772df7a",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% import libraries\n",
    "from sklearn.feature_selection import mutual_info_classif\n",
    "from sklearn.feature_selection import SelectKBest\n",
    "import pandas as pd\n",
    "from sklearn.datasets import load_iris"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "875e1f1c",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% load sample data\n",
    "iris = load_iris()\n",
    "feature_names = load_iris().feature_names\n",
    "X = pd.DataFrame(iris.data, columns=iris.feature_names)\n",
    "y = iris.target"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "35743852",
   "metadata": {},
   "outputs": [],
   "source": [
    "##%% create and fit feature selector\n",
    "selector = SelectKBest(score_func=mutual_info_classif, k=2)\n",
    "X_new = selector.fit_transform(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "2aec4113",
   "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>features</th>\n",
       "      <th>Scores</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>sepal length (cm)</td>\n",
       "      <td>0.488584</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>sepal width (cm)</td>\n",
       "      <td>0.229371</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>petal length (cm)</td>\n",
       "      <td>0.983380</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>petal width (cm)</td>\n",
       "      <td>0.995523</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            features    Scores\n",
       "0  sepal length (cm)  0.488584\n",
       "1   sepal width (cm)  0.229371\n",
       "2  petal length (cm)  0.983380\n",
       "3   petal width (cm)  0.995523"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#%% print mutual information per feature\n",
    "pd.DataFrame({'features': X.columns.values, 'Scores': selector.scores_})"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "68704446",
   "metadata": {},
   "source": [
    "### 1.4 Permutation Feature Importance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "9cbbc88f",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% load libraries\n",
    "import pandas as pd\n",
    "from sklearn.datasets import load_iris\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.inspection import permutation_importance\n",
    "from sklearn.neighbors import KNeighborsClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "3c0ed3d7",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% load sample data\n",
    "iris = load_iris()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "5d0489b6",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% split the data into training and testing, 0.25 test size default\n",
    "X_train, X_test, y_train,y_test = train_test_split(iris.data, iris.target)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "7921b731",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "KNeighborsClassifier()"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#%% create and fit a KNN model\n",
    "model = KNeighborsClassifier()\n",
    "model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "da737d90",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% assess feature importance by permutation \n",
    "feat_imp = permutation_importance(model, X_test, y_test, n_repeats=10, scoring='accuracy')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "8ee5bf9b",
   "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>features</th>\n",
       "      <th>importances_mean</th>\n",
       "      <th>importances_std</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>sepal length (cm)</td>\n",
       "      <td>-0.002632</td>\n",
       "      <td>0.018421</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>sepal width (cm)</td>\n",
       "      <td>0.013158</td>\n",
       "      <td>0.024262</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>petal length (cm)</td>\n",
       "      <td>0.565789</td>\n",
       "      <td>0.065789</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>petal width (cm)</td>\n",
       "      <td>0.107895</td>\n",
       "      <td>0.039823</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            features  importances_mean  importances_std\n",
       "0  sepal length (cm)         -0.002632         0.018421\n",
       "1   sepal width (cm)          0.013158         0.024262\n",
       "2  petal length (cm)          0.565789         0.065789\n",
       "3   petal width (cm)          0.107895         0.039823"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#%% display calculated feature importances\n",
    "#The scores in the table above show how much model performance and accuracy rate\n",
    "#decreased with a random permutation. Therefore, the most important features are\n",
    "#first “petal length,” then “petal width.” The other two features have very little importance\n",
    "#since their impact on the accuracy rate is very small. \n",
    "\n",
    "pd.DataFrame({'features': iris.feature_names, \n",
    " 'importances_mean': feat_imp['importances_mean'], \n",
    " 'importances_std': feat_imp['importances_std']})"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8050b167",
   "metadata": {},
   "source": [
    "## 2 Feature Variance"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1e51cbe7",
   "metadata": {},
   "source": [
    "#### Example 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "1743b521",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% load libraries\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from sklearn.feature_selection import VarianceThreshold\n",
    "from sklearn.datasets import load_iris\n",
    "from matplotlib import pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "1168719c",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% generate sample data\n",
    "X = np.array([[0, 2, 0, 3], [0, 3, 4, 3], \n",
    "    [0, 5, 1, 2]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "d34378b8",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% apply variance threshold\n",
    "selector = VarianceThreshold(threshold=0.4)\n",
    "Xs = selector.fit_transform(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "d4af7c46",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0.         1.55555556 2.88888889 0.22222222]\n"
     ]
    }
   ],
   "source": [
    "#%% show the variances per feature\n",
    "# (the ones above the threshold were chosen)\n",
    "print(selector.variances_)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "716eca7f",
   "metadata": {},
   "source": [
    "#### Example 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "6d61d736",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% load sample data\n",
    "iris = load_iris()\n",
    "X = pd.DataFrame(iris.data, columns=iris.feature_names)\n",
    "y = iris.target"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "8cf6ff54",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[5.1, 1.4],\n",
       "       [4.9, 1.4],\n",
       "       [4.7, 1.3],\n",
       "       [4.6, 1.5],\n",
       "       [5. , 1.4],\n",
       "       [5.4, 1.7],\n",
       "       [4.6, 1.4],\n",
       "       [5. , 1.5],\n",
       "       [4.4, 1.4],\n",
       "       [4.9, 1.5],\n",
       "       [5.4, 1.5],\n",
       "       [4.8, 1.6],\n",
       "       [4.8, 1.4],\n",
       "       [4.3, 1.1],\n",
       "       [5.8, 1.2],\n",
       "       [5.7, 1.5],\n",
       "       [5.4, 1.3],\n",
       "       [5.1, 1.4],\n",
       "       [5.7, 1.7],\n",
       "       [5.1, 1.5],\n",
       "       [5.4, 1.7],\n",
       "       [5.1, 1.5],\n",
       "       [4.6, 1. ],\n",
       "       [5.1, 1.7],\n",
       "       [4.8, 1.9],\n",
       "       [5. , 1.6],\n",
       "       [5. , 1.6],\n",
       "       [5.2, 1.5],\n",
       "       [5.2, 1.4],\n",
       "       [4.7, 1.6],\n",
       "       [4.8, 1.6],\n",
       "       [5.4, 1.5],\n",
       "       [5.2, 1.5],\n",
       "       [5.5, 1.4],\n",
       "       [4.9, 1.5],\n",
       "       [5. , 1.2],\n",
       "       [5.5, 1.3],\n",
       "       [4.9, 1.4],\n",
       "       [4.4, 1.3],\n",
       "       [5.1, 1.5],\n",
       "       [5. , 1.3],\n",
       "       [4.5, 1.3],\n",
       "       [4.4, 1.3],\n",
       "       [5. , 1.6],\n",
       "       [5.1, 1.9],\n",
       "       [4.8, 1.4],\n",
       "       [5.1, 1.6],\n",
       "       [4.6, 1.4],\n",
       "       [5.3, 1.5],\n",
       "       [5. , 1.4],\n",
       "       [7. , 4.7],\n",
       "       [6.4, 4.5],\n",
       "       [6.9, 4.9],\n",
       "       [5.5, 4. ],\n",
       "       [6.5, 4.6],\n",
       "       [5.7, 4.5],\n",
       "       [6.3, 4.7],\n",
       "       [4.9, 3.3],\n",
       "       [6.6, 4.6],\n",
       "       [5.2, 3.9],\n",
       "       [5. , 3.5],\n",
       "       [5.9, 4.2],\n",
       "       [6. , 4. ],\n",
       "       [6.1, 4.7],\n",
       "       [5.6, 3.6],\n",
       "       [6.7, 4.4],\n",
       "       [5.6, 4.5],\n",
       "       [5.8, 4.1],\n",
       "       [6.2, 4.5],\n",
       "       [5.6, 3.9],\n",
       "       [5.9, 4.8],\n",
       "       [6.1, 4. ],\n",
       "       [6.3, 4.9],\n",
       "       [6.1, 4.7],\n",
       "       [6.4, 4.3],\n",
       "       [6.6, 4.4],\n",
       "       [6.8, 4.8],\n",
       "       [6.7, 5. ],\n",
       "       [6. , 4.5],\n",
       "       [5.7, 3.5],\n",
       "       [5.5, 3.8],\n",
       "       [5.5, 3.7],\n",
       "       [5.8, 3.9],\n",
       "       [6. , 5.1],\n",
       "       [5.4, 4.5],\n",
       "       [6. , 4.5],\n",
       "       [6.7, 4.7],\n",
       "       [6.3, 4.4],\n",
       "       [5.6, 4.1],\n",
       "       [5.5, 4. ],\n",
       "       [5.5, 4.4],\n",
       "       [6.1, 4.6],\n",
       "       [5.8, 4. ],\n",
       "       [5. , 3.3],\n",
       "       [5.6, 4.2],\n",
       "       [5.7, 4.2],\n",
       "       [5.7, 4.2],\n",
       "       [6.2, 4.3],\n",
       "       [5.1, 3. ],\n",
       "       [5.7, 4.1],\n",
       "       [6.3, 6. ],\n",
       "       [5.8, 5.1],\n",
       "       [7.1, 5.9],\n",
       "       [6.3, 5.6],\n",
       "       [6.5, 5.8],\n",
       "       [7.6, 6.6],\n",
       "       [4.9, 4.5],\n",
       "       [7.3, 6.3],\n",
       "       [6.7, 5.8],\n",
       "       [7.2, 6.1],\n",
       "       [6.5, 5.1],\n",
       "       [6.4, 5.3],\n",
       "       [6.8, 5.5],\n",
       "       [5.7, 5. ],\n",
       "       [5.8, 5.1],\n",
       "       [6.4, 5.3],\n",
       "       [6.5, 5.5],\n",
       "       [7.7, 6.7],\n",
       "       [7.7, 6.9],\n",
       "       [6. , 5. ],\n",
       "       [6.9, 5.7],\n",
       "       [5.6, 4.9],\n",
       "       [7.7, 6.7],\n",
       "       [6.3, 4.9],\n",
       "       [6.7, 5.7],\n",
       "       [7.2, 6. ],\n",
       "       [6.2, 4.8],\n",
       "       [6.1, 4.9],\n",
       "       [6.4, 5.6],\n",
       "       [7.2, 5.8],\n",
       "       [7.4, 6.1],\n",
       "       [7.9, 6.4],\n",
       "       [6.4, 5.6],\n",
       "       [6.3, 5.1],\n",
       "       [6.1, 5.6],\n",
       "       [7.7, 6.1],\n",
       "       [6.3, 5.6],\n",
       "       [6.4, 5.5],\n",
       "       [6. , 4.8],\n",
       "       [6.9, 5.4],\n",
       "       [6.7, 5.6],\n",
       "       [6.9, 5.1],\n",
       "       [5.8, 5.1],\n",
       "       [6.8, 5.9],\n",
       "       [6.7, 5.7],\n",
       "       [6.7, 5.2],\n",
       "       [6.3, 5. ],\n",
       "       [6.5, 5.2],\n",
       "       [6.2, 5.4],\n",
       "       [5.9, 5.1]])"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#%% create and fit feature selector\n",
    "selector = VarianceThreshold(threshold=0.6)\n",
    "selector.fit_transform(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "ea8cc65b",
   "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>features</th>\n",
       "      <th>variances</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>sepal length (cm)</td>\n",
       "      <td>0.681122</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>sepal width (cm)</td>\n",
       "      <td>0.188713</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>petal length (cm)</td>\n",
       "      <td>3.095503</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>petal width (cm)</td>\n",
       "      <td>0.577133</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            features  variances\n",
       "0  sepal length (cm)   0.681122\n",
       "1   sepal width (cm)   0.188713\n",
       "2  petal length (cm)   3.095503\n",
       "3   petal width (cm)   0.577133"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#%%\n",
    "# show the variances per feature\n",
    "# (the ones above the threshold were chosen)\n",
    "pd.DataFrame({'features': iris.feature_names, \n",
    "    'variances': selector.variances_})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "07bca1c5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEICAYAAABS0fM3AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8QVMy6AAAACXBIWXMAAAsTAAALEwEAmpwYAAAf0UlEQVR4nO3debgcVZ3/8feHEAhIQoiJEEJCBKLzY5Etw+JGZFAh6ACKAi4ILhEUhEF0eBhZVHTQcQNRMkGZCLKjQpRdh31PYkgCCIZNAhkIW8IOge/vj3MuFJ2+fevem+rLvfV5PU8/XcupU986XV3f2rpaEYGZmdXXSn0dgJmZ9S0nAjOzmnMiMDOrOScCM7OacyIwM6s5JwIzs5pzIhjgJH1a0uU9mG5tSddIelrSj6uIrb+T9D5Jd/V1HP1dT9dRW3Hk3xH0b5LuB74YEX9ewfUeBWwJfDx6uZJImg4sjIhvrYjYzGzF8hHBACZp5V5Mvj5wR2+TwIrQy+WoxJsxpv7I7fgmERF+9eMXcD+wU+7eD7ge+CnwBHBcHnZdHq887lFgCTAX2LRJndOBl4GXgGeAnUg7DUcA9wCPA+cCIwrTnAf8X673GmCTPHxKQ11/zMMD2Khhnsfl7knAQuDfc52nt5o/MAT4bR7+FHArsHaT5ToCOL9h2AnAibl7f+BO4GngXuDLhXLNYppEOtIp1n9Pnv4OYI/CuP2A64AfAU8C9wG7FMaPAP4HeDiPv6Aw7iPAnLxsNwDvarE+bAJckT//R4Aj8/BVgZ/l+h/O3as2LNs387qxCNgdmAzcnes6sjCPY4HzgXPyss4GNu9GO/RoHQXWBE4DFgMPAN8CVirTvn51sR3p6wD86uUHuHwiWAYcDKwMrNbwJfswMAsYnr9w/w8Y3Um908kb5tx/KHATsF7eqPw3cFZh/OeBoYUNzpzO6srDukoEy4Af5PpWazV/4MvAH4HVgUHA1sCwJsu0PvBcx7hcdhGwXe7fFdgwt80OuexWLWKaxBsTwSeAdUlJay/g2Y72zZ/Dy8CX8nwPJG2QO07PXkTasK4FDAZ2yMO3Im0Ut83TfS5/5qs2Wb6heXm+TkqOQ4Ft87jv5PZ7GzCKlFC+27BsR+d5f4m0sT0z17EJ8AKwQS5/bF6WPXP5w0kb3sEl26FH6ygpCVyYYxpPSlJfKNO+fnWxHenrAPzq5Qe4fCL4R8P44pdsx/zl2Y68J9Wi3um8MRHcCfxLoX90/uKt3GTa4aQN/ZrN6srDukoELwFDysyflIRa7ikXprsO2Dd3fxC4p0XZC4BDWsQ0iUIiaDL9HGC3wuewoDBu9dwG6+RleRVYq0kdJ5M32IVhd5ETRcPwfYC/dhLLPcDkQv+HgfsLy/E8MCj3D82xbVsoPwvYPXcfC9xUGLcSKQG9r2Q7dHsdJW3cXwQ2Lgz7MnBVV+1b1XdvIL18jWDgebCzERHxv8BJwC+ARyRNkzSsZL3rA3+Q9JSkp0gb5leAtSUNknS8pHskLSUlJ4CRPV0IYHFEvFBm/qTTNJcBZ0t6WNIPJQ3upN4zSRtMgE/lfgAk7SLpJklP5HlMbliGxpjeQNK+kuYUYty0Yfr/6+iIiOdy5xrAWOCJiHiySbXrA1/vqDPXO5a0x91oLGmD38y6pNMpHR5oqOPxiHgldz+f3x8pjH8+x9rhtfUsIl4lnVpaF0q1Q0/W0ZHAKk2WYUyhv7P2tS44EQw8LS/uRsSJEbE16XD/HcA3Stb7IOmc6/DCa0hEPETaoO5GupawJumwHdKhfWcxPUfaa+uwThfL0en8I+LliPh2RGwMvJt0Tn3fTpbjPGCSpPWAPciJQNKqwO9I55jXjojhwMWFZehsOcjTrw+cAhwEvDVPP79h+s48CIyQNLyTcd9rWO7VI+KsTspu2Mk8HiYllQ7j8rCeGtvRIWkl0im7h0u2Q0/W0cdIR4CNy/BQL5bBMieCGpH0z5K2zXvLz5LO+77SxWQdpgLfy190JI2StFseN5R02P44aeP+/YZpHwE2aBg2B/hUPprYmXROvkfzl/QBSZtJGgQsJW0wmi5XRCwGriJdmL0vIu7Mo1YhnftfDCyTtAvwoS5iKnoLaQO3OMe0P2lPuEsRsQi4BPilpLUkDZb0/jz6FOCA/LlJ0lsk7SppaJOq/gSsI+lQSatKGipp2zzuLOBbud1Gkq4H/LYby9doa0kfy3f9HEr6/G+iF+2QyzddR/PRyrmkdWBoXg8O6+UyWOZEUC/DSBuWJ0mH1Y+T9oDLOAGYAVwu6WnSl75jI3Naru8h0l0iNzVM+2tg43yq4II87BDgo6Q7YT5NOh/f0/mvQ7qLZSnplNHVtN5AnEk6ennttFBEPA18jbSxeZJ0lDOji5heExF3AD8GbiQlvs1Id8eU9VlSAvsb6eLwobnemaQLoCfluBaQzoc3i+Fp0nWPj5JOk/wd+EAefRwwk3QXzjzSnT7HdSO+RheSLgQ/mWP/WD4y6207tFpHDyYlh3tJ13rOBE7txTJY5h+UmVm3SDqWdKH/M30di60YPiIwM6s5JwIzs5rzqSEzs5rzEYGZWc31uwc+jRw5MsaPH9/XYZiZ9SuzZs16LCJGNRvX7xLB+PHjmTlzZl+HYWbWr0h6oLNxPjVkZlZzTgRmZjXnRGBmVnNOBGZmNedEYGZWc04EZmY150RgZlZzTgRmZjXnRGBmVnP97pfFZv3Z+CMu6usQ+tT9x+/a1yFYEz4iMDOrOScCM7OaqywRSBoi6RZJt0m6XdK3m5SRpBMlLZA0V9JWVcVjZmbNVXmN4EVgx4h4RtJg4DpJl0RE8Y/NdwEm5Ne2wMm8/ofkZmbWBpUdEUTyTO4dnF+Nf4e2G3BaLnsTMFzS6KpiMjOz5VV6jUDSIElzgEeBKyLi5oYiY4AHC/0L87DGeqZImilp5uLFiyuL18ysjipNBBHxSkRsAawHbCNp04YiajZZk3qmRcTEiJg4alTTP9gxM7MeastdQxHxFHAVsHPDqIXA2EL/esDD7YjJzMySKu8aGiVpeO5eDdgJ+FtDsRnAvvnuoe2AJRGxqKqYzMxseVXeNTQa+I2kQaSEc25E/EnSAQARMRW4GJgMLACeA/avMB4zM2uiskQQEXOBLZsMn1roDuCrVcVgZmZd8y+LzcxqzonAzKzmnAjMzGrOicDMrOacCMzMas6JwMys5pwIzMxqzonAzKzmnAjMzGrOicDMrOacCMzMas6JwMys5pwIzMxqzonAzKzmnAjMzGrOicDMrOacCMzMas6JwMys5pwIzMxqzonAzKzmnAjMzGrOicDMrOacCMzMas6JwMys5ipLBJLGSrpS0p2Sbpd0SJMykyQtkTQnv46uKh4zM2tu5QrrXgZ8PSJmSxoKzJJ0RUTc0VDu2oj4SIVxmJlZC5UdEUTEooiYnbufBu4ExlQ1PzMz65m2XCOQNB7YEri5yejtJd0m6RJJm7QjHjMze12Vp4YAkLQG8Dvg0IhY2jB6NrB+RDwjaTJwATChSR1TgCkA48aNqzZgM7OaqfSIQNJgUhI4IyJ+3zg+IpZGxDO5+2JgsKSRTcpNi4iJETFx1KhRVYZsZlY7Vd41JODXwJ0R8ZNOyqyTyyFpmxzP41XFZGZmy6vy1NB7gM8C8yTNycOOBMYBRMRUYE/gQEnLgOeBvSMiKozJzMwaVJYIIuI6QF2UOQk4qaoYzMysa/5lsZlZzTkRmJnVnBOBmVnNORGYmdWcE4GZWc05EZiZ1ZwTgZlZzTkRmJnVnBOBmVnNORGYmdWcE4GZWc05EZiZ1ZwTgZlZzXWZCCStLukoSafk/gmS/GfzZmYDRJkjgv8BXgS2z/0LgeMqi8jMzNqqTCLYMCJ+CLwMEBHP08X/DJiZWf9RJhG8JGk1IAAkbUg6QjAzswGgzD+UHQNcCoyVdAbpLyj3qzIoMzNrny4TQURcIWk2sB3plNAhEfFY5ZGZmVlblLlraA9gWURcFBF/ApZJ2r3yyMzMrC3KXCM4JiKWdPRExFOk00VmZjYAlEkEzcqUubZgZmb9QJlEMFPSTyRtKGkDST8FZlUdmJmZtUeZRHAw8BJwDnAe8ALw1SqDMjOz9ilz19CzwBFtiMXMzPpAl4lA0juAw4HxxfIRsWMX040FTgPWAV4FpkXECQ1lBJwATAaeA/aLiNndWwQzM+uNMhd9zwOmAr8CXulG3cuAr0fEbElDgVmSroiIOwpldgEm5Ne2wMn53czM2qRMIlgWESd3t+KIWAQsyt1PS7oTGAMUE8FuwGkREcBNkoZLGp2nNTOzNihzsfiPkr4iabSkER2v7sxE0nhgS+DmhlFjgAcL/QvzsMbpp0iaKWnm4sWLuzNrMzPrQpkjgs/l928UhgWwQZkZSFoD+B1waEQsbRzdZJJYbkDENGAawMSJE5cbb2ZmPVfmrqG397RySYNJSeCMiPh9kyILgbGF/vWAh3s6PzMz675SvxCWtCmwMTCkY1hEnNbFNAJ+DdwZET/ppNgM4CBJZ5MuEi/x9QEzs/Yqc/voMcAkUiK4mHSnz3WkW0NbeQ/wWWCepDl52JHAOICImJrrmwwsIN0+un93F8DMzHqnzBHBnsDmwF8jYn9Ja5NuJW0pIq6ji38yy3cL+VfKZmZ9qMxdQ89HxKukx08PAx6l5IViMzN78ytzRDBT0nDgFNLD5p4BbqkyKDMza58ydw19JXdOlXQpMCwi5lYblpmZtUuniUDSP0XE3yRt1WTcVn4mkJnZwNDqiOAwYArw4ybjAmj50DkzM+sfOk0EETFF0krAtyLi+jbGZGZmbdTyrqF8t9CP2hSLmZn1gTK3j14u6eP5l8JmZjbAlLl99DDgLaTfEbxA+pFYRMSwSiMzM7O2KHP76NB2BGJmZn2j7EPn1iL9i1jxoXPXVBWUmZm1T5mHzn0ROIT0iOg5wHbAjfj2UTOzAaHMxeJDgH8GHoiID5D+acx/E2ZmNkCUSQQvRMQLAJJWjYi/Ae+sNiwzM2uXMtcIFuaHzl0AXCHpSfwvYmZmA0arZw0dDpwTEXvkQcdKuhJYE7i0HcGZmVn1Wh0RjAFukHQfcBZwXkRc3Z6wzMysXTq9RhAR/0b6W8mjgHcBcyVdImlfSf5tgZnZANHVs4YiIq6OiAOBscDPgH8DHmlDbGZm1gZlf1C2GbA3sBfwOOlP6M3MbABodbF4Amnjvw/wCnA28KGIuLdNsZmZWRu0OiK4jHSReK+ImNemeMzMrM1a/THNBu0MxMzM+kaZXxabmdkAVlkikHSqpEclze9k/CRJSyTNya+jq4rFzMw6VyoRSFpNUnefLzQd2LmLMtdGxBb59Z1u1m9mZitAl4lA0kdJj5++NPdvIWlGV9Pl/yt4orcBmplZtcocERwLbAM8BRARc4DxK2j+20u6Lf9ieZPOCkmaImmmpJmLF/sJ2GZmK1KZRLAsIpZUMO/ZwPoRsTnwc9LTTZuKiGkRMTEiJo4aNaqCUMzM6qtMIpgv6VPAIEkTJP0cuKG3M46IpRHxTO6+GBgsaWRv6zUzs+4pkwgOBjYBXgTOBJYAh/Z2xpLWkaTcvU2O5fHe1mtmZt3T8llDkgYBMyJiJ+A/ulOxpLOAScBISQuBY4DBABExFdgTOFDSMuB5YO+IiG4vgZmZ9UrLRBARr0h6TtKa3b1OEBH7dDH+JOCk7tRpZmYrXpmnj74AzJN0BfBsx8CI+FplUZmZWduUSQQX5ZeZmQ1AXSaCiPhNOwIxM7O+0WUiyP9ZvNxFXD+d1MxsYChzamhioXsI8AlgRDXhmJlZu3X5O4KIeLzweigifgbsWH1oZmbWDmVODW1V6F2JdIQwtLKIzMysrcqcGvpxoXsZcB/wyWrCMTOzdiuTCL7Q+If1kt5eUTxmZtZmZZ41dH7JYWZm1g91ekQg6Z9ID5tbU9LHCqOGke4eMjOzAaDVqaF3Ah8BhgMfLQx/GvhShTGZmVkbdZoIIuJC4EJJ20fEjW2MyczM2qjMxeK/Svoq6TTRa6eEIuLzlUVlZmZtU+Zi8enAOsCHgauB9Uinh8zMbAAokwg2ioijgGfzA+h2BTarNiwzM2uXMong5fz+lKRNgTWB8ZVFZGZmbVXmGsE0SWsBRwEzgDWAoyuNyszM2qbM/xH8KndeDfjR02ZmA0yXp4YkrS3p15Iuyf0bS/pC9aGZmVk7lLlGMB24DFg3998NHFpRPGZm1mZlEsHIiDgXeBUgIpYBr1QalZmZtU2ZRPCspLeS/65S0nbAkkqjMjOztilz19BhpLuFNpR0PTAK2LPSqMzMrG1aPX10XET8IyJmS9qB9BA6AXdFxMudTWdmZv1Lq1NDFxS6z4mI2yNiftkkIOlUSY9Kmt/JeEk6UdICSXMb/hLTzMzapFUiUKG7J78fmA7s3GL8LsCE/JoCnNyDeZiZWS+1SgTRSXcpEXEN8ESLIrsBp0VyEzBc0ujuzsfMzHqn1cXizSUtJR0ZrJa7yf0REcN6Oe8xwIOF/oV52KLGgpKmkI4aGDduXC9na2ZmRa3+mGZQxfNWk2FNjzwiYhowDWDixIndPjoxM7POlfkdQVUWAmML/esBD/dRLGZmtdWXiWAGsG++e2g7YElELHdayMzMqlXmB2U9IuksYBIwUtJC4BhgMEBETAUuBiYDC4DngP2risXMzDpXWSKIiH26GB/AV6uav5mZldOXp4bMzOxNwInAzKzmnAjMzGrOicDMrOacCMzMas6JwMys5pwIzMxqzonAzKzmnAjMzGrOicDMrOacCMzMas6JwMys5pwIzMxqzonAzKzmnAjMzGqusv8jeDMaf8RFfR1Cn7r/+F37OgQzexPyEYGZWc05EZiZ1ZwTgZlZzTkRmJnVnBOBmVnNORGYmdVcrW4fNbP+zbeAV3MLuI8IzMxqrtJEIGlnSXdJWiDpiCbjJ0laImlOfh1dZTxmZra8yk4NSRoE/AL4ILAQuFXSjIi4o6HotRHxkariMDOz1qo8ItgGWBAR90bES8DZwG4Vzs/MzHqgykQwBniw0L8wD2u0vaTbJF0iaZNmFUmaImmmpJmLFy+uIlYzs9qqMhGoybBo6J8NrB8RmwM/By5oVlFETIuIiRExcdSoUSs2SjOzmqsyESwExhb61wMeLhaIiKUR8UzuvhgYLGlkhTGZmVmDKhPBrcAESW+XtAqwNzCjWEDSOpKUu7fJ8TxeYUxmZtagsruGImKZpIOAy4BBwKkRcbukA/L4qcCewIGSlgHPA3tHROPpIzMzq1ClvyzOp3subhg2tdB9EnBSlTGYmVlr/mWxmVnNORGYmdWcE4GZWc05EZiZ1ZwTgZlZzTkRmJnVnBOBmVnNORGYmdWcE4GZWc05EZiZ1ZwTgZlZzTkRmJnVnBOBmVnNORGYmdWcE4GZWc1V+n8ENvCMP+Kivg6hT91//K59HYLZCucjAjOzmnMiMDOrOScCM7OacyIwM6s5JwIzs5pzIjAzqzknAjOzmnMiMDOrOScCM7OaqzQRSNpZ0l2SFkg6osl4SToxj58raasq4zEzs+VVlggkDQJ+AewCbAzsI2njhmK7ABPyawpwclXxmJlZc1UeEWwDLIiIeyPiJeBsYLeGMrsBp0VyEzBc0ugKYzIzswZVPnRuDPBgoX8hsG2JMmOARcVCkqaQjhgAnpF014oNtW1GAo/11cz1g76a8wrlNuwdt1/v9Of2W7+zEVUmAjUZFj0oQ0RMA6atiKD6kqSZETGxr+Poz9yGveP2652B2n5VnhpaCIwt9K8HPNyDMmZmVqEqE8GtwARJb5e0CrA3MKOhzAxg33z30HbAkohY1FiRmZlVp7JTQxGxTNJBwGXAIODUiLhd0gF5/FTgYmAysAB4Dti/qnjeJPr96a03Abdh77j9emdAtp8iljslb2ZmNeJfFpuZ1ZwTgZlZzTkRdEHSJEl/Kjt8Bcxv9+IvsCVdJanL29UkjV4R8UgaJenS3tbTpN4etZekdSWd38m419pG0pGF4eMlzS9Z/6GS9u1uXE3qOUhSqWtckvaTtG6JctMl7Vl2eG/1pzYsTNOrtiwx3QHNlq3YPpK2kDS5MO5YSYeXqFuS/lfSsO7G1aSuP0taq6fTOxG8+exOeiRHdx0GnNLbmUfEYmCRpPf0tq4VISIejogyX+Ajuy7yRpJWBj4PnNntwJZ3KvC1kmX3A7rcePWB/tSGHfajwraMiKkRcVoXxbYg3fTSXZOB2yJiaQ+mbXQ68JWeTtzvE4Gkt0i6SNJtkuZL2isP31rS1ZJmSbqs49EVeS/yZ5JuyOW3ycO3ycP+mt/f2c0YTpV0a55+tzx8P0m/l3SppL9L+mFhmi9IujvHc4qkkyS9G/hX4L8kzZG0YS7+CUm35PLv6ySMjwOX5roHSfqRpHn5YX4H5+H3S/q+pBslzZS0VW6beyR9raMdgR2BY9rRjpIulvSu3P1XSUfn7u9K+mLDntdqks7Oy3QOsFoefjywWm6zM3LVg3K73i7pckmrNZn9jsDsiFiW69ko71ndJmm2pA2VjmSulnRubv/jJX06fx7zJL1f0t9Iz8naQNJfJK3eWdsp7ZVOBM7I8a4m6ei87syXNE1Ssx9adtZ+rT6fHzSuN5JWz8syV9I5km6WNLGP27BjPX8bsFFeD+dKOr/KtpT0NkmzcvfmkkLSuNx/T26r1/bucwy3SboR+GoetgrwHWCvHMNeufqN82dwr6TOktungQsL8eybl/s2SafnYdMlnSzpylzXDkrbmjslTS/UNQPYp7Nl7VJE9OsXaQN4SqF/TWAwcAMwKg/bi3T7KsBVHeWB9wPzc/cwYOXcvRPwu9w9CfhTk/m+Nhz4PvCZ3D0cuBt4C2lv5d4c0xDgAdIP6NYF7gdG5FivBU7K008H9izM5yrgx7l7MvDnJrG8HZhV6D8Q+F1heUbk9/uBA3P3T4G5wFBgFLCk0C5jgNvb1I5HkL5Uw0i/PbksD78SeCcwvlD3YYX5vwtYBkzM/c8U6hyfx22R+8/t+Hwa5v1t4OBC/83AHrl7CLB6jvspYDSwKvAQ8O1c5hDSXmwA7wH+A7gFOLxE200szHdEoft04KPN1oVCmenAniXmsdx6k2P779y96ZukDX9WmGcAP8/9p7ahLW8nrXsHkda/T5MexXBjHn8scHjungvskLv/i9fXy/3I39/CNDfkZR0JPA4MbjLvB4ChuXsT4C5gZMN3djrpOW0iPZttKbAZaSd+Vsfnk8v+HXhrT7ajVT5iol3mAT+S9APShuZaSZuSVvIr8g7BIN74/KKzACLiGknDJA0nbRB/I2kCaWUc3I0YPgT8q14/LzgEGJe7/xIRSwAk3UFayUYCV0fEE3n4ecA7WtT/+/w+i/RlaTQaWFzo3wmYGnkvrWM+WceP+uYBa0TE08DTkp4FPpTb8RJgHdKGuOp2vJZ0OuA+4CLgg3kvcHxE3CWpuLzvB07M85wraW6Leu+LiDm5u1W73QkgaSgwJiL+kOt/IQ8HuDXyDx0l3QNcnqefB+wKPBgR1ytd27kfeC/p6KxV2xV9QNI3SRvNEaSN0x9bLFuHrj6fZuvNe4ET8jLOf5O04QcK9T0BvJS7f0taN6psyxtISfz9pB26nUkb3WuLhSStCQyPiKvzoNNJT0/uzEUR8SLwoqRHgbVJT1IoGpG/f5COrM6PiMdgue/sHyMiJM0DHomIeTmm20mfyZxc7lHSTubjLeJqqt8ngoi4W9LWpL2e/5R0OfAH4PaI2L6zyZr0fxe4MiL2yBufq7oRhoCPR8QbHoYnaVvgxcKgV0htXvrQP+uoo2P6Rs+Tkk8xns5+INJR16sNsb1ESiDbAf9ZiLPqdryVdHh/L3AFKUl+ibThKTPPzjS2e7PTGsV2a/WZFOsqtturpI1SR0xD8rjBdN12aabSEOCXpL3aByUdyxs/y5aTdzGPZutNd9a9drVhcZ1WrrNDUG1bXgu8j7SDdiHw73mejTc2tPpONdPse99omaSVIuLVLurv7Dvb2HZDeGPblTYQrhGsCzwXEb8FfgRsRTrEGiVp+1xmsKRNCpN1XEd4L+mxFktIp28eyuP362YYlwEHd5yPlLRlF+VvAXaQtJbSxbaPF8Y9Tdqr7o67eePe2uXAAbluJI0oUccg4PncjucDL9CGdoz0iPIHgU8CN5G+mIfTsEeWXUM6dCcf9b2rMO5lSd05ioO0J7tRjmMpsFDS7rn+VTvOT5cwLrfRO0h7ZNfRuu2Kn3HHhuoxSWuQTvmU1dXn08x1pLYmH8FsVhjXl23YYS1e39jtQ/VteQ3wGeDveYP8BGmn8vpioYh4CliS13XI62GTGLrjLmCD3P0X4JOS3gqlv7OvyduedUhHpN3W7xMBaUW+RdIc0jna4/LGZU/gB0oXP+cA7y5M86SkG4CpwBfysB+SjiiuJ20Uu+O7pL3AuUoXNr/bqnBEPEQ6DL0Z+DNwB+kcPaTzgd9QunC6YSdVNNb3LHCPpI3yoF8B/8jx3AZ8qkQ1g4HLczseCvymje14LemQ97ncvR7NE8HJwBr5dMY3SQm1wzTS8p7RZLrOXEI6JdDhs8DXcv03kL5YZdwJfA74IulRKSd30XbTgam5rV8k3e01D7iAdIRUSonPp5lfkjaqc0l7v3N5fd3ryzbs8AKwYZ5+BBW3ZUTcnzuvye/XAU9FxJNNiu8P/ELpYnFxz/tK0sXh4sXiMi4iXT8hIm4HvgdcnZfxJ92oB2Br4KaO08Hd1pMLC/35RcPFpT6MY438vjLpHOYevaxvD1ISXBGxXQOs1R/acQUs6x+ACb2YfjwwH9gSOL2vl6dEvIOAIbl7Q9Ie5Cp92YaFeiaTNsJ93k5t+ixGA1esoLpOAP6lp9P3+2sE/dixknYiHc5eTtqD6bGI+EPHYWVvSBoF/CSa7xENREeQvpB/72U9I4Gjeh9O5VYHrsyngES6i+ylLqbpyopqwxHAI72so9+IiEVKt+cOi97/lmB+RPylpxP7oXNmZjU3EK4RmJlZLzgRmJnVnBOBmVnNORGYmdWcE4GZWc39f6FiqUA1f/V0AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#%% create a barplot visual based on the variances\n",
    "plt.bar(x=iris.feature_names, height=selector.variances_)\n",
    "plt.ylabel('Feature Variance')\n",
    "plt.title('Iris features variance comparison')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "0995e897",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "* sepal length (cm)\n",
      "* petal length (cm)\n"
     ]
    }
   ],
   "source": [
    "#%% Select features with variance greater than 0.6 and display the names of selected features. \n",
    "#The method get_support()\n",
    "for selected_feature in selector.get_support(indices=True):\n",
    "    print('* ' + feature_names[selected_feature])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1b0e31e3",
   "metadata": {},
   "source": [
    "## 3 Correlation matrix"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "27f2fffa",
   "metadata": {},
   "source": [
    "### 3.1 Pearson’s Correlation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "650e403e",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% load libraries\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.datasets import load_iris\n",
    "from matplotlib import pyplot as plt\n",
    "import seaborn as sns\n",
    "from scipy import stats"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "80a9dd41",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% load sample data\n",
    "iris = load_iris()\n",
    "X = pd.DataFrame(iris.data, \n",
    "    columns=iris.feature_names)\n",
    "y = pd.DataFrame(iris.target, \n",
    "    columns=['Response_variable'])\n",
    "XY = pd.concat([X, y], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "4c9e1174",
   "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>sepal length (cm)</th>\n",
       "      <th>sepal width (cm)</th>\n",
       "      <th>petal length (cm)</th>\n",
       "      <th>petal width (cm)</th>\n",
       "      <th>Response_variable</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>sepal length (cm)</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.117570</td>\n",
       "      <td>0.871754</td>\n",
       "      <td>0.817941</td>\n",
       "      <td>0.782561</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sepal width (cm)</th>\n",
       "      <td>-0.117570</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.428440</td>\n",
       "      <td>-0.366126</td>\n",
       "      <td>-0.426658</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>petal length (cm)</th>\n",
       "      <td>0.871754</td>\n",
       "      <td>-0.428440</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.962865</td>\n",
       "      <td>0.949035</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>petal width (cm)</th>\n",
       "      <td>0.817941</td>\n",
       "      <td>-0.366126</td>\n",
       "      <td>0.962865</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.956547</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Response_variable</th>\n",
       "      <td>0.782561</td>\n",
       "      <td>-0.426658</td>\n",
       "      <td>0.949035</td>\n",
       "      <td>0.956547</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                   sepal length (cm)  sepal width (cm)  petal length (cm)  \\\n",
       "sepal length (cm)           1.000000         -0.117570           0.871754   \n",
       "sepal width (cm)           -0.117570          1.000000          -0.428440   \n",
       "petal length (cm)           0.871754         -0.428440           1.000000   \n",
       "petal width (cm)            0.817941         -0.366126           0.962865   \n",
       "Response_variable           0.782561         -0.426658           0.949035   \n",
       "\n",
       "                   petal width (cm)  Response_variable  \n",
       "sepal length (cm)          0.817941           0.782561  \n",
       "sepal width (cm)          -0.366126          -0.426658  \n",
       "petal length (cm)          0.962865           0.949035  \n",
       "petal width (cm)           1.000000           0.956547  \n",
       "Response_variable          0.956547           1.000000  "
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#%% calcualte the correlation matrix\n",
    "cor_mat = XY.corr(method='pearson')\n",
    "cor_mat"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "6e878b28",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#%% correlation matrix as heatmap\n",
    "ax = sns.heatmap(cor_mat, vmin=-1, vmax=1, \n",
    "    annot=True, fmt=\"f\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2a4da663",
   "metadata": {},
   "source": [
    "### 3.2 Spearman correlation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "fff0ffb7",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% generate sampe data\n",
    "dat = pd.DataFrame({'x': np.arange(0,10), \n",
    "    'y': np.exp(-np.arange(0,10))})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "eb1011b7",
   "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>x</th>\n",
       "      <th>y</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>x</th>\n",
       "      <td>1.00000</td>\n",
       "      <td>-0.71687</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>y</th>\n",
       "      <td>-0.71687</td>\n",
       "      <td>1.00000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         x        y\n",
       "x  1.00000 -0.71687\n",
       "y -0.71687  1.00000"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#%%\n",
    "# compute Pearson’s correlation and\n",
    "# display the correlation matrix\n",
    "dat.corr(method='pearson')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "830a6d8c",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\scipy\\stats\\stats.py:1604: UserWarning: kurtosistest only valid for n>=20 ... continuing anyway, n=10\n",
      "  \"anyway, n=%i\" % int(n))\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "NormaltestResult(statistic=2.02697581498966, pvalue=0.362950830342156)"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#%% test for normal distribution of the data\n",
    "stats.normaltest(dat['x'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "f21f3ddd",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "NormaltestResult(statistic=19.779358749097575, pvalue=5.0695197559354735e-05)"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "stats.normaltest(dat['y'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "aa58c38e",
   "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>x</th>\n",
       "      <th>y</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>x</th>\n",
       "      <td>1.0</td>\n",
       "      <td>-1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>y</th>\n",
       "      <td>-1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     x    y\n",
       "x  1.0 -1.0\n",
       "y -1.0  1.0"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#%%\n",
    "# compute Spearman’s rank correlation and\n",
    "# display the correlation matrix\n",
    "dat.corr(method='spearman')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "acbac87a",
   "metadata": {},
   "source": [
    "## 4 Recursive Feature Selection"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5626a183",
   "metadata": {},
   "source": [
    "### 4.1 Exclusive Feature Selection (EFS)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "1c8b6e0f",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Recursive feature selection\n",
    "# Exclusive Feature Selection (EFS)\n",
    "\n",
    "#%% import libraries\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.datasets import load_iris\n",
    "from mlxtend.feature_selection \\\n",
    "    import ExhaustiveFeatureSelector as EFS\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "c22e9156",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% load sample data\n",
    "iris = load_iris()\n",
    "x = pd.DataFrame(iris.data, \n",
    "    columns=iris.feature_names)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "33d96b8f",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% create a logistic regression object\n",
    "lr = LogisticRegression()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "fc92da8f",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% create an EFS object, cross-validations = 5\n",
    "efs = EFS(estimator=lr,        \n",
    "          min_features=1,      \n",
    "          max_features=3,      \n",
    "          scoring='accuracy',  \n",
    "          cv=5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "8316583a",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Features: 14/14"
     ]
    }
   ],
   "source": [
    "#%% fit the model\n",
    "efs = efs.fit(x, iris.target)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "d0186ec0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "('sepal length (cm)', 'petal length (cm)', 'petal width (cm)')"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#%% show the selected features\n",
    "efs.best_feature_names_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "d591e374",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% show a full report on the feature selection\n",
    "efs_results = pd.DataFrame(efs.get_metric_dict()).T.sort_values(by='avg_score', ascending=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "ad36c772",
   "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>feature_idx</th>\n",
       "      <th>cv_scores</th>\n",
       "      <th>avg_score</th>\n",
       "      <th>feature_names</th>\n",
       "      <th>ci_bound</th>\n",
       "      <th>std_dev</th>\n",
       "      <th>std_err</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>(0, 2, 3)</td>\n",
       "      <td>[0.9666666666666667, 0.9666666666666667, 0.933...</td>\n",
       "      <td>0.966667</td>\n",
       "      <td>(sepal length (cm), petal length (cm), petal w...</td>\n",
       "      <td>0.027096</td>\n",
       "      <td>0.021082</td>\n",
       "      <td>0.010541</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>(1, 2, 3)</td>\n",
       "      <td>[0.9666666666666667, 0.9666666666666667, 0.933...</td>\n",
       "      <td>0.966667</td>\n",
       "      <td>(sepal width (cm), petal length (cm), petal wi...</td>\n",
       "      <td>0.027096</td>\n",
       "      <td>0.021082</td>\n",
       "      <td>0.010541</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>(3,)</td>\n",
       "      <td>[1.0, 0.9666666666666667, 0.9, 0.9333333333333...</td>\n",
       "      <td>0.96</td>\n",
       "      <td>(petal width (cm),)</td>\n",
       "      <td>0.049963</td>\n",
       "      <td>0.038873</td>\n",
       "      <td>0.019437</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>(2, 3)</td>\n",
       "      <td>[0.9666666666666667, 0.9666666666666667, 0.933...</td>\n",
       "      <td>0.96</td>\n",
       "      <td>(petal length (cm), petal width (cm))</td>\n",
       "      <td>0.032061</td>\n",
       "      <td>0.024944</td>\n",
       "      <td>0.012472</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>(2,)</td>\n",
       "      <td>[0.9333333333333333, 1.0, 0.9, 0.9333333333333...</td>\n",
       "      <td>0.953333</td>\n",
       "      <td>(petal length (cm),)</td>\n",
       "      <td>0.051412</td>\n",
       "      <td>0.04</td>\n",
       "      <td>0.02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>(0, 2)</td>\n",
       "      <td>[0.9333333333333333, 1.0, 0.9, 0.9333333333333...</td>\n",
       "      <td>0.953333</td>\n",
       "      <td>(sepal length (cm), petal length (cm))</td>\n",
       "      <td>0.051412</td>\n",
       "      <td>0.04</td>\n",
       "      <td>0.02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>(0, 3)</td>\n",
       "      <td>[0.9333333333333333, 0.9666666666666667, 0.933...</td>\n",
       "      <td>0.953333</td>\n",
       "      <td>(sepal length (cm), petal width (cm))</td>\n",
       "      <td>0.034274</td>\n",
       "      <td>0.026667</td>\n",
       "      <td>0.013333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>(1, 2)</td>\n",
       "      <td>[0.9333333333333333, 1.0, 0.9, 0.9333333333333...</td>\n",
       "      <td>0.953333</td>\n",
       "      <td>(sepal width (cm), petal length (cm))</td>\n",
       "      <td>0.051412</td>\n",
       "      <td>0.04</td>\n",
       "      <td>0.02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>(0, 1, 2)</td>\n",
       "      <td>[0.9333333333333333, 1.0, 0.9, 0.9333333333333...</td>\n",
       "      <td>0.953333</td>\n",
       "      <td>(sepal length (cm), sepal width (cm), petal le...</td>\n",
       "      <td>0.051412</td>\n",
       "      <td>0.04</td>\n",
       "      <td>0.02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>(0, 1, 3)</td>\n",
       "      <td>[0.9, 0.9666666666666667, 0.9333333333333333, ...</td>\n",
       "      <td>0.946667</td>\n",
       "      <td>(sepal length (cm), sepal width (cm), petal wi...</td>\n",
       "      <td>0.043691</td>\n",
       "      <td>0.033993</td>\n",
       "      <td>0.016997</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>(1, 3)</td>\n",
       "      <td>[0.9333333333333333, 0.9666666666666667, 0.9, ...</td>\n",
       "      <td>0.94</td>\n",
       "      <td>(sepal width (cm), petal width (cm))</td>\n",
       "      <td>0.032061</td>\n",
       "      <td>0.024944</td>\n",
       "      <td>0.012472</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>(0, 1)</td>\n",
       "      <td>[0.7333333333333333, 0.8333333333333334, 0.766...</td>\n",
       "      <td>0.813333</td>\n",
       "      <td>(sepal length (cm), sepal width (cm))</td>\n",
       "      <td>0.069612</td>\n",
       "      <td>0.05416</td>\n",
       "      <td>0.02708</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>(0,)</td>\n",
       "      <td>[0.6666666666666666, 0.7333333333333333, 0.766...</td>\n",
       "      <td>0.753333</td>\n",
       "      <td>(sepal length (cm),)</td>\n",
       "      <td>0.069612</td>\n",
       "      <td>0.05416</td>\n",
       "      <td>0.02708</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>(1,)</td>\n",
       "      <td>[0.5333333333333333, 0.5666666666666667, 0.533...</td>\n",
       "      <td>0.566667</td>\n",
       "      <td>(sepal width (cm),)</td>\n",
       "      <td>0.046932</td>\n",
       "      <td>0.036515</td>\n",
       "      <td>0.018257</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   feature_idx                                          cv_scores avg_score  \\\n",
       "12   (0, 2, 3)  [0.9666666666666667, 0.9666666666666667, 0.933...  0.966667   \n",
       "13   (1, 2, 3)  [0.9666666666666667, 0.9666666666666667, 0.933...  0.966667   \n",
       "3         (3,)  [1.0, 0.9666666666666667, 0.9, 0.9333333333333...      0.96   \n",
       "9       (2, 3)  [0.9666666666666667, 0.9666666666666667, 0.933...      0.96   \n",
       "2         (2,)  [0.9333333333333333, 1.0, 0.9, 0.9333333333333...  0.953333   \n",
       "5       (0, 2)  [0.9333333333333333, 1.0, 0.9, 0.9333333333333...  0.953333   \n",
       "6       (0, 3)  [0.9333333333333333, 0.9666666666666667, 0.933...  0.953333   \n",
       "7       (1, 2)  [0.9333333333333333, 1.0, 0.9, 0.9333333333333...  0.953333   \n",
       "10   (0, 1, 2)  [0.9333333333333333, 1.0, 0.9, 0.9333333333333...  0.953333   \n",
       "11   (0, 1, 3)  [0.9, 0.9666666666666667, 0.9333333333333333, ...  0.946667   \n",
       "8       (1, 3)  [0.9333333333333333, 0.9666666666666667, 0.9, ...      0.94   \n",
       "4       (0, 1)  [0.7333333333333333, 0.8333333333333334, 0.766...  0.813333   \n",
       "0         (0,)  [0.6666666666666666, 0.7333333333333333, 0.766...  0.753333   \n",
       "1         (1,)  [0.5333333333333333, 0.5666666666666667, 0.533...  0.566667   \n",
       "\n",
       "                                        feature_names  ci_bound   std_dev  \\\n",
       "12  (sepal length (cm), petal length (cm), petal w...  0.027096  0.021082   \n",
       "13  (sepal width (cm), petal length (cm), petal wi...  0.027096  0.021082   \n",
       "3                                 (petal width (cm),)  0.049963  0.038873   \n",
       "9               (petal length (cm), petal width (cm))  0.032061  0.024944   \n",
       "2                                (petal length (cm),)  0.051412      0.04   \n",
       "5              (sepal length (cm), petal length (cm))  0.051412      0.04   \n",
       "6               (sepal length (cm), petal width (cm))  0.034274  0.026667   \n",
       "7               (sepal width (cm), petal length (cm))  0.051412      0.04   \n",
       "10  (sepal length (cm), sepal width (cm), petal le...  0.051412      0.04   \n",
       "11  (sepal length (cm), sepal width (cm), petal wi...  0.043691  0.033993   \n",
       "8                (sepal width (cm), petal width (cm))  0.032061  0.024944   \n",
       "4               (sepal length (cm), sepal width (cm))  0.069612   0.05416   \n",
       "0                                (sepal length (cm),)  0.069612   0.05416   \n",
       "1                                 (sepal width (cm),)  0.046932  0.036515   \n",
       "\n",
       "     std_err  \n",
       "12  0.010541  \n",
       "13  0.010541  \n",
       "3   0.019437  \n",
       "9   0.012472  \n",
       "2       0.02  \n",
       "5       0.02  \n",
       "6   0.013333  \n",
       "7       0.02  \n",
       "10      0.02  \n",
       "11  0.016997  \n",
       "8   0.012472  \n",
       "4    0.02708  \n",
       "0    0.02708  \n",
       "1   0.018257  "
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "efs_results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "00b1193f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#%% show feature importance visually\n",
    "# create figure and axes\n",
    "fig, ax = plt.subplots()\n",
    "# plot bars\n",
    "y_pos = np.arange(len(efs_results))\n",
    "ax.barh(y_pos, efs_results['avg_score'], \n",
    "    xerr=efs_results['std_err'])\n",
    "# set axis ticks and labels\n",
    "ax.set_yticks(y_pos)\n",
    "ax.set_yticklabels(efs_results['feature_names'])\n",
    "ax.set_xlabel('Accuracy')\n",
    "\n",
    "# show the plot\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "403a0c8a",
   "metadata": {},
   "source": [
    "### 4.2 Sequential Forward Feature Selection (SFS)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "id": "c2495fda",
   "metadata": {},
   "outputs": [],
   "source": [
    "# import libraries\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.datasets import load_iris\n",
    "from mlxtend.feature_selection import SequentialFeatureSelector as SFS\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "395a81e3",
   "metadata": {},
   "outputs": [],
   "source": [
    "# load sample data\n",
    "iris = load_iris()\n",
    "x = pd.DataFrame(iris.data, columns=iris.feature_names)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "id": "646d782e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create a logistic regression model\n",
    "lr = LogisticRegression()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "id": "7f26abb2",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% create an SFS object\n",
    "sfs = SFS(estimator=lr,\n",
    "          k_features=(1, 3),  \n",
    "          forward=True,       \n",
    "          scoring='accuracy', \n",
    "          cv=5)   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "id": "f3e1efdf",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% fit the model\n",
    "sfs = sfs.fit(x, iris.target)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "id": "e44e1efa",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "('sepal length (cm)', 'petal length (cm)', 'petal width (cm)')"
      ]
     },
     "execution_count": 73,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#%% show the selected features\n",
    "sfs.k_feature_names_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "id": "6c1df8ef",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% show a full report on the feature selection\n",
    "sfs_results = pd.DataFrame(sfs.get_metric_dict()).T.sort_values(by='avg_score', ascending=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "id": "7a3d04eb",
   "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>feature_idx</th>\n",
       "      <th>cv_scores</th>\n",
       "      <th>avg_score</th>\n",
       "      <th>feature_names</th>\n",
       "      <th>ci_bound</th>\n",
       "      <th>std_dev</th>\n",
       "      <th>std_err</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>(0, 2, 3)</td>\n",
       "      <td>[0.9666666666666667, 0.9666666666666667, 0.933...</td>\n",
       "      <td>0.966667</td>\n",
       "      <td>(sepal length (cm), petal length (cm), petal w...</td>\n",
       "      <td>0.027096</td>\n",
       "      <td>0.021082</td>\n",
       "      <td>0.010541</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>(3,)</td>\n",
       "      <td>[1.0, 0.9666666666666667, 0.9, 0.9333333333333...</td>\n",
       "      <td>0.96</td>\n",
       "      <td>(petal width (cm),)</td>\n",
       "      <td>0.049963</td>\n",
       "      <td>0.038873</td>\n",
       "      <td>0.019437</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>(2, 3)</td>\n",
       "      <td>[0.9666666666666667, 0.9666666666666667, 0.933...</td>\n",
       "      <td>0.96</td>\n",
       "      <td>(petal length (cm), petal width (cm))</td>\n",
       "      <td>0.032061</td>\n",
       "      <td>0.024944</td>\n",
       "      <td>0.012472</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  feature_idx                                          cv_scores avg_score  \\\n",
       "3   (0, 2, 3)  [0.9666666666666667, 0.9666666666666667, 0.933...  0.966667   \n",
       "1        (3,)  [1.0, 0.9666666666666667, 0.9, 0.9333333333333...      0.96   \n",
       "2      (2, 3)  [0.9666666666666667, 0.9666666666666667, 0.933...      0.96   \n",
       "\n",
       "                                       feature_names  ci_bound   std_dev  \\\n",
       "3  (sepal length (cm), petal length (cm), petal w...  0.027096  0.021082   \n",
       "1                                (petal width (cm),)  0.049963  0.038873   \n",
       "2              (petal length (cm), petal width (cm))  0.032061  0.024944   \n",
       "\n",
       "    std_err  \n",
       "3  0.010541  \n",
       "1  0.019437  \n",
       "2  0.012472  "
      ]
     },
     "execution_count": 75,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sfs_results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "id": "fa2dd1af",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAApgAAAEGCAYAAADBteVnAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8QVMy6AAAACXBIWXMAAAsTAAALEwEAmpwYAAAh9ElEQVR4nO3de7gdVX3/8ffHRMK1KOAlIhpAEC8oyEUR1FCxrcUKIhYE+Ymo/KhFVIo+sbWK9iKt/lpp8Ya2YhXUqkARKwV5DBcVJFwDCKKSCgqC3BUQCN/fH7OO2TmcW3Lm5JyE9+t5znNmz6xZs9bM7OzPWTOzk6pCkiRJ6stjprsBkiRJWrMYMCVJktQrA6YkSZJ6ZcCUJElSrwyYkiRJ6tXs6W6AJPVlk002qXnz5k13MyRptXLxxRf/qqqe0GedBkxJa4x58+axaNGi6W6GJK1Wkvxv33V6iVySJEm9MmBKkiSpVwZMSZIk9cqAKUmSpF4ZMCVJktQrA6YkSZJ6ZcCUJElSrwyYkiRJ6pVftC5pjbH453cxb8E3p7sZWglLjtlzupsgqUeOYEqSJKlXBkxJkiT1yoApSZKkXhkwJUmS1CsDpiRJknplwJQkSVKvDJiSJEnqlQFTkiRJvTJgSpIkqVcGTEmSJPXKgClJkqReGTAlSZLUKwOmJEmSemXAlCRJUq8MmJIkSeqVAVOSJEm9MmBKkiSpVwZMSZIk9cqAKUmSpF4ZMCVJktQrA6YkSZJ6ZcCUJElSrwyYkiRJ6pUBU5IkSb0yYK5mkqyT5Jwks5LMS7JwEnX95QTLLUmyyUTnT0br0wEDrw9OctwE1/1aki16aMOXk2w18HrJBNZZmGTeSm5v7yTPnkC5o5MctRL1vzrJglGW/br9nrH7PclaSc5NMnuy25AkrRoGzNXPIcDJVbW0h7omFDBXsXnAAeMVGi7Jc4BZVfXTHtrwSeA9PdQzUXsD4wbMlVVVp1XVMeMUm8cM3e9V9QBwNrBfD9uQJK0Cjgisfg5kWRBYCtwO3YgT8BpgDrA5cFJVfbAtewNwBLAWcCHwNuDvgHWSXAZcVVUHJjkV2AxYGzi2qo6faKNG2kZVLW0jZMcCrwLuA/aqql8m2RI4EZgFfAs4sqrWB44BntXa9XngDuApSc4AtgROqaqRwt+BwH8NtOePgL9v9f+qql6e5Oi2b+YCWwNHAi8CXgn8HPiTqnoQOA84IcnsqnoIuHUCu+B2uuMxNCr4aWD31v79q+rW1uePA08A7gXeCmwEvBp4WZL3Aa8Ffh84tO3LHwMHVdW9I200ySzgurZvNmztmF9V5yY5D3gTsBuwY1UdnmRz4CS69/4ZA1XN9P1+KvBhunNmVA/ediM3nzTiYK1muPkXfGS6m6DVyMKFC6e7CRqHI5irkSRrAVtU1RKAqrqhqvYZKLIz3Qf+dsDrkuyY5Fl0Iz+7VtV2dCHowKpaANxXVdtV1YFt/UOqagdgR+CIJBtPsF0jbqMtXg+4oKqeD5xLF6qgC53HVtVOwC8GqlsAnNfa9c9t3nat/m2B/ZJsNkIzdgUubu15AvAZ4LVtu68bKLclsCewF/BF4DtVtS1d+N0ToKoepgt2z2+vdxpvH1TVPlV1w0CfL6mqFwDnAB9o848H3t728VHAJ6rqe8BpwLtbn39CN0K9U2v7D4E3j7HdpcCP6EZAd2v74CVJ5gBPraofD1vlWOCTrU83D8yf6fv9SmDE45Dk0CSLkiyqh/sY2JckTZYjmKuXTYA7x1h+VlXdBpDkZLrA8RCwA3BREoB1gFtGWf+IJK9p05sBWwG3TaBdLx9jGw8Ap7fpi4FXtOld6C4NQzei9tEx6j+7qu5q/boaeDpww7Ayc1k24vUi4Nyquh6gqm4fKPetqnowyWK6UbahUbzFdJeJh9wCPKW1eUU9DHylTX8RODnJ+sCLga+2fQTdaPNInpvkb4HHAesD/zPO9s4DXko3SvhhuhB/DnDRCGV3pRslBfgC8A9j1Dtj9nsbDX8gyQZVdc+wZcfThXfmzN2qnnzAeHcDaCZaeMye090EST0yYK5e7qO7fD2aGuF1gM9X1XvHqjjJfGAPYJequrc9PDTWtpZbfYxtPFhVQ+1aysqdc78dmB6tjsF9Ex65L5arq6oeTjLYtoeH1bt2q7MPRXe14M42wjueE4C9q+ryduvD/HHKnwccRhfM3g+8u61z7hjtmYiZtt/nAPdPoN2SpGnmJfLVSFXdAcxKMlrwe0WSjZKsQzc6+F26hyP2TfJEgLb86a38g0ke26Y3BO5o4XIbutGoiRprG6O5gGUjafsPzL8H2GAFtj3kh8Az2vT36e5p3HyoPStR39bAVcNnJjk7yabjrPsYYN82fQBwflXdDVyf5HWtniR5fiszvM8bADe1Y3Mg47uQbnT04aq6H7gM+L90wXO477Jsfw/WPdP3+8bAre1eTUnSDGfAXP2cSXfpeyTn0132vAz4elUtqqqrgfcBZya5AjiL7rImdJcVr0hyIt0ly9mtzN/QBcAJGWcbo3kncGSSH7Syd7X5VwAPJbk8ybsm2gbgm7SRvqq6le4hmZOTXM6yy9UTkuRJdPen3jRs/mPowtTtI664zG+A5yS5mO6BnQ+1+QcCb25tuorufkSALwPvTnJpexDor+lC41nANeO1t6p+S3fpeuiYnUcXFhePUPwdwJ8nuYjuj4ohM3a/N7sD/70i9UmSpk+WXanS6iDJ9nRPXB80bP7BtCeFp6VhKyjJunRhopLsD7y+qvYab70x6lsH+A7dg0aTetKjBay7q+rfhs1/Lt2DUEeOs/6v2xPxa7xVsd/bspOB91bVtWPVMWfuVjX3jR+bTDM0TZZ4D6Y0bZJcXFU79lmn92CuZqrq0iTfSTKrp+/CnC47AMele+LlTrrv91xpVXVfkg8AmwI/m2Tb7qQbCR6+jSvpvmJHzarY7+3bE04dL1xKkmYORzAlrTEcwVx9OYIpTZ+pGMH0HkxJkiT1yoApSZKkXhkwJUmS1CsDpiRJknplwJQkSVKvDJiSJEnqlQFTkiRJvTJgSpIkqVcGTEmSJPXKgClJkqReGTAlSZLUKwOmJEmSemXAlCRJUq8MmJIkSeqVAVOSJEm9MmBKkiSpVwZMSZIk9cqAKUmSpF4ZMCVJktQrA6YkSZJ6ZcCUJElSr2ZPdwMkqS/bbrohi47Zc7qbIUmPeo5gSpIkqVcGTEmSJPXKgClJkqReGTAlSZLUKwOmJEmSemXAlCRJUq8MmJIkSeqVAVOSJEm9MmBKkiSpVwZMSZIk9cqAKUmSpF4ZMCVJktQrA6YkSZJ6NXu6GyBJfVn887uYt+Cb090MrYQlx+w53U2Q1CNHMCVJktQrA6YkSZJ6ZcCUJElSrwyYkiRJ6pUBU5IkSb0yYEqSJKlXBkxJkiT1yoApSZKkXhkwJUmS1CsDpiRJknplwJQkSVKvDJiSJEnqlQFTkiRJvTJgSpIkqVcGTEmSJPXKgClJkqReGTAlSZLUKwOmJEmSemXAlCRJUq8MmJIkSeqVAVOSJEm9MmBKkiSpVwZMSZIk9cqAKUmSpF4ZMKUJSrJOknOSzEoyL8nCSdT1lxMstyTJJitR/2eTPHuE+QcnOa5N7z1YJsnCJDtOoO65SU5f0TaNUM8Tkpwx8Hp+khNGKLftSPMlSTOXAVOauEOAk6tqaQ91TShgrqyqektVXT1Osb2BR4TQCTgS+MxKrLecqroVuCnJruOUWww8NcnTJrtNSdKqMXu6GyCtRg4EDmjTS4HboRsVBF4DzAE2B06qqg+2ZW8AjgDWAi4E3gb8HbBOksuAq6rqwCSnApsBawPHVtXxozUiyZ8CL6qqI5O8A3hHVW2RZEvg81W1WxtdPaqqFiV5E/Be4CbgR8Bvk7wYeDXwsiTvA17bqn9dkk8AjwPeXFXnjdCE1wLva22ZBfwD8IdAAZ+pqn9NsgQ4CdgdeCxwKPBh4BnAR6rqU62uU9t+/S7wAHDXKN3+BrA/8I+j7ReAB2+7kZtPWjBWEc1Q8y/4yHQ3QVqjLVy4cJVuzxFMaQKSrAVsUVVLAKrqhqraZ6DIznRBaTu6kLZjkmcB+wG7VtV2dKH0wKpaANxXVdtV1YFt/UOqagdgR+CIJBuP0ZxzgZe06ZcAtyXZFNgNWC4QJpkLfBDYFXgFbcSyqr4HnAa8u7XjJ22V2VW1M/BO4AMj7IfNgTuq6rdt1qF0oXr7qnoecOJA8RuqapfWphOAfYEXAR8aKLNoqC9V9b2qescoff5duRHadGiSRUkW1cN9DC5LkibLEUxpYjYB7hxj+VlVdRtAkpPpwt5DwA7ARUkA1gFuGWX9I5K8pk1vBmwF3DZSwaq6Ocn6STZoZU8CXkoXwE4eVvyFwMJ2OZokXwG2HqMfQ+tfDMwbYflc4NaB13sAn6qqh1rbbh9Ydlr7vRhYv6ruAe5Jcn+Sx1XVnXT74yljtGfIqOXaaO/xAHPmblVPPuCYCVSnmWbhMXtOdxMk9ciAKU3MfXSXr0dTI7wO3SXr945VcZL5dEFtl6q6t13eHmtbAN8H3gRcSzdCeAiwC/AXE2jbWIZGJpcy8r8Pw/dDxqh/qK6HB6aHXg/VvXarczwTLSdJmgG8RC5NQFXdAcxKMlrwe0WSjZKsQ/fwzHeBs4F9kzwRoC1/eiv/YJLHtukN6S4735tkG7rLyOM5Fziq/b6U7l7H31bV8HsYLwTmJ9m4be91A8vuATaYwLYG/YjlRzbPBA5LMhu6Pq5gfVsDVw6fmWTnJP8xXjlJ0sxkwJQm7ky6S98jOR/4AnAZ8PWqWtSe4n4fcGaSK4Cz6C4xQ3dJ94okJwJnALNbmb8BLphAW86juzx+bnuq/YbWhuVU1U3A0XQjnt8GLhlY/GXg3UkubQ8IjauqfgP8JMkz2qzPAj9rfbmcZQ9BTdTuwDdHmP80lh+xHK2cJGkGStWKXD2THr2SbA8cWVUHDZt/MLBjVR0+LQ1bxdq9ojtU1ft6qOtcYK82Qjw4/yPAF6rqiiRzgHOA3Ybu9RzNnLlb1dw3fmyyzdI0WOI9mNK0SXJxVY37PcgrwnswpQmqqkuTfCfJrJ6+C3O1VFWnjPOU+4QkeQLwT8PDZdvGuwdePg1YMF64lCTNHAZMaQVU1b+PMO8Euq/hedSoqs/2UMetdN+DOV6564DrJrs9SdKq4z2YkiRJ6pUBU5IkSb0yYEqSJKlXBkxJkiT1yoApSZKkXhkwJUmS1CsDpiRJknplwJQkSVKvDJiSJEnqlQFTkiRJvTJgSpIkqVcGTEmSJPXKgClJkqReGTAlSZLUKwOmJEmSemXAlCRJUq8MmJIkSeqVAVOSJEm9MmBKkiSpVwZMSZIk9Wr2dDdAkvqy7aYbsuiYPae7GZL0qOcIpiRJknplwJQkSVKvDJiSJEnqlQFTkiRJvTJgSpIkqVcGTEmSJPXKgClJkqReGTAlSZLUKwOmJEmSemXAlCRJUq8MmJIkSeqVAVOSJEm9MmBKkiSpV7OnuwGS1JfFP7+LeQu+Od3NkKbUkmP2nO4mSONyBFOSJEm9MmBKkiSpVwZMSZIk9cqAKUmSpF4ZMCVJktQrA6YkSZJ6ZcCUJElSrwyYkiRJ6pUBU5IkSb0yYEqSJKlXBkxJkiT1yoApSZKkXhkwJUmS1CsDpiRJknplwJQkSVKvDJiSJEnqlQFTkiRJvTJgSpIkqVcGTEmSJPXKgClJkqReGTAlSZLUKwOmJEmSemXAlCRJUq8MmJIkSeqVAVOSJEm9GjdgJlknyTlJZiWZl2ThKmgXSY5OctRE5/ewvXcmWXfg9a8nuN7eSd7fw/ZfleSDA6+PTnLwOOscnOToldzevCQHTLDclROdP1lJ5id58cDrE5LsO4H1fneeTnL7ayU5N8ns9npC53ySJZPY5nLn3hjlFibZcSXq/1CSPUaYPz/J6QPTM3K/J9k2yQmTqV+StGpNZATzEODkqlo61Y2ZZu8Exv2QH8F7gE/0sP1vAq+eSNDoyTxg3IA5DeYDLx6v0Ah6OU+r6gHgbGC/ydSzgt7Jyp17E1JV76+qb49TbD4zdL9X1WLgqUmeNpltSJJWndkTKHMgy4LIUuB2gCTPAT4HrEUXVF9bVdcleQNwRJt/IfC2qlraRgQ/DewO3AHsX1W3JnkrcGgr/2PgoKq6dyKNT7Il8HHgCcC9wFur6po22nE3sCPwZOA9VfW1JI8BjgNeBlzf2v3vwFPaz3eS/Kqqdm/1/x3wKuA+YK+q+uWw7W8N/LaqftVePwn4FLBFK/JnwC+AM4DzgRcBl7f99kHgicCBVfWDqqo2YvMq4D+BX7ftjuW+Vo7W5/uB5wBPAo6sqtPbyNIxdAFiDvDxqvp0m/esJJcBnwdOAb4ArNfqPryqvjfO9of2w4jbSDIfOBr4FfBc4GLgDa2vfwz8U1t2SdtnhwOHAUvbefT2tomXJjmSgWM5QjMGz1OSvAc4CHgY+FZVLWj791JgB7pz5v8A7wW2Bb5SVe9rq58KfBg4kYFzfhy3tu3OBz4E3AY8EziX7j3wcJI/oDvuc4CfAG+iC2jLnXtJPgnsBKwDfK2qPjDaRpPsDCyoqn2S7AV8GdiQ7ty+uqq2aOfG6e098EfAx1i230kyj5m/378B7A/842j7Qo9080kLprsJmgLzL/jIdDdBPVu4cOF0N6F3Y45gJlkL2KKqlgBU1Q1VtU9bfBhwbFVtRxfkbkzyLLoRiF3b/KV0H0DQBZdLquoFwDnA0IfmyVW1U1U9H/gh8OYVaP/xwNuragfgKJYfSZwL7EYX2I5p8/ahG7nbFngLsEvr17/QBcHdh8Jla+8FrV3nAm8dYfu70j6km38BzmnrvAC4qs1/BnAs8DxgG7oP5N1am/9yYP1FwEtamz5aVV8Zq/NV9ZWq+ujArHl04XlP4FNJ1qbbn3dV1U50oeWtSTYHFgDnVdV2VfXPwC3AK9rx2a/1ZaJG2wbA9nQjdM+mC5G7tnZ9GnhlVe1GFzpo59mngH9u7Tqv1THSsfyd4edpklcCewMvbMdiMJQ8UFUvbdv5L+DP6cLvwUk2bmWubP0Yfs6PqvV9yM7AX9CdZ1sC+yTZBHgfsEfbx4vo/ggY6dz7q6rake58eVmS542x6Uvo9jF0585Q219I9wfe77T9/hngT1rZJ7e2L2Hm7/ffvTdGaMehSRYlWbT03rtG3EmSpFVrvBHMTYA7R1n2feCvkjyVLiRel+TldKMUFyWBbgTmllb+YWAoMH0ROLlNPzfJ3wKPA9YH/mciDU+yPt0lva+2bUE3MjTk1Kp6GLi6jSxC92H51Tb/5iTfGWMTDwCnt+mLgVeMUGYubeSq+X260RnaJcO7kjweuL5d5iPJVcDZbRRvMV0oHHIL3WjWyvrP1rfrkvyULsz+AfC8gfvpNgS2av0b9FjguCTb0f1hsPUKbHesbfygqm4EaKOl8+hGXX9aVde38l+iG8UezUjHctDw83QP4HNDI+FVNTgSdlr7vRi4qqpuam37KbAZcFsbcX8gyQZVdc94nR/BD6rqp63eL9Gdd/fThezvtvN1Lbr30Ej+NMmhdO/PuW29K0YqWFUPJflx++NuZ7pR4ZcCs4DzhhXfhu5cvK617YusPvt91PdGVR1P98cmc+ZuVWP051HnyQc84u8CrQEWHrPndDdBGtd4AfM+YO2RFlTVSUkupBst+58kbwECfL6q3juBbQ99EJwA7F1Vl6d7qGX+BNaFbvT1zjZSOpLfDkxn2O+JeLCqhtq4lJH31X10YWo8g215eOD1w8PqXZvxL4uPZfiHa9H1+e1VtVxwb5dyB70L+CXwfLp9e/8KbHesbQz2fWg/rshxgJGP5aDh52l45L4YXtfgcRh6PXgs5rBi+2DQaMfhrKp6/VgrtpHfo4CdquqOdnl7xPfggPOAVwIPAt+me0/NavWM17axzKT9Ptn3hiRpFRrzEnlV3QHMapfWlpNkC7pRqH+hG514Ht1N+vsmeWIrs1GSpw9sa2iE6wC6exIBNgBuSvJYll1OH1dV3Q1cn+R1bVtJ8vxxVjsfeG2Sx7QRmfkDy+5pbVkRP6S7/D3kbLr7Lkn31P3vrWB9W9NdJlxOksOTHD6B9V/X+rYl3eXoa+lGhP+s7V+SbJ1kPR7Z3w2Bm9qI1UF0AWWiRtvGaK4Btmj3/sHyD3as8HEY4Tw9Ezhk6IGpJButSH3tku2tVfXgsPmbJjl7AlXsnGTzds/vfnTn3QV0twc8o9W1bruHF5bv8+8Bv6Eb/X4SXXAcz7l0tyF8v6puBTamG628ali5a4DN2/kBMBh2Z+x+b0Z8b0iSZqaJPEV+Jt0lvuH2A65slz23Af6jqq6mu8/szCRXAGfRXeKD7kPzOUkupruU/KE2/6/p7hU7i+4DcEUcCLw5yeV0H6Z7jVP+68CNdB9Un27bHbpp63jgW+NcNh/uXGD7LLtG/w5g93bp+2K6B25WxO50T5MPtw3dQyPjuZbu/tZvAYdV1f3AZ4GrgUvSfa3Qp+lGjK4AHkpyeZJ30d2/+sYkF9B9mP9mBdo92jZGVFX3AW8DzkhyPt3I6dBx+AbwmiSXJRnxnrtR/O48raoz6P7oWdTOzxX9Wqvdgf8eYf5c4KEJrP99unsWr6R7mOyUFvwOBr7U3hsX0B1XGDj3qupyugdirqJ7AO27E9jehXQPdp3bXl8BXDEwAg9AOx8OBb7Z9vv/Diyeyft9aNlI7w1J0gyUYZ9BjyyQbE/3MMJBk9pQ8uuqWn8ydfQhyfpV9es2WvIDugeSbp5EfccC36jxvwZmvHqeBJxUVS8fYdnpwD7VfZXLaOufQHtSeDLtWFUGjkPovgnguvaw0crW18t52uo6GXhvVV07bP7hwM+q6rSR1/zdbQFHVdWrJtuO1cEq2u9z6P5w2q2qxgz4c+ZuVXPf+LHJNkWa0ZZ4D6Z6luTi9nBpb8b9mqKqujTJd5LMqjXjuzBPT/I4uocs/mYy4bL5e7ondifraXRPHj/CGhpW3prkjXTH4VK6Uc+V1td5mu7J6FOHh5y2jeMm08Y10arY73TvjQXjhUtJ0swx7gimJK0uHMHUo4EjmOrbVIxg+n+RS5IkqVcGTEmSJPXKgClJkqReGTAlSZLUKwOmJEmSemXAlCRJUq8MmJIkSeqVAVOSJEm9MmBKkiSpVwZMSZIk9cqAKUmSpF4ZMCVJktQrA6YkSZJ6ZcCUJElSrwyYkiRJ6pUBU5IkSb0yYEqSJKlXBkxJkiT1yoApSZKkXhkwJUmS1KvZ090ASerLtptuyKJj9pzuZkjSo54jmJIkSeqVAVOSJEm9MmBKkiSpVwZMSZIk9cqAKUmSpF4ZMCVJktQrA6YkSZJ6ZcCUJElSrwyYkiRJ6lWqarrbIEm9SHIPcO10t2MKbQL8arobMUXW5L6B/Vvdren9e2ZVbdBnhf5XkZLWJNdW1Y7T3YipkmTRmtq/NblvYP9Wd4+G/vVdp5fIJUmS1CsDpiRJknplwJS0Jjl+uhswxdbk/q3JfQP7t7qzfyvIh3wkSZLUK0cwJUmS1CsDpiRJknplwJQ0IyX5oyTXJvlxkgUjLH98klOSXJHkB0meO7BsSZLFSS4b/PqNJBslOSvJde3341dVf4abov4dneTnbf5lSf54VfVnuEn273FJvpbkmiQ/TLJLmz8jjt8U9W21P3ZJnjnQ/suS3J3knW3ZjDh2rS1T0b/V/vi1Ze9KclWSK5N8Kcnabf6KH7+q8scff/yZUT/ALOAnwBbAWsDlwLOHlfkI8IE2vQ1w9sCyJcAmI9T7j8CCNr0A+Ic1rH9HA0etAcfv88Bb2vRawONmyvGbwr6tEcduWD03A0+fKcduivu32h8/YFPgemCd9vo/gYNX9vg5gilpJtoZ+HFV/bSqHgC+DOw1rMyzgbMBquoaYF6SJ41T7150H/C033v31uIVM1X9mylWun9Jfg94KfBvbdkDVXVnW2cmHL+p6ttM0de5+XLgJ1X1v+31TDh2MHX9mykm27/ZwDpJZgPrAr9o81f4+BkwJc1EmwI3DLy+sc0bdDmwD0CSnYGnA09tywo4M8nFSQ4dWOdJVXUTQPv9xClo+0RMVf8ADm+Xvv59Gi9DTqZ/WwC3Ap9LcmmSzyZZr60zE47fVPUNVv9jN2h/4EsDr2fCsYOp6x+s5sevqn4OfBT4GXATcFdVndnWWeHjZ8CUNBNlhHnDv1PtGODxSS4D3g5cCjzUlu1aVS8AXgn8eZKXTlVDV9JU9e+TwJbAdnQfEP+v32ZP2GT6Nxt4AfDJqtoe+A3dJbmZYqr6tiYcu66CZC3g1cBXp6iNkzFV/Vvtj18LxXsBmwNPAdZL8oaVbYj/F7mkmehGYLOB109l2aUaAKrqbuBNAElCd+/Q9W3ZL9rvW5KcQnfZ6Fzgl0nmVtVNSeYCt0x1R0YxJf2rql8OrZ/kM8DpU9iHsUymf+sCN1bVha3o11gWwmbC8ZuSvq0hx27IK4FLBvvEzDh2MEX9W0OO3x8C11fVrW3ZycCLgS+yEsfPEUxJM9FFwFZJNm+jBfsDpw0WSPc07lrt5VvoAtbdSdZLskErsx7wB8CVrdxpwBvb9BuB/5rifoxmSvrX/uEf8hqW9XtVW+n+VdXNwA1JntmWvRy4uk3PhOM3JX1bE47dQJHX88jLxzPh2MEU9W8NOX4/A16UZN0WPF8O/LCVW/HjN91PPPnjjz/+jPQD/DHwI7onIv+qzTsMOKxN7wJcB1wDnAw8vs3fgu4eo8uBq4bWbcs2pru5/br2e6M1rH9fABYDV7QPhLmrW//asu2ARa0fpw70fUYcvynq25py7NYFbgM2HFbnjDh2U9i/NeX4fbDNv7L1ac7KHj//q0hJkiT1ykvkkiRJ6pUBU5IkSb0yYEqSJKlXBkxJkiT1yoApSZKkXhkwJUkakOQ1SSrJNtPdFml1ZcCUJGl5rwfOp/uS6imRZNZU1S3NBAZMSZKaJOsDuwJvpgXMJLOSfDTJ4iRXJHl7m79Tku8luTzJD5JskOTgJMcN1Hd6kvlt+tdJPpTkQmCXJO9PclGSK5Mc3/73FJI8I8m3W72XJNkyyReS7DVQ74lJXr2q9ou0ogyYkiQtszdwRlX9CLg9yQuAQ4HNge2r6nnAie2/2vsK8I6qej6wB3DfOHWvB1xZVS+sqvOB46pqp6p6LrAO8KpW7kTg463eFwM3AZ9l2f8fvWGb/999dVrqmwFTkqRlXg98uU1/ub3eA/hUVT0EUFW3A88Ebqqqi9q8u4eWj2Ep8PWB17snuTDJYuD3gee0/2d+06o6pdV7f1XdW1XnAM9I8sTWpq9PYHvStJk93Q2QJGkmSLIxXdB7bpICZgEFXNx+L1d8hHkAD7H84M3aA9P3V9XStq21gU8AO1bVDUmObmUzRhO/ABxId+n+kAl2S5oWjmBKktTZF/iPqnp6Vc2rqs2A64FLgMOSzAZIshFwDfCUJDu1eRu05UuA7ZI8JslmwM6jbGsoeP6q3fe5L3QjocCNSfZu9c5Jsm4rewLwzlbuqt56LU0BA6YkSZ3XA6cMm/d14CnAz4ArklwOHFBVDwD7Af/a5p1FFxq/SxdKFwMfpQunj1BVdwKfaeVOBS4aWHwQcESSK4DvAU9u6/wS+CHwuUn2U5pyqRpphF+SJM0kbSRzMfCCqrprutsjjcURTEmSZrgke9Bdlv9Xw6VWB45gSpIkqVeOYEqSJKlXBkxJkiT1yoApSZKkXhkwJUmS1CsDpiRJknr1/wHPdEjpFslOTAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#%% show feature importance visually\n",
    "# create figure and axes\n",
    "fig, ax = plt.subplots()\n",
    "\n",
    "# plot bars\n",
    "y_pos = np.arange(len(sfs_results))\n",
    "ax.barh(y_pos, sfs_results['avg_score'], \\\n",
    "    xerr=sfs_results['std_err'])\n",
    "\n",
    "# set axis ticks and labels\n",
    "ax.set_yticks(y_pos)\n",
    "ax.set_yticklabels(sfs_results['feature_names'])\n",
    "ax.set_xlabel('Accuracy')\n",
    "\n",
    "# limit range to overimpose differences\n",
    "plt.xlim([0.95, 0.98])\n",
    "\n",
    "# show the plot\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bf1751fa",
   "metadata": {},
   "source": [
    "### 4.3 Sequential Backward Feature Selection (SBS)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "id": "bd77e8f8",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Sequential Backward Feature Selection (SBS)\n",
    "\n",
    "#%% import libraries\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.datasets import load_iris\n",
    "from mlxtend.feature_selection     import SequentialFeatureSelector as SBS\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "id": "b8adcfd2",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% load sample data\n",
    "iris = load_iris()\n",
    "x = pd.DataFrame(iris.data, \n",
    "    columns=iris.feature_names)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "id": "2d8b8ad0",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% create a logistic regression object\n",
    "lr = LogisticRegression()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "id": "5763766b",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% create an SBS object\n",
    "sbs = SBS(estimator=lr,\n",
    "          k_features=(1, 3),  \n",
    "          forward=False,       \n",
    "          scoring='accuracy', \n",
    "          cv=5)    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "id": "ed7ad5a4",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\sklearn\\linear_model\\_logistic.py:818: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  extra_warning_msg=_LOGISTIC_SOLVER_CONVERGENCE_MSG,\n"
     ]
    }
   ],
   "source": [
    "#%% fit the model\n",
    "sbs = sbs.fit(x, iris.target)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "id": "b3d21d7b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "('sepal length (cm)', 'petal length (cm)', 'petal width (cm)')"
      ]
     },
     "execution_count": 83,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#%% show the selected features\n",
    "sbs.k_feature_names_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "id": "e3d0d64f",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% show a full report on the feature selection\n",
    "sbs_results = pd.DataFrame(sbs.get_metric_dict()).T.sort_values(by='avg_score', ascending=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "id": "8980a8a9",
   "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>feature_idx</th>\n",
       "      <th>cv_scores</th>\n",
       "      <th>avg_score</th>\n",
       "      <th>feature_names</th>\n",
       "      <th>ci_bound</th>\n",
       "      <th>std_dev</th>\n",
       "      <th>std_err</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>(0, 1, 2, 3)</td>\n",
       "      <td>[0.9666666666666667, 1.0, 0.9333333333333333, ...</td>\n",
       "      <td>0.973333</td>\n",
       "      <td>(sepal length (cm), sepal width (cm), petal le...</td>\n",
       "      <td>0.032061</td>\n",
       "      <td>0.024944</td>\n",
       "      <td>0.012472</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>(0, 2, 3)</td>\n",
       "      <td>[0.9666666666666667, 0.9666666666666667, 0.933...</td>\n",
       "      <td>0.966667</td>\n",
       "      <td>(sepal length (cm), petal length (cm), petal w...</td>\n",
       "      <td>0.027096</td>\n",
       "      <td>0.021082</td>\n",
       "      <td>0.010541</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>(2, 3)</td>\n",
       "      <td>[0.9666666666666667, 0.9666666666666667, 0.933...</td>\n",
       "      <td>0.96</td>\n",
       "      <td>(petal length (cm), petal width (cm))</td>\n",
       "      <td>0.032061</td>\n",
       "      <td>0.024944</td>\n",
       "      <td>0.012472</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>(3,)</td>\n",
       "      <td>[1.0, 0.9666666666666667, 0.9, 0.9333333333333...</td>\n",
       "      <td>0.96</td>\n",
       "      <td>(petal width (cm),)</td>\n",
       "      <td>0.049963</td>\n",
       "      <td>0.038873</td>\n",
       "      <td>0.019437</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    feature_idx                                          cv_scores avg_score  \\\n",
       "4  (0, 1, 2, 3)  [0.9666666666666667, 1.0, 0.9333333333333333, ...  0.973333   \n",
       "3     (0, 2, 3)  [0.9666666666666667, 0.9666666666666667, 0.933...  0.966667   \n",
       "2        (2, 3)  [0.9666666666666667, 0.9666666666666667, 0.933...      0.96   \n",
       "1          (3,)  [1.0, 0.9666666666666667, 0.9, 0.9333333333333...      0.96   \n",
       "\n",
       "                                       feature_names  ci_bound   std_dev  \\\n",
       "4  (sepal length (cm), sepal width (cm), petal le...  0.032061  0.024944   \n",
       "3  (sepal length (cm), petal length (cm), petal w...  0.027096  0.021082   \n",
       "2              (petal length (cm), petal width (cm))  0.032061  0.024944   \n",
       "1                                (petal width (cm),)  0.049963  0.038873   \n",
       "\n",
       "    std_err  \n",
       "4  0.012472  \n",
       "3  0.010541  \n",
       "2  0.012472  \n",
       "1  0.019437  "
      ]
     },
     "execution_count": 85,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sbs_results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "id": "c93b401e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#%% show feature importance visually\n",
    "# create figure and axes\n",
    "fig, ax = plt.subplots()\n",
    "# plot bars\n",
    "y_pos = np.arange(len(sbs_results))\n",
    "ax.barh(y_pos, sbs_results['avg_score'], \\\n",
    "    xerr=sbs_results['std_err'])\n",
    "\n",
    "# set axis ticks and labels\n",
    "ax.set_yticks(y_pos)\n",
    "ax.set_yticklabels(sbs_results['feature_names'])\n",
    "ax.set_xlabel('Accuracy')\n",
    "\n",
    "# limit range to overimpose differences\n",
    "plt.xlim([0.95, 0.98])\n",
    "\n",
    "# show the plot\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f932c628",
   "metadata": {},
   "source": [
    "### 4.4 Recursive Feature Elimination (RFE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "id": "3bfee205",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Recursive Feature Elimination\n",
    "\n",
    "#%% import libraries\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.datasets import load_iris\n",
    "from sklearn.feature_selection import RFE\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.feature_selection import RFECV"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "id": "06c7eab1",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% load sample data\n",
    "iris = load_iris()\n",
    "x = pd.DataFrame(iris.data,columns=iris.feature_names)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "id": "f9a15cf2",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%% create a logistic regression object\n",
    "lr = LogisticRegression()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "id": "893d322d",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\sklearn\\linear_model\\_logistic.py:818: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  extra_warning_msg=_LOGISTIC_SOLVER_CONVERGENCE_MSG,\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "RFE(estimator=LogisticRegression(), n_features_to_select=3)"
      ]
     },
     "execution_count": 102,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#%% create and fit logistic regressor with RFE\n",
    "rfe = RFE(estimator = lr, n_features_to_select =3)\n",
    "rfe.fit(x, iris.target)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "id": "d393c6b3",
   "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>features</th>\n",
       "      <th>Selected features</th>\n",
       "      <th>Feature ranks</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>sepal length (cm)</td>\n",
       "      <td>False</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>sepal width (cm)</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>petal length (cm)</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>petal width (cm)</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            features  Selected features  Feature ranks\n",
       "0  sepal length (cm)              False              2\n",
       "1   sepal width (cm)               True              1\n",
       "2  petal length (cm)               True              1\n",
       "3   petal width (cm)               True              1"
      ]
     },
     "execution_count": 103,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#%% show which feature were selected\n",
    "pd.DataFrame({'features': iris.feature_names,\n",
    "    'Selected features': rfe.support_, \n",
    "    'Feature ranks': rfe.ranking_})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "id": "0b4884f6",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\sklearn\\linear_model\\_logistic.py:818: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  extra_warning_msg=_LOGISTIC_SOLVER_CONVERGENCE_MSG,\n",
      "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\sklearn\\linear_model\\_logistic.py:818: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  extra_warning_msg=_LOGISTIC_SOLVER_CONVERGENCE_MSG,\n",
      "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\sklearn\\linear_model\\_logistic.py:818: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  extra_warning_msg=_LOGISTIC_SOLVER_CONVERGENCE_MSG,\n"
     ]
    }
   ],
   "source": [
    "#Other option: It is possible to find the number of features to be selected using cross validation\n",
    "#%% REF with cross-validation = 5\n",
    "#step: the number of features to be eliminated at each iteration\n",
    "rfecv = RFECV(estimator=lr, step=1, cv=5, \n",
    "    scoring='accuracy', \n",
    "    min_features_to_select= 3).fit(x, iris.target)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "id": "f282bcf5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[0.96666667 0.96666667 0.93333333 0.96666667 1.        ]\n",
      " [0.96666667 1.         0.93333333 0.96666667 1.        ]]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\sklearn\\utils\\deprecation.py:103: FutureWarning: The `grid_scores_` attribute is deprecated in version 1.0 in favor of `cv_results_` and will be removed in version 1.2.\n",
      "  warnings.warn(msg, category=FutureWarning)\n"
     ]
    }
   ],
   "source": [
    "#%% show grid scores for selected featues\n",
    "print(rfecv.grid_scores_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "id": "9600dba5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'mean_test_score': array([0.96666667, 0.97333333]),\n",
       " 'std_test_score': array([0.02108185, 0.02494438]),\n",
       " 'split0_test_score': array([0.96666667, 0.96666667]),\n",
       " 'split1_test_score': array([0.96666667, 1.        ]),\n",
       " 'split2_test_score': array([0.93333333, 0.93333333]),\n",
       " 'split3_test_score': array([0.96666667, 0.96666667]),\n",
       " 'split4_test_score': array([1., 1.])}"
      ]
     },
     "execution_count": 108,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rfecv.cv_results_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "id": "fb920460",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "4"
      ]
     },
     "execution_count": 109,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# The number of selected features with cross-validation.\n",
    "rfecv.n_features_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9a7234d7",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.7.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}