WAZE PART III, XGBoost.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "DtNBZFHO3M7n"
   },
   "source": [
    "![](WAZE.jpg)\n",
    "# Waze User Churn Project: Random Forest and XGBoost models\n",
    "Waze is a free navigation app that makes it easier for drivers around the world to get to where they want to go. Waze leadership wants to optimize the company’s retention strategy, enhance user experience, and make data-driven decisions about product development. They would like an analysis of WAZE data to understand their users better and the development of a machine learning model that predicts user churn. (Churn is understood to be the number of users who have uninstalled the Waze app or stopped using it.) \n",
    "\n",
    "This project is part of a larger effort at Waze to increase growth. It assumes that high retention rates indicate satisfied users who repeatedly employ the Waze app over time. Identifying and predicting which users are likely to churn will allow the WAZE team to target such individuals to induce their retention, thereby allowing Waze to grow its business. \n",
    "\n",
    "**Data**    \n",
    "Waze’s free navigation app makes it easier for drivers around the world to get to where they want to go. Waze’s community of map editors, beta testers, translators, partners, and users help make each drive better and safer. Waze partners with cities, transportation authorities, broadcasters, businesses, and first responders to help as many people as possible travel more efficiently and safely. The data set is in-house from Waze for Cities (https://www.transportation.gov/office-policy/transportation-policy/faq-waze-data).\n",
    "\n",
    "**Deliverables**   \n",
    "(Since this is an exercise, all models are predetermined.)\n",
    "* I.\tAn analysis of WAZE data to understand their users better\n",
    "* II.\tThe development of a machine learning model that predicts user churn\n",
    "    * a.\ta binomial logistic regression model\n",
    "    * b.\ta winning tree-based model\n",
    "* Appendix: A 2 sample t-test based on a sample of user data determining if there is a statistically significant difference in the mean number of rides between iPhone® users and Android™ users"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# PART III: Build and test two tree-based models: random forest and XGBoost\n",
    "* basic EDA\n",
    "* prepare data for modelling\n",
    "* build models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "id": "fKhnX2Puf4Bt"
   },
   "outputs": [],
   "source": [
    "# Import standard operational packages\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "# Import preprocessing, modeling, and evaluation packages\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "from sklearn.model_selection import train_test_split, GridSearchCV, PredefinedSplit\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.metrics import recall_score, precision_score, f1_score, accuracy_score, classification_report\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "# Import from xgboost\n",
    "from xgboost import XGBClassifier\n",
    "from xgboost import plot_importance\n",
    "\n",
    "# Import visualization package\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from sklearn.tree import plot_tree\n",
    "from sklearn.metrics import confusion_matrix, ConfusionMatrixDisplay, RocCurveDisplay, roc_auc_score\n",
    "\n",
    "# Save models once fitted.\n",
    "import pickle\n",
    "\n",
    "# Set Jupyter to display all of the columns (no redaction)\n",
    "pd.set_option('display.max_columns', None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "id": "5weTXGKqa_iG"
   },
   "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>label</th>\n",
       "      <th>sessions</th>\n",
       "      <th>drives</th>\n",
       "      <th>total_sessions</th>\n",
       "      <th>n_days_after_onboarding</th>\n",
       "      <th>total_navigations_fav1</th>\n",
       "      <th>total_navigations_fav2</th>\n",
       "      <th>driven_km_drives</th>\n",
       "      <th>duration_minutes_drives</th>\n",
       "      <th>activity_days</th>\n",
       "      <th>driving_days</th>\n",
       "      <th>device</th>\n",
       "      <th>percent_sessions_in_month</th>\n",
       "      <th>total_sessions_per_day</th>\n",
       "      <th>kms_driven_per_day_during_the_month</th>\n",
       "      <th>km_per_drive</th>\n",
       "      <th>km_per_hour</th>\n",
       "      <th>professional_driver</th>\n",
       "      <th>label2</th>\n",
       "      <th>device2</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>retained</td>\n",
       "      <td>243</td>\n",
       "      <td>200</td>\n",
       "      <td>296.748273</td>\n",
       "      <td>2276</td>\n",
       "      <td>208</td>\n",
       "      <td>0</td>\n",
       "      <td>2628.845068</td>\n",
       "      <td>1985.775061</td>\n",
       "      <td>28</td>\n",
       "      <td>19</td>\n",
       "      <td>Android</td>\n",
       "      <td>0.936313</td>\n",
       "      <td>0.130381</td>\n",
       "      <td>138.360267</td>\n",
       "      <td>11.632058</td>\n",
       "      <td>79.430298</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>retained</td>\n",
       "      <td>133</td>\n",
       "      <td>107</td>\n",
       "      <td>326.896596</td>\n",
       "      <td>1225</td>\n",
       "      <td>19</td>\n",
       "      <td>64</td>\n",
       "      <td>8898.716275</td>\n",
       "      <td>3160.472914</td>\n",
       "      <td>13</td>\n",
       "      <td>11</td>\n",
       "      <td>iPhone</td>\n",
       "      <td>0.406856</td>\n",
       "      <td>0.266854</td>\n",
       "      <td>1246.901868</td>\n",
       "      <td>128.186173</td>\n",
       "      <td>260.389902</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>retained</td>\n",
       "      <td>114</td>\n",
       "      <td>95</td>\n",
       "      <td>135.522926</td>\n",
       "      <td>2651</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>3059.148818</td>\n",
       "      <td>1610.735904</td>\n",
       "      <td>14</td>\n",
       "      <td>8</td>\n",
       "      <td>Android</td>\n",
       "      <td>0.841186</td>\n",
       "      <td>0.051121</td>\n",
       "      <td>382.393602</td>\n",
       "      <td>32.201567</td>\n",
       "      <td>113.953460</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>retained</td>\n",
       "      <td>49</td>\n",
       "      <td>40</td>\n",
       "      <td>67.589221</td>\n",
       "      <td>15</td>\n",
       "      <td>322</td>\n",
       "      <td>7</td>\n",
       "      <td>913.591123</td>\n",
       "      <td>587.196542</td>\n",
       "      <td>7</td>\n",
       "      <td>3</td>\n",
       "      <td>iPhone</td>\n",
       "      <td>0.724968</td>\n",
       "      <td>1.089743</td>\n",
       "      <td>304.530374</td>\n",
       "      <td>22.839778</td>\n",
       "      <td>93.351141</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>retained</td>\n",
       "      <td>84</td>\n",
       "      <td>68</td>\n",
       "      <td>168.247020</td>\n",
       "      <td>1562</td>\n",
       "      <td>166</td>\n",
       "      <td>5</td>\n",
       "      <td>3950.202008</td>\n",
       "      <td>1219.555924</td>\n",
       "      <td>27</td>\n",
       "      <td>18</td>\n",
       "      <td>Android</td>\n",
       "      <td>0.499266</td>\n",
       "      <td>0.107713</td>\n",
       "      <td>219.455667</td>\n",
       "      <td>58.091206</td>\n",
       "      <td>194.342970</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      label  sessions  drives  total_sessions  n_days_after_onboarding  \\\n",
       "0  retained       243     200      296.748273                     2276   \n",
       "1  retained       133     107      326.896596                     1225   \n",
       "2  retained       114      95      135.522926                     2651   \n",
       "3  retained        49      40       67.589221                       15   \n",
       "4  retained        84      68      168.247020                     1562   \n",
       "\n",
       "   total_navigations_fav1  total_navigations_fav2  driven_km_drives  \\\n",
       "0                     208                       0       2628.845068   \n",
       "1                      19                      64       8898.716275   \n",
       "2                       0                       0       3059.148818   \n",
       "3                     322                       7        913.591123   \n",
       "4                     166                       5       3950.202008   \n",
       "\n",
       "   duration_minutes_drives  activity_days  driving_days   device  \\\n",
       "0              1985.775061             28            19  Android   \n",
       "1              3160.472914             13            11   iPhone   \n",
       "2              1610.735904             14             8  Android   \n",
       "3               587.196542              7             3   iPhone   \n",
       "4              1219.555924             27            18  Android   \n",
       "\n",
       "   percent_sessions_in_month  total_sessions_per_day  \\\n",
       "0                   0.936313                0.130381   \n",
       "1                   0.406856                0.266854   \n",
       "2                   0.841186                0.051121   \n",
       "3                   0.724968                1.089743   \n",
       "4                   0.499266                0.107713   \n",
       "\n",
       "   kms_driven_per_day_during_the_month  km_per_drive  km_per_hour  \\\n",
       "0                           138.360267     11.632058    79.430298   \n",
       "1                          1246.901868    128.186173   260.389902   \n",
       "2                           382.393602     32.201567   113.953460   \n",
       "3                           304.530374     22.839778    93.351141   \n",
       "4                           219.455667     58.091206   194.342970   \n",
       "\n",
       "   professional_driver  label2  device2  \n",
       "0                    1       0        0  \n",
       "1                    1       0        1  \n",
       "2                    0       0        0  \n",
       "3                    0       0        1  \n",
       "4                    0       0        0  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Import data; create df\n",
    "df0 = pd.read_csv('WAZE_dataset_transformed_2.csv', on_bad_lines='skip')\n",
    "df0.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### BASIC EDA revisted\n",
    "* check for nulls\n",
    "* check dtypes: decision trees need numeric columns, so any must be converted to ordinals\n",
    "* no need to be concerned with outliers, since DTs can handle them\n",
    "* encode variables"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "id": "teUeCF-yf_6o"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 14299 entries, 0 to 14298\n",
      "Data columns (total 20 columns):\n",
      " #   Column                               Non-Null Count  Dtype  \n",
      "---  ------                               --------------  -----  \n",
      " 0   label                                14299 non-null  object \n",
      " 1   sessions                             14299 non-null  int64  \n",
      " 2   drives                               14299 non-null  int64  \n",
      " 3   total_sessions                       14299 non-null  float64\n",
      " 4   n_days_after_onboarding              14299 non-null  int64  \n",
      " 5   total_navigations_fav1               14299 non-null  int64  \n",
      " 6   total_navigations_fav2               14299 non-null  int64  \n",
      " 7   driven_km_drives                     14299 non-null  float64\n",
      " 8   duration_minutes_drives              14299 non-null  float64\n",
      " 9   activity_days                        14299 non-null  int64  \n",
      " 10  driving_days                         14299 non-null  int64  \n",
      " 11  device                               14299 non-null  object \n",
      " 12  percent_sessions_in_month            14299 non-null  float64\n",
      " 13  total_sessions_per_day               14299 non-null  float64\n",
      " 14  kms_driven_per_day_during_the_month  14299 non-null  float64\n",
      " 15  km_per_drive                         14299 non-null  float64\n",
      " 16  km_per_hour                          14299 non-null  float64\n",
      " 17  professional_driver                  14299 non-null  int64  \n",
      " 18  label2                               14299 non-null  int64  \n",
      " 19  device2                              14299 non-null  int64  \n",
      "dtypes: float64(8), int64(10), object(2)\n",
      "memory usage: 2.2+ MB\n"
     ]
    }
   ],
   "source": [
    "# Call info on new df\n",
    "df0.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "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>sessions</th>\n",
       "      <th>drives</th>\n",
       "      <th>total_sessions</th>\n",
       "      <th>n_days_after_onboarding</th>\n",
       "      <th>total_navigations_fav1</th>\n",
       "      <th>total_navigations_fav2</th>\n",
       "      <th>driven_km_drives</th>\n",
       "      <th>duration_minutes_drives</th>\n",
       "      <th>activity_days</th>\n",
       "      <th>driving_days</th>\n",
       "      <th>percent_sessions_in_month</th>\n",
       "      <th>total_sessions_per_day</th>\n",
       "      <th>kms_driven_per_day_during_the_month</th>\n",
       "      <th>km_per_drive</th>\n",
       "      <th>km_per_hour</th>\n",
       "      <th>professional_driver</th>\n",
       "      <th>label2</th>\n",
       "      <th>device2</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>14299.000000</td>\n",
       "      <td>14299.000000</td>\n",
       "      <td>14299.000000</td>\n",
       "      <td>14299.000000</td>\n",
       "      <td>14299.000000</td>\n",
       "      <td>14299.000000</td>\n",
       "      <td>14299.000000</td>\n",
       "      <td>14299.000000</td>\n",
       "      <td>14299.000000</td>\n",
       "      <td>14299.000000</td>\n",
       "      <td>14299.000000</td>\n",
       "      <td>14299.000000</td>\n",
       "      <td>14299.000000</td>\n",
       "      <td>14299.000000</td>\n",
       "      <td>14299.000000</td>\n",
       "      <td>14299.000000</td>\n",
       "      <td>14299.000000</td>\n",
       "      <td>14299.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>76.539688</td>\n",
       "      <td>63.964683</td>\n",
       "      <td>183.717304</td>\n",
       "      <td>1751.822505</td>\n",
       "      <td>114.562767</td>\n",
       "      <td>27.187216</td>\n",
       "      <td>3944.558631</td>\n",
       "      <td>1792.911210</td>\n",
       "      <td>15.544653</td>\n",
       "      <td>12.182530</td>\n",
       "      <td>0.448226</td>\n",
       "      <td>0.203300</td>\n",
       "      <td>484.488544</td>\n",
       "      <td>168.147169</td>\n",
       "      <td>165.698737</td>\n",
       "      <td>0.228338</td>\n",
       "      <td>0.177355</td>\n",
       "      <td>0.645150</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>67.243178</td>\n",
       "      <td>55.127927</td>\n",
       "      <td>118.720520</td>\n",
       "      <td>1008.663834</td>\n",
       "      <td>124.378550</td>\n",
       "      <td>36.715302</td>\n",
       "      <td>2218.358258</td>\n",
       "      <td>1224.329759</td>\n",
       "      <td>9.016088</td>\n",
       "      <td>7.833835</td>\n",
       "      <td>0.283954</td>\n",
       "      <td>0.267737</td>\n",
       "      <td>570.872328</td>\n",
       "      <td>232.903899</td>\n",
       "      <td>110.185653</td>\n",
       "      <td>0.419776</td>\n",
       "      <td>0.381982</td>\n",
       "      <td>0.478485</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.220211</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>60.441250</td>\n",
       "      <td>18.282082</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000298</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>72.013095</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>23.000000</td>\n",
       "      <td>20.000000</td>\n",
       "      <td>90.457733</td>\n",
       "      <td>878.500000</td>\n",
       "      <td>10.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>2217.319909</td>\n",
       "      <td>840.181344</td>\n",
       "      <td>8.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>0.196890</td>\n",
       "      <td>0.050818</td>\n",
       "      <td>136.168003</td>\n",
       "      <td>32.486524</td>\n",
       "      <td>90.649402</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>56.000000</td>\n",
       "      <td>48.000000</td>\n",
       "      <td>158.718571</td>\n",
       "      <td>1749.000000</td>\n",
       "      <td>71.000000</td>\n",
       "      <td>9.000000</td>\n",
       "      <td>3496.545617</td>\n",
       "      <td>1479.394387</td>\n",
       "      <td>16.000000</td>\n",
       "      <td>12.000000</td>\n",
       "      <td>0.424310</td>\n",
       "      <td>0.100457</td>\n",
       "      <td>273.301012</td>\n",
       "      <td>72.947059</td>\n",
       "      <td>122.105845</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>111.000000</td>\n",
       "      <td>93.000000</td>\n",
       "      <td>253.540450</td>\n",
       "      <td>2627.500000</td>\n",
       "      <td>178.000000</td>\n",
       "      <td>43.000000</td>\n",
       "      <td>5299.972162</td>\n",
       "      <td>2466.928876</td>\n",
       "      <td>23.000000</td>\n",
       "      <td>19.000000</td>\n",
       "      <td>0.687251</td>\n",
       "      <td>0.215210</td>\n",
       "      <td>558.018761</td>\n",
       "      <td>179.347527</td>\n",
       "      <td>193.238449</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>243.000000</td>\n",
       "      <td>200.000000</td>\n",
       "      <td>455.439492</td>\n",
       "      <td>3500.000000</td>\n",
       "      <td>422.000000</td>\n",
       "      <td>124.000000</td>\n",
       "      <td>8898.716275</td>\n",
       "      <td>4668.180092</td>\n",
       "      <td>31.000000</td>\n",
       "      <td>30.000000</td>\n",
       "      <td>0.936313</td>\n",
       "      <td>1.089743</td>\n",
       "      <td>2258.494484</td>\n",
       "      <td>913.878779</td>\n",
       "      <td>488.824224</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           sessions        drives  total_sessions  n_days_after_onboarding  \\\n",
       "count  14299.000000  14299.000000    14299.000000             14299.000000   \n",
       "mean      76.539688     63.964683      183.717304              1751.822505   \n",
       "std       67.243178     55.127927      118.720520              1008.663834   \n",
       "min        0.000000      0.000000        0.220211                 4.000000   \n",
       "25%       23.000000     20.000000       90.457733               878.500000   \n",
       "50%       56.000000     48.000000      158.718571              1749.000000   \n",
       "75%      111.000000     93.000000      253.540450              2627.500000   \n",
       "max      243.000000    200.000000      455.439492              3500.000000   \n",
       "\n",
       "       total_navigations_fav1  total_navigations_fav2  driven_km_drives  \\\n",
       "count            14299.000000            14299.000000      14299.000000   \n",
       "mean               114.562767               27.187216       3944.558631   \n",
       "std                124.378550               36.715302       2218.358258   \n",
       "min                  0.000000                0.000000         60.441250   \n",
       "25%                 10.000000                0.000000       2217.319909   \n",
       "50%                 71.000000                9.000000       3496.545617   \n",
       "75%                178.000000               43.000000       5299.972162   \n",
       "max                422.000000              124.000000       8898.716275   \n",
       "\n",
       "       duration_minutes_drives  activity_days  driving_days  \\\n",
       "count             14299.000000   14299.000000  14299.000000   \n",
       "mean               1792.911210      15.544653     12.182530   \n",
       "std                1224.329759       9.016088      7.833835   \n",
       "min                  18.282082       0.000000      0.000000   \n",
       "25%                 840.181344       8.000000      5.000000   \n",
       "50%                1479.394387      16.000000     12.000000   \n",
       "75%                2466.928876      23.000000     19.000000   \n",
       "max                4668.180092      31.000000     30.000000   \n",
       "\n",
       "       percent_sessions_in_month  total_sessions_per_day  \\\n",
       "count               14299.000000            14299.000000   \n",
       "mean                    0.448226                0.203300   \n",
       "std                     0.283954                0.267737   \n",
       "min                     0.000000                0.000298   \n",
       "25%                     0.196890                0.050818   \n",
       "50%                     0.424310                0.100457   \n",
       "75%                     0.687251                0.215210   \n",
       "max                     0.936313                1.089743   \n",
       "\n",
       "       kms_driven_per_day_during_the_month  km_per_drive   km_per_hour  \\\n",
       "count                         14299.000000  14299.000000  14299.000000   \n",
       "mean                            484.488544    168.147169    165.698737   \n",
       "std                             570.872328    232.903899    110.185653   \n",
       "min                               0.000000      0.000000     72.013095   \n",
       "25%                             136.168003     32.486524     90.649402   \n",
       "50%                             273.301012     72.947059    122.105845   \n",
       "75%                             558.018761    179.347527    193.238449   \n",
       "max                            2258.494484    913.878779    488.824224   \n",
       "\n",
       "       professional_driver        label2       device2  \n",
       "count         14299.000000  14299.000000  14299.000000  \n",
       "mean              0.228338      0.177355      0.645150  \n",
       "std               0.419776      0.381982      0.478485  \n",
       "min               0.000000      0.000000      0.000000  \n",
       "25%               0.000000      0.000000      0.000000  \n",
       "50%               0.000000      0.000000      1.000000  \n",
       "75%               0.000000      0.000000      1.000000  \n",
       "max               1.000000      1.000000      1.000000  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df0.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "id": "2TdA6SnGWJY-"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "label                                  0\n",
      "sessions                               0\n",
      "drives                                 0\n",
      "total_sessions                         0\n",
      "n_days_after_onboarding                0\n",
      "total_navigations_fav1                 0\n",
      "total_navigations_fav2                 0\n",
      "driven_km_drives                       0\n",
      "duration_minutes_drives                0\n",
      "activity_days                          0\n",
      "driving_days                           0\n",
      "device                                 0\n",
      "percent_sessions_in_month              0\n",
      "total_sessions_per_day                 0\n",
      "kms_driven_per_day_during_the_month    0\n",
      "km_per_drive                           0\n",
      "km_per_hour                            0\n",
      "professional_driver                    0\n",
      "label2                                 0\n",
      "device2                                0\n",
      "dtype: int64\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(14299, 20)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check for missing values\n",
    "print(df0.isnull().sum())\n",
    "df0.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "kxBYyXDSWPkw"
   },
   "source": [
    "#### Encode variables and feature selection\n",
    "* drop device and label (these have been previously encoded as device2 and label2)\n",
    "* no need to drop others (tree-based models can handle multicollinearity)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "id": "kf3uGtUQaWSL"
   },
   "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>sessions</th>\n",
       "      <th>drives</th>\n",
       "      <th>total_sessions</th>\n",
       "      <th>n_days_after_onboarding</th>\n",
       "      <th>total_navigations_fav1</th>\n",
       "      <th>total_navigations_fav2</th>\n",
       "      <th>driven_km_drives</th>\n",
       "      <th>duration_minutes_drives</th>\n",
       "      <th>activity_days</th>\n",
       "      <th>driving_days</th>\n",
       "      <th>percent_sessions_in_month</th>\n",
       "      <th>total_sessions_per_day</th>\n",
       "      <th>kms_driven_per_day_during_the_month</th>\n",
       "      <th>km_per_drive</th>\n",
       "      <th>km_per_hour</th>\n",
       "      <th>professional_driver</th>\n",
       "      <th>label2</th>\n",
       "      <th>device2</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>243</td>\n",
       "      <td>200</td>\n",
       "      <td>296.748273</td>\n",
       "      <td>2276</td>\n",
       "      <td>208</td>\n",
       "      <td>0</td>\n",
       "      <td>2628.845068</td>\n",
       "      <td>1985.775061</td>\n",
       "      <td>28</td>\n",
       "      <td>19</td>\n",
       "      <td>0.936313</td>\n",
       "      <td>0.130381</td>\n",
       "      <td>138.360267</td>\n",
       "      <td>11.632058</td>\n",
       "      <td>79.430298</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>133</td>\n",
       "      <td>107</td>\n",
       "      <td>326.896596</td>\n",
       "      <td>1225</td>\n",
       "      <td>19</td>\n",
       "      <td>64</td>\n",
       "      <td>8898.716275</td>\n",
       "      <td>3160.472914</td>\n",
       "      <td>13</td>\n",
       "      <td>11</td>\n",
       "      <td>0.406856</td>\n",
       "      <td>0.266854</td>\n",
       "      <td>1246.901868</td>\n",
       "      <td>128.186173</td>\n",
       "      <td>260.389902</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>114</td>\n",
       "      <td>95</td>\n",
       "      <td>135.522926</td>\n",
       "      <td>2651</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>3059.148818</td>\n",
       "      <td>1610.735904</td>\n",
       "      <td>14</td>\n",
       "      <td>8</td>\n",
       "      <td>0.841186</td>\n",
       "      <td>0.051121</td>\n",
       "      <td>382.393602</td>\n",
       "      <td>32.201567</td>\n",
       "      <td>113.953460</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>49</td>\n",
       "      <td>40</td>\n",
       "      <td>67.589221</td>\n",
       "      <td>15</td>\n",
       "      <td>322</td>\n",
       "      <td>7</td>\n",
       "      <td>913.591123</td>\n",
       "      <td>587.196542</td>\n",
       "      <td>7</td>\n",
       "      <td>3</td>\n",
       "      <td>0.724968</td>\n",
       "      <td>1.089743</td>\n",
       "      <td>304.530374</td>\n",
       "      <td>22.839778</td>\n",
       "      <td>93.351141</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>84</td>\n",
       "      <td>68</td>\n",
       "      <td>168.247020</td>\n",
       "      <td>1562</td>\n",
       "      <td>166</td>\n",
       "      <td>5</td>\n",
       "      <td>3950.202008</td>\n",
       "      <td>1219.555924</td>\n",
       "      <td>27</td>\n",
       "      <td>18</td>\n",
       "      <td>0.499266</td>\n",
       "      <td>0.107713</td>\n",
       "      <td>219.455667</td>\n",
       "      <td>58.091206</td>\n",
       "      <td>194.342970</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   sessions  drives  total_sessions  n_days_after_onboarding  \\\n",
       "0       243     200      296.748273                     2276   \n",
       "1       133     107      326.896596                     1225   \n",
       "2       114      95      135.522926                     2651   \n",
       "3        49      40       67.589221                       15   \n",
       "4        84      68      168.247020                     1562   \n",
       "\n",
       "   total_navigations_fav1  total_navigations_fav2  driven_km_drives  \\\n",
       "0                     208                       0       2628.845068   \n",
       "1                      19                      64       8898.716275   \n",
       "2                       0                       0       3059.148818   \n",
       "3                     322                       7        913.591123   \n",
       "4                     166                       5       3950.202008   \n",
       "\n",
       "   duration_minutes_drives  activity_days  driving_days  \\\n",
       "0              1985.775061             28            19   \n",
       "1              3160.472914             13            11   \n",
       "2              1610.735904             14             8   \n",
       "3               587.196542              7             3   \n",
       "4              1219.555924             27            18   \n",
       "\n",
       "   percent_sessions_in_month  total_sessions_per_day  \\\n",
       "0                   0.936313                0.130381   \n",
       "1                   0.406856                0.266854   \n",
       "2                   0.841186                0.051121   \n",
       "3                   0.724968                1.089743   \n",
       "4                   0.499266                0.107713   \n",
       "\n",
       "   kms_driven_per_day_during_the_month  km_per_drive  km_per_hour  \\\n",
       "0                           138.360267     11.632058    79.430298   \n",
       "1                          1246.901868    128.186173   260.389902   \n",
       "2                           382.393602     32.201567   113.953460   \n",
       "3                           304.530374     22.839778    93.351141   \n",
       "4                           219.455667     58.091206   194.342970   \n",
       "\n",
       "   professional_driver  label2  device2  \n",
       "0                    1       0        0  \n",
       "1                    1       0        1  \n",
       "2                    0       0        0  \n",
       "3                    0       0        1  \n",
       "4                    0       0        0  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Drop `device and label` columns since they don't contain any information relevant to churn\n",
    "df = df0.drop(['device', 'label'], axis=1)\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ajj50RCCaXrF"
   },
   "source": [
    "## Prepare data for modelling\n",
    "* choose evaluation metric: recall, since the logistic regression model had a terrible recall\n",
    "* no need to check for multicollinearity - DTs can handle them\n",
    "\n",
    "Split the data into:\n",
    "* features and target variable\n",
    "* training data(80%) and test data(20%)\n",
    "* split separate validation data. Set 75/25 into a training set and a validation set to get 75% of 80%\n",
    "* stratify= y to ensures that splitting maintains same proportion of different classes or groups that are present in the target variable y \n",
    "* set a random seed so work can be reproduced (42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "label2\n",
      "0    11763\n",
      "1     2536\n",
      "Name: count, dtype: int64\n",
      "label2\n",
      "0    0.822645\n",
      "1    0.177355\n",
      "Name: proportion, dtype: float64\n"
     ]
    }
   ],
   "source": [
    "# Check class balance for target variable label2\n",
    "print(df0['label2'].value_counts())\n",
    "print(df0['label2'].value_counts(normalize=True))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "w9vnV1wtaZWJ"
   },
   "source": [
    "Class rebalancing   \n",
    "~ 18% of the users in this dataset churned and ~82% retained. This is a slightly unbalanced dataset, but the DTs can handle it and so it can be modeled without any class rebalancing.    \n",
    "\n",
    "Metrics  \n",
    "*Accuracy* can be misleading if the data set is unbalanced. The algorithm could be poor at balancing false positive and false negatives because it does not account as well for the distribution of classes (churned/retained) as in the case of the logistic regression model. The model could be excellent at predicting one class (e.g., the majority[retained]) and terrible at predicting the other (e.g., the minority[churned]) and return a very good accuracy score. \n",
    "\n",
    "*Precision* In its goal to be precise (not classify people as churned when they are retained, the algorithm can become more selective. A higher rate of precision may mean that more samples are classified as retained, when they churn (false-negative/Type II errors). Thus, users set to churn can be missed.   \n",
    "\n",
    "*F1* balances both recall and precision, but here concentrating on recall would be more productive\n",
    "\n",
    "*Recall* is the chosen eveluation metric"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "id": "qLbapbSWDUL-"
   },
   "outputs": [],
   "source": [
    "# Subset data into dependent y(target) and independent X variables\n",
    "X = df.drop(columns=['label2'])\n",
    "y = df['label2']\n",
    "\n",
    "# Split into train and test sets\n",
    "X_tr, X_test, y_tr, y_test = train_test_split(X, y, stratify=y, test_size=0.2, random_state=42)\n",
    "\n",
    "# Split into train and validate sets\n",
    "X_train, X_val, y_train, y_val = train_test_split(X_tr, y_tr, stratify=y_tr, test_size=0.25, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "id": "qWIog8v_ckIg"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X\n",
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 14299 entries, 0 to 14298\n",
      "Data columns (total 17 columns):\n",
      " #   Column                               Non-Null Count  Dtype  \n",
      "---  ------                               --------------  -----  \n",
      " 0   sessions                             14299 non-null  int64  \n",
      " 1   drives                               14299 non-null  int64  \n",
      " 2   total_sessions                       14299 non-null  float64\n",
      " 3   n_days_after_onboarding              14299 non-null  int64  \n",
      " 4   total_navigations_fav1               14299 non-null  int64  \n",
      " 5   total_navigations_fav2               14299 non-null  int64  \n",
      " 6   driven_km_drives                     14299 non-null  float64\n",
      " 7   duration_minutes_drives              14299 non-null  float64\n",
      " 8   activity_days                        14299 non-null  int64  \n",
      " 9   driving_days                         14299 non-null  int64  \n",
      " 10  percent_sessions_in_month            14299 non-null  float64\n",
      " 11  total_sessions_per_day               14299 non-null  float64\n",
      " 12  kms_driven_per_day_during_the_month  14299 non-null  float64\n",
      " 13  km_per_drive                         14299 non-null  float64\n",
      " 14  km_per_hour                          14299 non-null  float64\n",
      " 15  professional_driver                  14299 non-null  int64  \n",
      " 16  device2                              14299 non-null  int64  \n",
      "dtypes: float64(8), int64(9)\n",
      "memory usage: 1.9 MB\n",
      "None\n",
      "y\n",
      "<class 'pandas.core.series.Series'>\n",
      "RangeIndex: 14299 entries, 0 to 14298\n",
      "Series name: label2\n",
      "Non-Null Count  Dtype\n",
      "--------------  -----\n",
      "14299 non-null  int64\n",
      "dtypes: int64(1)\n",
      "memory usage: 111.8 KB\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "# Verify number of samples in the partitioned data\n",
    "print(\"X\")\n",
    "print(X.info())\n",
    "print(\"y\")\n",
    "print(y.info())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((8579, 17), (2860, 17), (2860, 17), (8579,), (2860,), (2860,))"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Get balances of each training, validation, and testing sets\n",
    "X_train.shape, X_val.shape, X_test.shape, y_train.shape, y_val.shape, y_test.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "0x-4vGANcki4"
   },
   "source": [
    "This aligns with expectations."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Build random forest model\n",
    "* instantiate RF model\n",
    "* define and tune hyperparameters\n",
    "* define scoring criteria: use recall as metric\n",
    "* construct GridSearch cross-validation\n",
    "* fit model to training data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "id": "Vj5rJWOv5O3d"
   },
   "outputs": [],
   "source": [
    "# Instantiate the random forest classifier\n",
    "rf = RandomForestClassifier(random_state=42)\n",
    "\n",
    "# Determine set of hyperparameters\n",
    "cv_params = {'max_depth': [None, 2, 5, 6],\n",
    "             'max_features': [0.7, 2, 4],\n",
    "             'max_samples': [0.5, 0.7, 1],\n",
    "             'min_samples_leaf': [1, 2, 4],\n",
    "             'min_samples_split': [0.001, 0.01, 2, 4, 5], # small fractional values allows can help capture patterns in minority classes \n",
    "                                                        # fixed values (like 4) can prevent the creation of very small nodes that could lead to overfitting on the majority class.\n",
    "             'n_estimators': [50, 100, 125], # how deep the tree is allowed to grow\n",
    "             }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import precision_score, make_scorer\n",
    "\n",
    "# Define scoring criteria\n",
    "precision_scorer = make_scorer(precision_score, zero_division=0)\n",
    "                             # Instructs scikit-learn to assign a precision of 0 when no predicted positives, to migate continued percision warnings when fitting model\n",
    "scoring = {'accuracy': 'accuracy', 'precision': precision_scorer, 'recall': 'recall', 'f1': 'f1'}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Instantiate the GridSearchCV object\n",
    "# refit: use recall as metric\n",
    "RF_cv = GridSearchCV(rf, cv_params, scoring=scoring, cv=4, refit='recall')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "id": "OXuBiTGi5ZHn"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 2 µs, sys: 0 ns, total: 2 µs\n",
      "Wall time: 5.96 µs\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<style>#sk-container-id-1 {\n",
       "  /* Definition of color scheme common for light and dark mode */\n",
       "  --sklearn-color-text: black;\n",
       "  --sklearn-color-line: gray;\n",
       "  /* Definition of color scheme for unfitted estimators */\n",
       "  --sklearn-color-unfitted-level-0: #fff5e6;\n",
       "  --sklearn-color-unfitted-level-1: #f6e4d2;\n",
       "  --sklearn-color-unfitted-level-2: #ffe0b3;\n",
       "  --sklearn-color-unfitted-level-3: chocolate;\n",
       "  /* Definition of color scheme for fitted estimators */\n",
       "  --sklearn-color-fitted-level-0: #f0f8ff;\n",
       "  --sklearn-color-fitted-level-1: #d4ebff;\n",
       "  --sklearn-color-fitted-level-2: #b3dbfd;\n",
       "  --sklearn-color-fitted-level-3: cornflowerblue;\n",
       "\n",
       "  /* Specific color for light theme */\n",
       "  --sklearn-color-text-on-default-background: var(--sg-text-color, var(--theme-code-foreground, var(--jp-content-font-color1, black)));\n",
       "  --sklearn-color-background: var(--sg-background-color, var(--theme-background, var(--jp-layout-color0, white)));\n",
       "  --sklearn-color-border-box: var(--sg-text-color, var(--theme-code-foreground, var(--jp-content-font-color1, black)));\n",
       "  --sklearn-color-icon: #696969;\n",
       "\n",
       "  @media (prefers-color-scheme: dark) {\n",
       "    /* Redefinition of color scheme for dark theme */\n",
       "    --sklearn-color-text-on-default-background: var(--sg-text-color, var(--theme-code-foreground, var(--jp-content-font-color1, white)));\n",
       "    --sklearn-color-background: var(--sg-background-color, var(--theme-background, var(--jp-layout-color0, #111)));\n",
       "    --sklearn-color-border-box: var(--sg-text-color, var(--theme-code-foreground, var(--jp-content-font-color1, white)));\n",
       "    --sklearn-color-icon: #878787;\n",
       "  }\n",
       "}\n",
       "\n",
       "#sk-container-id-1 {\n",
       "  color: var(--sklearn-color-text);\n",
       "}\n",
       "\n",
       "#sk-container-id-1 pre {\n",
       "  padding: 0;\n",
       "}\n",
       "\n",
       "#sk-container-id-1 input.sk-hidden--visually {\n",
       "  border: 0;\n",
       "  clip: rect(1px 1px 1px 1px);\n",
       "  clip: rect(1px, 1px, 1px, 1px);\n",
       "  height: 1px;\n",
       "  margin: -1px;\n",
       "  overflow: hidden;\n",
       "  padding: 0;\n",
       "  position: absolute;\n",
       "  width: 1px;\n",
       "}\n",
       "\n",
       "#sk-container-id-1 div.sk-dashed-wrapped {\n",
       "  border: 1px dashed var(--sklearn-color-line);\n",
       "  margin: 0 0.4em 0.5em 0.4em;\n",
       "  box-sizing: border-box;\n",
       "  padding-bottom: 0.4em;\n",
       "  background-color: var(--sklearn-color-background);\n",
       "}\n",
       "\n",
       "#sk-container-id-1 div.sk-container {\n",
       "  /* jupyter's `normalize.less` sets `[hidden] { display: none; }`\n",
       "     but bootstrap.min.css set `[hidden] { display: none !important; }`\n",
       "     so we also need the `!important` here to be able to override the\n",
       "     default hidden behavior on the sphinx rendered scikit-learn.org.\n",
       "     See: https://github.com/scikit-learn/scikit-learn/issues/21755 */\n",
       "  display: inline-block !important;\n",
       "  position: relative;\n",
       "}\n",
       "\n",
       "#sk-container-id-1 div.sk-text-repr-fallback {\n",
       "  display: none;\n",
       "}\n",
       "\n",
       "div.sk-parallel-item,\n",
       "div.sk-serial,\n",
       "div.sk-item {\n",
       "  /* draw centered vertical line to link estimators */\n",
       "  background-image: linear-gradient(var(--sklearn-color-text-on-default-background), var(--sklearn-color-text-on-default-background));\n",
       "  background-size: 2px 100%;\n",
       "  background-repeat: no-repeat;\n",
       "  background-position: center center;\n",
       "}\n",
       "\n",
       "/* Parallel-specific style estimator block */\n",
       "\n",
       "#sk-container-id-1 div.sk-parallel-item::after {\n",
       "  content: \"\";\n",
       "  width: 100%;\n",
       "  border-bottom: 2px solid var(--sklearn-color-text-on-default-background);\n",
       "  flex-grow: 1;\n",
       "}\n",
       "\n",
       "#sk-container-id-1 div.sk-parallel {\n",
       "  display: flex;\n",
       "  align-items: stretch;\n",
       "  justify-content: center;\n",
       "  background-color: var(--sklearn-color-background);\n",
       "  position: relative;\n",
       "}\n",
       "\n",
       "#sk-container-id-1 div.sk-parallel-item {\n",
       "  display: flex;\n",
       "  flex-direction: column;\n",
       "}\n",
       "\n",
       "#sk-container-id-1 div.sk-parallel-item:first-child::after {\n",
       "  align-self: flex-end;\n",
       "  width: 50%;\n",
       "}\n",
       "\n",
       "#sk-container-id-1 div.sk-parallel-item:last-child::after {\n",
       "  align-self: flex-start;\n",
       "  width: 50%;\n",
       "}\n",
       "\n",
       "#sk-container-id-1 div.sk-parallel-item:only-child::after {\n",
       "  width: 0;\n",
       "}\n",
       "\n",
       "/* Serial-specific style estimator block */\n",
       "\n",
       "#sk-container-id-1 div.sk-serial {\n",
       "  display: flex;\n",
       "  flex-direction: column;\n",
       "  align-items: center;\n",
       "  background-color: var(--sklearn-color-background);\n",
       "  padding-right: 1em;\n",
       "  padding-left: 1em;\n",
       "}\n",
       "\n",
       "\n",
       "/* Toggleable style: style used for estimator/Pipeline/ColumnTransformer box that is\n",
       "clickable and can be expanded/collapsed.\n",
       "- Pipeline and ColumnTransformer use this feature and define the default style\n",
       "- Estimators will overwrite some part of the style using the `sk-estimator` class\n",
       "*/\n",
       "\n",
       "/* Pipeline and ColumnTransformer style (default) */\n",
       "\n",
       "#sk-container-id-1 div.sk-toggleable {\n",
       "  /* Default theme specific background. It is overwritten whether we have a\n",
       "  specific estimator or a Pipeline/ColumnTransformer */\n",
       "  background-color: var(--sklearn-color-background);\n",
       "}\n",
       "\n",
       "/* Toggleable label */\n",
       "#sk-container-id-1 label.sk-toggleable__label {\n",
       "  cursor: pointer;\n",
       "  display: block;\n",
       "  width: 100%;\n",
       "  margin-bottom: 0;\n",
       "  padding: 0.5em;\n",
       "  box-sizing: border-box;\n",
       "  text-align: center;\n",
       "}\n",
       "\n",
       "#sk-container-id-1 label.sk-toggleable__label-arrow:before {\n",
       "  /* Arrow on the left of the label */\n",
       "  content: \"▸\";\n",
       "  float: left;\n",
       "  margin-right: 0.25em;\n",
       "  color: var(--sklearn-color-icon);\n",
       "}\n",
       "\n",
       "#sk-container-id-1 label.sk-toggleable__label-arrow:hover:before {\n",
       "  color: var(--sklearn-color-text);\n",
       "}\n",
       "\n",
       "/* Toggleable content - dropdown */\n",
       "\n",
       "#sk-container-id-1 div.sk-toggleable__content {\n",
       "  max-height: 0;\n",
       "  max-width: 0;\n",
       "  overflow: hidden;\n",
       "  text-align: left;\n",
       "  /* unfitted */\n",
       "  background-color: var(--sklearn-color-unfitted-level-0);\n",
       "}\n",
       "\n",
       "#sk-container-id-1 div.sk-toggleable__content.fitted {\n",
       "  /* fitted */\n",
       "  background-color: var(--sklearn-color-fitted-level-0);\n",
       "}\n",
       "\n",
       "#sk-container-id-1 div.sk-toggleable__content pre {\n",
       "  margin: 0.2em;\n",
       "  border-radius: 0.25em;\n",
       "  color: var(--sklearn-color-text);\n",
       "  /* unfitted */\n",
       "  background-color: var(--sklearn-color-unfitted-level-0);\n",
       "}\n",
       "\n",
       "#sk-container-id-1 div.sk-toggleable__content.fitted pre {\n",
       "  /* unfitted */\n",
       "  background-color: var(--sklearn-color-fitted-level-0);\n",
       "}\n",
       "\n",
       "#sk-container-id-1 input.sk-toggleable__control:checked~div.sk-toggleable__content {\n",
       "  /* Expand drop-down */\n",
       "  max-height: 200px;\n",
       "  max-width: 100%;\n",
       "  overflow: auto;\n",
       "}\n",
       "\n",
       "#sk-container-id-1 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {\n",
       "  content: \"▾\";\n",
       "}\n",
       "\n",
       "/* Pipeline/ColumnTransformer-specific style */\n",
       "\n",
       "#sk-container-id-1 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {\n",
       "  color: var(--sklearn-color-text);\n",
       "  background-color: var(--sklearn-color-unfitted-level-2);\n",
       "}\n",
       "\n",
       "#sk-container-id-1 div.sk-label.fitted input.sk-toggleable__control:checked~label.sk-toggleable__label {\n",
       "  background-color: var(--sklearn-color-fitted-level-2);\n",
       "}\n",
       "\n",
       "/* Estimator-specific style */\n",
       "\n",
       "/* Colorize estimator box */\n",
       "#sk-container-id-1 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {\n",
       "  /* unfitted */\n",
       "  background-color: var(--sklearn-color-unfitted-level-2);\n",
       "}\n",
       "\n",
       "#sk-container-id-1 div.sk-estimator.fitted input.sk-toggleable__control:checked~label.sk-toggleable__label {\n",
       "  /* fitted */\n",
       "  background-color: var(--sklearn-color-fitted-level-2);\n",
       "}\n",
       "\n",
       "#sk-container-id-1 div.sk-label label.sk-toggleable__label,\n",
       "#sk-container-id-1 div.sk-label label {\n",
       "  /* The background is the default theme color */\n",
       "  color: var(--sklearn-color-text-on-default-background);\n",
       "}\n",
       "\n",
       "/* On hover, darken the color of the background */\n",
       "#sk-container-id-1 div.sk-label:hover label.sk-toggleable__label {\n",
       "  color: var(--sklearn-color-text);\n",
       "  background-color: var(--sklearn-color-unfitted-level-2);\n",
       "}\n",
       "\n",
       "/* Label box, darken color on hover, fitted */\n",
       "#sk-container-id-1 div.sk-label.fitted:hover label.sk-toggleable__label.fitted {\n",
       "  color: var(--sklearn-color-text);\n",
       "  background-color: var(--sklearn-color-fitted-level-2);\n",
       "}\n",
       "\n",
       "/* Estimator label */\n",
       "\n",
       "#sk-container-id-1 div.sk-label label {\n",
       "  font-family: monospace;\n",
       "  font-weight: bold;\n",
       "  display: inline-block;\n",
       "  line-height: 1.2em;\n",
       "}\n",
       "\n",
       "#sk-container-id-1 div.sk-label-container {\n",
       "  text-align: center;\n",
       "}\n",
       "\n",
       "/* Estimator-specific */\n",
       "#sk-container-id-1 div.sk-estimator {\n",
       "  font-family: monospace;\n",
       "  border: 1px dotted var(--sklearn-color-border-box);\n",
       "  border-radius: 0.25em;\n",
       "  box-sizing: border-box;\n",
       "  margin-bottom: 0.5em;\n",
       "  /* unfitted */\n",
       "  background-color: var(--sklearn-color-unfitted-level-0);\n",
       "}\n",
       "\n",
       "#sk-container-id-1 div.sk-estimator.fitted {\n",
       "  /* fitted */\n",
       "  background-color: var(--sklearn-color-fitted-level-0);\n",
       "}\n",
       "\n",
       "/* on hover */\n",
       "#sk-container-id-1 div.sk-estimator:hover {\n",
       "  /* unfitted */\n",
       "  background-color: var(--sklearn-color-unfitted-level-2);\n",
       "}\n",
       "\n",
       "#sk-container-id-1 div.sk-estimator.fitted:hover {\n",
       "  /* fitted */\n",
       "  background-color: var(--sklearn-color-fitted-level-2);\n",
       "}\n",
       "\n",
       "/* Specification for estimator info (e.g. \"i\" and \"?\") */\n",
       "\n",
       "/* Common style for \"i\" and \"?\" */\n",
       "\n",
       ".sk-estimator-doc-link,\n",
       "a:link.sk-estimator-doc-link,\n",
       "a:visited.sk-estimator-doc-link {\n",
       "  float: right;\n",
       "  font-size: smaller;\n",
       "  line-height: 1em;\n",
       "  font-family: monospace;\n",
       "  background-color: var(--sklearn-color-background);\n",
       "  border-radius: 1em;\n",
       "  height: 1em;\n",
       "  width: 1em;\n",
       "  text-decoration: none !important;\n",
       "  margin-left: 1ex;\n",
       "  /* unfitted */\n",
       "  border: var(--sklearn-color-unfitted-level-1) 1pt solid;\n",
       "  color: var(--sklearn-color-unfitted-level-1);\n",
       "}\n",
       "\n",
       ".sk-estimator-doc-link.fitted,\n",
       "a:link.sk-estimator-doc-link.fitted,\n",
       "a:visited.sk-estimator-doc-link.fitted {\n",
       "  /* fitted */\n",
       "  border: var(--sklearn-color-fitted-level-1) 1pt solid;\n",
       "  color: var(--sklearn-color-fitted-level-1);\n",
       "}\n",
       "\n",
       "/* On hover */\n",
       "div.sk-estimator:hover .sk-estimator-doc-link:hover,\n",
       ".sk-estimator-doc-link:hover,\n",
       "div.sk-label-container:hover .sk-estimator-doc-link:hover,\n",
       ".sk-estimator-doc-link:hover {\n",
       "  /* unfitted */\n",
       "  background-color: var(--sklearn-color-unfitted-level-3);\n",
       "  color: var(--sklearn-color-background);\n",
       "  text-decoration: none;\n",
       "}\n",
       "\n",
       "div.sk-estimator.fitted:hover .sk-estimator-doc-link.fitted:hover,\n",
       ".sk-estimator-doc-link.fitted:hover,\n",
       "div.sk-label-container:hover .sk-estimator-doc-link.fitted:hover,\n",
       ".sk-estimator-doc-link.fitted:hover {\n",
       "  /* fitted */\n",
       "  background-color: var(--sklearn-color-fitted-level-3);\n",
       "  color: var(--sklearn-color-background);\n",
       "  text-decoration: none;\n",
       "}\n",
       "\n",
       "/* Span, style for the box shown on hovering the info icon */\n",
       ".sk-estimator-doc-link span {\n",
       "  display: none;\n",
       "  z-index: 9999;\n",
       "  position: relative;\n",
       "  font-weight: normal;\n",
       "  right: .2ex;\n",
       "  padding: .5ex;\n",
       "  margin: .5ex;\n",
       "  width: min-content;\n",
       "  min-width: 20ex;\n",
       "  max-width: 50ex;\n",
       "  color: var(--sklearn-color-text);\n",
       "  box-shadow: 2pt 2pt 4pt #999;\n",
       "  /* unfitted */\n",
       "  background: var(--sklearn-color-unfitted-level-0);\n",
       "  border: .5pt solid var(--sklearn-color-unfitted-level-3);\n",
       "}\n",
       "\n",
       ".sk-estimator-doc-link.fitted span {\n",
       "  /* fitted */\n",
       "  background: var(--sklearn-color-fitted-level-0);\n",
       "  border: var(--sklearn-color-fitted-level-3);\n",
       "}\n",
       "\n",
       ".sk-estimator-doc-link:hover span {\n",
       "  display: block;\n",
       "}\n",
       "\n",
       "/* \"?\"-specific style due to the `<a>` HTML tag */\n",
       "\n",
       "#sk-container-id-1 a.estimator_doc_link {\n",
       "  float: right;\n",
       "  font-size: 1rem;\n",
       "  line-height: 1em;\n",
       "  font-family: monospace;\n",
       "  background-color: var(--sklearn-color-background);\n",
       "  border-radius: 1rem;\n",
       "  height: 1rem;\n",
       "  width: 1rem;\n",
       "  text-decoration: none;\n",
       "  /* unfitted */\n",
       "  color: var(--sklearn-color-unfitted-level-1);\n",
       "  border: var(--sklearn-color-unfitted-level-1) 1pt solid;\n",
       "}\n",
       "\n",
       "#sk-container-id-1 a.estimator_doc_link.fitted {\n",
       "  /* fitted */\n",
       "  border: var(--sklearn-color-fitted-level-1) 1pt solid;\n",
       "  color: var(--sklearn-color-fitted-level-1);\n",
       "}\n",
       "\n",
       "/* On hover */\n",
       "#sk-container-id-1 a.estimator_doc_link:hover {\n",
       "  /* unfitted */\n",
       "  background-color: var(--sklearn-color-unfitted-level-3);\n",
       "  color: var(--sklearn-color-background);\n",
       "  text-decoration: none;\n",
       "}\n",
       "\n",
       "#sk-container-id-1 a.estimator_doc_link.fitted:hover {\n",
       "  /* fitted */\n",
       "  background-color: var(--sklearn-color-fitted-level-3);\n",
       "}\n",
       "</style><div id=\"sk-container-id-1\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>GridSearchCV(cv=4, estimator=RandomForestClassifier(random_state=42),\n",
       "             param_grid={&#x27;max_depth&#x27;: [None, 2, 5, 6],\n",
       "                         &#x27;max_features&#x27;: [0.7, 2, 4],\n",
       "                         &#x27;max_samples&#x27;: [0.5, 0.7, 1],\n",
       "                         &#x27;min_samples_leaf&#x27;: [1, 2, 4],\n",
       "                         &#x27;min_samples_split&#x27;: [0.001, 0.01, 2, 4, 5],\n",
       "                         &#x27;n_estimators&#x27;: [50, 100, 125]},\n",
       "             refit=&#x27;recall&#x27;,\n",
       "             scoring={&#x27;accuracy&#x27;: &#x27;accuracy&#x27;, &#x27;f1&#x27;: &#x27;f1&#x27;,\n",
       "                      &#x27;precision&#x27;: make_scorer(precision_score, response_method=&#x27;predict&#x27;, zero_division=0),\n",
       "                      &#x27;recall&#x27;: &#x27;recall&#x27;})</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label fitted sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-1\" type=\"checkbox\" ><label for=\"sk-estimator-id-1\" class=\"sk-toggleable__label fitted sk-toggleable__label-arrow fitted\">&nbsp;&nbsp;GridSearchCV<a class=\"sk-estimator-doc-link fitted\" rel=\"noreferrer\" target=\"_blank\" href=\"https://scikit-learn.org/1.4/modules/generated/sklearn.model_selection.GridSearchCV.html\">?<span>Documentation for GridSearchCV</span></a><span class=\"sk-estimator-doc-link fitted\">i<span>Fitted</span></span></label><div class=\"sk-toggleable__content fitted\"><pre>GridSearchCV(cv=4, estimator=RandomForestClassifier(random_state=42),\n",
       "             param_grid={&#x27;max_depth&#x27;: [None, 2, 5, 6],\n",
       "                         &#x27;max_features&#x27;: [0.7, 2, 4],\n",
       "                         &#x27;max_samples&#x27;: [0.5, 0.7, 1],\n",
       "                         &#x27;min_samples_leaf&#x27;: [1, 2, 4],\n",
       "                         &#x27;min_samples_split&#x27;: [0.001, 0.01, 2, 4, 5],\n",
       "                         &#x27;n_estimators&#x27;: [50, 100, 125]},\n",
       "             refit=&#x27;recall&#x27;,\n",
       "             scoring={&#x27;accuracy&#x27;: &#x27;accuracy&#x27;, &#x27;f1&#x27;: &#x27;f1&#x27;,\n",
       "                      &#x27;precision&#x27;: make_scorer(precision_score, response_method=&#x27;predict&#x27;, zero_division=0),\n",
       "                      &#x27;recall&#x27;: &#x27;recall&#x27;})</pre></div> </div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label fitted sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-2\" type=\"checkbox\" ><label for=\"sk-estimator-id-2\" class=\"sk-toggleable__label fitted sk-toggleable__label-arrow fitted\">estimator: RandomForestClassifier</label><div class=\"sk-toggleable__content fitted\"><pre>RandomForestClassifier(random_state=42)</pre></div> </div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator fitted sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-3\" type=\"checkbox\" ><label for=\"sk-estimator-id-3\" class=\"sk-toggleable__label fitted sk-toggleable__label-arrow fitted\">&nbsp;RandomForestClassifier<a class=\"sk-estimator-doc-link fitted\" rel=\"noreferrer\" target=\"_blank\" href=\"https://scikit-learn.org/1.4/modules/generated/sklearn.ensemble.RandomForestClassifier.html\">?<span>Documentation for RandomForestClassifier</span></a></label><div class=\"sk-toggleable__content fitted\"><pre>RandomForestClassifier(random_state=42)</pre></div> </div></div></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "GridSearchCV(cv=4, estimator=RandomForestClassifier(random_state=42),\n",
       "             param_grid={'max_depth': [None, 2, 5, 6],\n",
       "                         'max_features': [0.7, 2, 4],\n",
       "                         'max_samples': [0.5, 0.7, 1],\n",
       "                         'min_samples_leaf': [1, 2, 4],\n",
       "                         'min_samples_split': [0.001, 0.01, 2, 4, 5],\n",
       "                         'n_estimators': [50, 100, 125]},\n",
       "             refit='recall',\n",
       "             scoring={'accuracy': 'accuracy', 'f1': 'f1',\n",
       "                      'precision': make_scorer(precision_score, response_method='predict', zero_division=0),\n",
       "                      'recall': 'recall'})"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Fit model to the training data & call magic function time to time execution\n",
    "%time\n",
    "RF_cv.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "id": "YtAgrH0zy4CE"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.12876260533222822"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Examine best average score across all the validation folds\n",
    "RF_cv.best_score_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "id": "kazNtYG4fQOI"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'max_depth': None,\n",
       " 'max_features': 0.7,\n",
       " 'max_samples': 0.5,\n",
       " 'min_samples_leaf': 1,\n",
       " 'min_samples_split': 5,\n",
       " 'n_estimators': 50}"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Examine best combination of hyperparameters\n",
    "RF_cv.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create a metrics results table with make_results function which returns a df with metrics for the RF model with the best mean 'metric' score across all validation folds\n",
    "# In this case: the best mean recall score\n",
    "\n",
    "def make_results(model_name:str, model_object, metric:str):  # model_name: name, model_object: a fit GridSearchCV object, metric:str\n",
    "    \n",
    "    # Create dictionary that maps input metric to actual metric name in GridSearchCV\n",
    "    metric_dict = {'precision': 'mean_test_precision',\n",
    "                   'recall': 'mean_test_recall',\n",
    "                   'f1': 'mean_test_f1',\n",
    "                   'accuracy': 'mean_test_accuracy',\n",
    "                   }\n",
    "    \n",
    "    cv_results = pd.DataFrame(model_object.cv_results_) # Create df for all results\n",
    "    \n",
    "    best_estimator_results = cv_results.iloc[cv_results[metric_dict[metric]].idxmax(), :] # idxmax() returns a series with the index of the max value for each col\n",
    "    \n",
    "    f1 = best_estimator_results.mean_test_f1\n",
    "    recall = best_estimator_results.mean_test_recall\n",
    "    precision = best_estimator_results.mean_test_precision\n",
    "    accuracy = best_estimator_results.mean_test_accuracy\n",
    "\n",
    "# Create table of best estimator row results\n",
    "    table = pd.DataFrame({'Model': [model_name],\n",
    "                          'Precision': [precision],\n",
    "                          'Recall': [recall],\n",
    "                          'Accuracy': [accuracy],\n",
    "                          'F1': [f1]\n",
    "                         },)\n",
    "    return table"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "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>Model</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>F1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Random Forest cv</td>\n",
       "      <td>0.454665</td>\n",
       "      <td>0.128763</td>\n",
       "      <td>0.81816</td>\n",
       "      <td>0.200545</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              Model  Precision    Recall  Accuracy        F1\n",
       "0  Random Forest cv   0.454665  0.128763   0.81816  0.200545"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Pass the GridSearch object to the make_results() function.\n",
    "RF_cv_results = make_results('Random Forest cv', RF_cv, 'recall')\n",
    "RF_cv_results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "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>Model</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>F1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Logistic Regression</td>\n",
       "      <td>0.535088</td>\n",
       "      <td>0.096215</td>\n",
       "      <td>0.824895</td>\n",
       "      <td>0.163102</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 Model  Precision    Recall  Accuracy        F1\n",
       "0  Logistic Regression   0.535088  0.096215  0.824895  0.163102"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Read in master results table\n",
    "models_table = pd.read_csv('Retention_Results.csv')\n",
    "models_table"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "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>Model</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>F1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Logistic Regression</td>\n",
       "      <td>0.535088</td>\n",
       "      <td>0.096215</td>\n",
       "      <td>0.824895</td>\n",
       "      <td>0.163102</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Random Forest cv</td>\n",
       "      <td>0.454665</td>\n",
       "      <td>0.128763</td>\n",
       "      <td>0.818160</td>\n",
       "      <td>0.200545</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 Model  Precision    Recall  Accuracy        F1\n",
       "0  Logistic Regression   0.535088  0.096215  0.824895  0.163102\n",
       "0     Random Forest cv   0.454665  0.128763  0.818160  0.200545"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Concatentate decision tree model results table with master results table\n",
    "models_table = pd.concat([models_table, RF_cv_results])\n",
    "models_table"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "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>Model</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>F1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Logistic Regression</td>\n",
       "      <td>0.535088</td>\n",
       "      <td>0.096215</td>\n",
       "      <td>0.824895</td>\n",
       "      <td>0.163102</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Random Forest cv</td>\n",
       "      <td>0.454665</td>\n",
       "      <td>0.128763</td>\n",
       "      <td>0.818160</td>\n",
       "      <td>0.200545</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 Model  Precision    Recall  Accuracy        F1\n",
       "0  Logistic Regression   0.535088  0.096215  0.824895  0.163102\n",
       "0     Random Forest cv   0.454665  0.128763  0.818160  0.200545"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Save the master results table\n",
    "models_table.to_csv('Retention_Results.csv', index=False)\n",
    "models_table.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Pickle the model (save)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Pickle the model\n",
    "with open('RF_cv_WAZE.pickle', 'wb') as to_write: # wb = write binary\n",
    "    pickle.dump(RF_cv, to_write)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Read in pickled model\n",
    "with open('RF_cv_WAZE.pickle', 'rb') as to_read: # rb = read binary\n",
    "    RF_cv = pickle.load(to_read)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Build an XGBoost model\n",
    "* instantiate XGBoost model\n",
    "* define and tune hyperparameters\n",
    "* define scoring criteria: use recall as metric\n",
    "* construct GridSearch cross-validation\n",
    "* fit model to training data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Instantiate the XGBoost classifier\n",
    "xgb = XGBClassifier(objective='binary:logistic', random_state=42)\n",
    "\n",
    "# Determine set of hyperparameters\n",
    "cv_params = {'max_depth': [2, 6, 12],\n",
    "             'min_child_weight': [0.5, 3, 5],\n",
    "             'learning_rate': [0.01, 0.1],\n",
    "             'n_estimators': [150, 200, 300]\n",
    "             }\n",
    "\n",
    "from sklearn.metrics import precision_score, make_scorer\n",
    "\n",
    "# Define scoring criteria\n",
    "precision_scorer = make_scorer(precision_score, zero_division=0)\n",
    "                             # Instructs scikit-learn to assign a precision of 0 when no predicted positives, to migate continued percision warnings when fitting model\n",
    "scoring = {'accuracy': 'accuracy', 'precision': precision_scorer, 'recall': 'recall', 'f1': 'f1'}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Instantiate the GridSearchCV object\n",
    "# refit: use recall as metric\n",
    "XGB_cv = GridSearchCV(xgb, cv_params, scoring=scoring, cv=4, refit='recall')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 2 µs, sys: 0 ns, total: 2 µs\n",
      "Wall time: 3.81 µs\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<style>#sk-container-id-2 {\n",
       "  /* Definition of color scheme common for light and dark mode */\n",
       "  --sklearn-color-text: black;\n",
       "  --sklearn-color-line: gray;\n",
       "  /* Definition of color scheme for unfitted estimators */\n",
       "  --sklearn-color-unfitted-level-0: #fff5e6;\n",
       "  --sklearn-color-unfitted-level-1: #f6e4d2;\n",
       "  --sklearn-color-unfitted-level-2: #ffe0b3;\n",
       "  --sklearn-color-unfitted-level-3: chocolate;\n",
       "  /* Definition of color scheme for fitted estimators */\n",
       "  --sklearn-color-fitted-level-0: #f0f8ff;\n",
       "  --sklearn-color-fitted-level-1: #d4ebff;\n",
       "  --sklearn-color-fitted-level-2: #b3dbfd;\n",
       "  --sklearn-color-fitted-level-3: cornflowerblue;\n",
       "\n",
       "  /* Specific color for light theme */\n",
       "  --sklearn-color-text-on-default-background: var(--sg-text-color, var(--theme-code-foreground, var(--jp-content-font-color1, black)));\n",
       "  --sklearn-color-background: var(--sg-background-color, var(--theme-background, var(--jp-layout-color0, white)));\n",
       "  --sklearn-color-border-box: var(--sg-text-color, var(--theme-code-foreground, var(--jp-content-font-color1, black)));\n",
       "  --sklearn-color-icon: #696969;\n",
       "\n",
       "  @media (prefers-color-scheme: dark) {\n",
       "    /* Redefinition of color scheme for dark theme */\n",
       "    --sklearn-color-text-on-default-background: var(--sg-text-color, var(--theme-code-foreground, var(--jp-content-font-color1, white)));\n",
       "    --sklearn-color-background: var(--sg-background-color, var(--theme-background, var(--jp-layout-color0, #111)));\n",
       "    --sklearn-color-border-box: var(--sg-text-color, var(--theme-code-foreground, var(--jp-content-font-color1, white)));\n",
       "    --sklearn-color-icon: #878787;\n",
       "  }\n",
       "}\n",
       "\n",
       "#sk-container-id-2 {\n",
       "  color: var(--sklearn-color-text);\n",
       "}\n",
       "\n",
       "#sk-container-id-2 pre {\n",
       "  padding: 0;\n",
       "}\n",
       "\n",
       "#sk-container-id-2 input.sk-hidden--visually {\n",
       "  border: 0;\n",
       "  clip: rect(1px 1px 1px 1px);\n",
       "  clip: rect(1px, 1px, 1px, 1px);\n",
       "  height: 1px;\n",
       "  margin: -1px;\n",
       "  overflow: hidden;\n",
       "  padding: 0;\n",
       "  position: absolute;\n",
       "  width: 1px;\n",
       "}\n",
       "\n",
       "#sk-container-id-2 div.sk-dashed-wrapped {\n",
       "  border: 1px dashed var(--sklearn-color-line);\n",
       "  margin: 0 0.4em 0.5em 0.4em;\n",
       "  box-sizing: border-box;\n",
       "  padding-bottom: 0.4em;\n",
       "  background-color: var(--sklearn-color-background);\n",
       "}\n",
       "\n",
       "#sk-container-id-2 div.sk-container {\n",
       "  /* jupyter's `normalize.less` sets `[hidden] { display: none; }`\n",
       "     but bootstrap.min.css set `[hidden] { display: none !important; }`\n",
       "     so we also need the `!important` here to be able to override the\n",
       "     default hidden behavior on the sphinx rendered scikit-learn.org.\n",
       "     See: https://github.com/scikit-learn/scikit-learn/issues/21755 */\n",
       "  display: inline-block !important;\n",
       "  position: relative;\n",
       "}\n",
       "\n",
       "#sk-container-id-2 div.sk-text-repr-fallback {\n",
       "  display: none;\n",
       "}\n",
       "\n",
       "div.sk-parallel-item,\n",
       "div.sk-serial,\n",
       "div.sk-item {\n",
       "  /* draw centered vertical line to link estimators */\n",
       "  background-image: linear-gradient(var(--sklearn-color-text-on-default-background), var(--sklearn-color-text-on-default-background));\n",
       "  background-size: 2px 100%;\n",
       "  background-repeat: no-repeat;\n",
       "  background-position: center center;\n",
       "}\n",
       "\n",
       "/* Parallel-specific style estimator block */\n",
       "\n",
       "#sk-container-id-2 div.sk-parallel-item::after {\n",
       "  content: \"\";\n",
       "  width: 100%;\n",
       "  border-bottom: 2px solid var(--sklearn-color-text-on-default-background);\n",
       "  flex-grow: 1;\n",
       "}\n",
       "\n",
       "#sk-container-id-2 div.sk-parallel {\n",
       "  display: flex;\n",
       "  align-items: stretch;\n",
       "  justify-content: center;\n",
       "  background-color: var(--sklearn-color-background);\n",
       "  position: relative;\n",
       "}\n",
       "\n",
       "#sk-container-id-2 div.sk-parallel-item {\n",
       "  display: flex;\n",
       "  flex-direction: column;\n",
       "}\n",
       "\n",
       "#sk-container-id-2 div.sk-parallel-item:first-child::after {\n",
       "  align-self: flex-end;\n",
       "  width: 50%;\n",
       "}\n",
       "\n",
       "#sk-container-id-2 div.sk-parallel-item:last-child::after {\n",
       "  align-self: flex-start;\n",
       "  width: 50%;\n",
       "}\n",
       "\n",
       "#sk-container-id-2 div.sk-parallel-item:only-child::after {\n",
       "  width: 0;\n",
       "}\n",
       "\n",
       "/* Serial-specific style estimator block */\n",
       "\n",
       "#sk-container-id-2 div.sk-serial {\n",
       "  display: flex;\n",
       "  flex-direction: column;\n",
       "  align-items: center;\n",
       "  background-color: var(--sklearn-color-background);\n",
       "  padding-right: 1em;\n",
       "  padding-left: 1em;\n",
       "}\n",
       "\n",
       "\n",
       "/* Toggleable style: style used for estimator/Pipeline/ColumnTransformer box that is\n",
       "clickable and can be expanded/collapsed.\n",
       "- Pipeline and ColumnTransformer use this feature and define the default style\n",
       "- Estimators will overwrite some part of the style using the `sk-estimator` class\n",
       "*/\n",
       "\n",
       "/* Pipeline and ColumnTransformer style (default) */\n",
       "\n",
       "#sk-container-id-2 div.sk-toggleable {\n",
       "  /* Default theme specific background. It is overwritten whether we have a\n",
       "  specific estimator or a Pipeline/ColumnTransformer */\n",
       "  background-color: var(--sklearn-color-background);\n",
       "}\n",
       "\n",
       "/* Toggleable label */\n",
       "#sk-container-id-2 label.sk-toggleable__label {\n",
       "  cursor: pointer;\n",
       "  display: block;\n",
       "  width: 100%;\n",
       "  margin-bottom: 0;\n",
       "  padding: 0.5em;\n",
       "  box-sizing: border-box;\n",
       "  text-align: center;\n",
       "}\n",
       "\n",
       "#sk-container-id-2 label.sk-toggleable__label-arrow:before {\n",
       "  /* Arrow on the left of the label */\n",
       "  content: \"▸\";\n",
       "  float: left;\n",
       "  margin-right: 0.25em;\n",
       "  color: var(--sklearn-color-icon);\n",
       "}\n",
       "\n",
       "#sk-container-id-2 label.sk-toggleable__label-arrow:hover:before {\n",
       "  color: var(--sklearn-color-text);\n",
       "}\n",
       "\n",
       "/* Toggleable content - dropdown */\n",
       "\n",
       "#sk-container-id-2 div.sk-toggleable__content {\n",
       "  max-height: 0;\n",
       "  max-width: 0;\n",
       "  overflow: hidden;\n",
       "  text-align: left;\n",
       "  /* unfitted */\n",
       "  background-color: var(--sklearn-color-unfitted-level-0);\n",
       "}\n",
       "\n",
       "#sk-container-id-2 div.sk-toggleable__content.fitted {\n",
       "  /* fitted */\n",
       "  background-color: var(--sklearn-color-fitted-level-0);\n",
       "}\n",
       "\n",
       "#sk-container-id-2 div.sk-toggleable__content pre {\n",
       "  margin: 0.2em;\n",
       "  border-radius: 0.25em;\n",
       "  color: var(--sklearn-color-text);\n",
       "  /* unfitted */\n",
       "  background-color: var(--sklearn-color-unfitted-level-0);\n",
       "}\n",
       "\n",
       "#sk-container-id-2 div.sk-toggleable__content.fitted pre {\n",
       "  /* unfitted */\n",
       "  background-color: var(--sklearn-color-fitted-level-0);\n",
       "}\n",
       "\n",
       "#sk-container-id-2 input.sk-toggleable__control:checked~div.sk-toggleable__content {\n",
       "  /* Expand drop-down */\n",
       "  max-height: 200px;\n",
       "  max-width: 100%;\n",
       "  overflow: auto;\n",
       "}\n",
       "\n",
       "#sk-container-id-2 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {\n",
       "  content: \"▾\";\n",
       "}\n",
       "\n",
       "/* Pipeline/ColumnTransformer-specific style */\n",
       "\n",
       "#sk-container-id-2 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {\n",
       "  color: var(--sklearn-color-text);\n",
       "  background-color: var(--sklearn-color-unfitted-level-2);\n",
       "}\n",
       "\n",
       "#sk-container-id-2 div.sk-label.fitted input.sk-toggleable__control:checked~label.sk-toggleable__label {\n",
       "  background-color: var(--sklearn-color-fitted-level-2);\n",
       "}\n",
       "\n",
       "/* Estimator-specific style */\n",
       "\n",
       "/* Colorize estimator box */\n",
       "#sk-container-id-2 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {\n",
       "  /* unfitted */\n",
       "  background-color: var(--sklearn-color-unfitted-level-2);\n",
       "}\n",
       "\n",
       "#sk-container-id-2 div.sk-estimator.fitted input.sk-toggleable__control:checked~label.sk-toggleable__label {\n",
       "  /* fitted */\n",
       "  background-color: var(--sklearn-color-fitted-level-2);\n",
       "}\n",
       "\n",
       "#sk-container-id-2 div.sk-label label.sk-toggleable__label,\n",
       "#sk-container-id-2 div.sk-label label {\n",
       "  /* The background is the default theme color */\n",
       "  color: var(--sklearn-color-text-on-default-background);\n",
       "}\n",
       "\n",
       "/* On hover, darken the color of the background */\n",
       "#sk-container-id-2 div.sk-label:hover label.sk-toggleable__label {\n",
       "  color: var(--sklearn-color-text);\n",
       "  background-color: var(--sklearn-color-unfitted-level-2);\n",
       "}\n",
       "\n",
       "/* Label box, darken color on hover, fitted */\n",
       "#sk-container-id-2 div.sk-label.fitted:hover label.sk-toggleable__label.fitted {\n",
       "  color: var(--sklearn-color-text);\n",
       "  background-color: var(--sklearn-color-fitted-level-2);\n",
       "}\n",
       "\n",
       "/* Estimator label */\n",
       "\n",
       "#sk-container-id-2 div.sk-label label {\n",
       "  font-family: monospace;\n",
       "  font-weight: bold;\n",
       "  display: inline-block;\n",
       "  line-height: 1.2em;\n",
       "}\n",
       "\n",
       "#sk-container-id-2 div.sk-label-container {\n",
       "  text-align: center;\n",
       "}\n",
       "\n",
       "/* Estimator-specific */\n",
       "#sk-container-id-2 div.sk-estimator {\n",
       "  font-family: monospace;\n",
       "  border: 1px dotted var(--sklearn-color-border-box);\n",
       "  border-radius: 0.25em;\n",
       "  box-sizing: border-box;\n",
       "  margin-bottom: 0.5em;\n",
       "  /* unfitted */\n",
       "  background-color: var(--sklearn-color-unfitted-level-0);\n",
       "}\n",
       "\n",
       "#sk-container-id-2 div.sk-estimator.fitted {\n",
       "  /* fitted */\n",
       "  background-color: var(--sklearn-color-fitted-level-0);\n",
       "}\n",
       "\n",
       "/* on hover */\n",
       "#sk-container-id-2 div.sk-estimator:hover {\n",
       "  /* unfitted */\n",
       "  background-color: var(--sklearn-color-unfitted-level-2);\n",
       "}\n",
       "\n",
       "#sk-container-id-2 div.sk-estimator.fitted:hover {\n",
       "  /* fitted */\n",
       "  background-color: var(--sklearn-color-fitted-level-2);\n",
       "}\n",
       "\n",
       "/* Specification for estimator info (e.g. \"i\" and \"?\") */\n",
       "\n",
       "/* Common style for \"i\" and \"?\" */\n",
       "\n",
       ".sk-estimator-doc-link,\n",
       "a:link.sk-estimator-doc-link,\n",
       "a:visited.sk-estimator-doc-link {\n",
       "  float: right;\n",
       "  font-size: smaller;\n",
       "  line-height: 1em;\n",
       "  font-family: monospace;\n",
       "  background-color: var(--sklearn-color-background);\n",
       "  border-radius: 1em;\n",
       "  height: 1em;\n",
       "  width: 1em;\n",
       "  text-decoration: none !important;\n",
       "  margin-left: 1ex;\n",
       "  /* unfitted */\n",
       "  border: var(--sklearn-color-unfitted-level-1) 1pt solid;\n",
       "  color: var(--sklearn-color-unfitted-level-1);\n",
       "}\n",
       "\n",
       ".sk-estimator-doc-link.fitted,\n",
       "a:link.sk-estimator-doc-link.fitted,\n",
       "a:visited.sk-estimator-doc-link.fitted {\n",
       "  /* fitted */\n",
       "  border: var(--sklearn-color-fitted-level-1) 1pt solid;\n",
       "  color: var(--sklearn-color-fitted-level-1);\n",
       "}\n",
       "\n",
       "/* On hover */\n",
       "div.sk-estimator:hover .sk-estimator-doc-link:hover,\n",
       ".sk-estimator-doc-link:hover,\n",
       "div.sk-label-container:hover .sk-estimator-doc-link:hover,\n",
       ".sk-estimator-doc-link:hover {\n",
       "  /* unfitted */\n",
       "  background-color: var(--sklearn-color-unfitted-level-3);\n",
       "  color: var(--sklearn-color-background);\n",
       "  text-decoration: none;\n",
       "}\n",
       "\n",
       "div.sk-estimator.fitted:hover .sk-estimator-doc-link.fitted:hover,\n",
       ".sk-estimator-doc-link.fitted:hover,\n",
       "div.sk-label-container:hover .sk-estimator-doc-link.fitted:hover,\n",
       ".sk-estimator-doc-link.fitted:hover {\n",
       "  /* fitted */\n",
       "  background-color: var(--sklearn-color-fitted-level-3);\n",
       "  color: var(--sklearn-color-background);\n",
       "  text-decoration: none;\n",
       "}\n",
       "\n",
       "/* Span, style for the box shown on hovering the info icon */\n",
       ".sk-estimator-doc-link span {\n",
       "  display: none;\n",
       "  z-index: 9999;\n",
       "  position: relative;\n",
       "  font-weight: normal;\n",
       "  right: .2ex;\n",
       "  padding: .5ex;\n",
       "  margin: .5ex;\n",
       "  width: min-content;\n",
       "  min-width: 20ex;\n",
       "  max-width: 50ex;\n",
       "  color: var(--sklearn-color-text);\n",
       "  box-shadow: 2pt 2pt 4pt #999;\n",
       "  /* unfitted */\n",
       "  background: var(--sklearn-color-unfitted-level-0);\n",
       "  border: .5pt solid var(--sklearn-color-unfitted-level-3);\n",
       "}\n",
       "\n",
       ".sk-estimator-doc-link.fitted span {\n",
       "  /* fitted */\n",
       "  background: var(--sklearn-color-fitted-level-0);\n",
       "  border: var(--sklearn-color-fitted-level-3);\n",
       "}\n",
       "\n",
       ".sk-estimator-doc-link:hover span {\n",
       "  display: block;\n",
       "}\n",
       "\n",
       "/* \"?\"-specific style due to the `<a>` HTML tag */\n",
       "\n",
       "#sk-container-id-2 a.estimator_doc_link {\n",
       "  float: right;\n",
       "  font-size: 1rem;\n",
       "  line-height: 1em;\n",
       "  font-family: monospace;\n",
       "  background-color: var(--sklearn-color-background);\n",
       "  border-radius: 1rem;\n",
       "  height: 1rem;\n",
       "  width: 1rem;\n",
       "  text-decoration: none;\n",
       "  /* unfitted */\n",
       "  color: var(--sklearn-color-unfitted-level-1);\n",
       "  border: var(--sklearn-color-unfitted-level-1) 1pt solid;\n",
       "}\n",
       "\n",
       "#sk-container-id-2 a.estimator_doc_link.fitted {\n",
       "  /* fitted */\n",
       "  border: var(--sklearn-color-fitted-level-1) 1pt solid;\n",
       "  color: var(--sklearn-color-fitted-level-1);\n",
       "}\n",
       "\n",
       "/* On hover */\n",
       "#sk-container-id-2 a.estimator_doc_link:hover {\n",
       "  /* unfitted */\n",
       "  background-color: var(--sklearn-color-unfitted-level-3);\n",
       "  color: var(--sklearn-color-background);\n",
       "  text-decoration: none;\n",
       "}\n",
       "\n",
       "#sk-container-id-2 a.estimator_doc_link.fitted:hover {\n",
       "  /* fitted */\n",
       "  background-color: var(--sklearn-color-fitted-level-3);\n",
       "}\n",
       "</style><div id=\"sk-container-id-2\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>GridSearchCV(cv=4,\n",
       "             estimator=XGBClassifier(base_score=None, booster=None,\n",
       "                                     callbacks=None, colsample_bylevel=None,\n",
       "                                     colsample_bynode=None,\n",
       "                                     colsample_bytree=None, device=None,\n",
       "                                     early_stopping_rounds=None,\n",
       "                                     enable_categorical=False, eval_metric=None,\n",
       "                                     feature_types=None, gamma=None,\n",
       "                                     grow_policy=None, importance_type=None,\n",
       "                                     interaction_constraints=None,\n",
       "                                     learning_rate=None,...\n",
       "                                     multi_strategy=None, n_estimators=None,\n",
       "                                     n_jobs=None, num_parallel_tree=None,\n",
       "                                     random_state=42, ...),\n",
       "             param_grid={&#x27;learning_rate&#x27;: [0.01, 0.1], &#x27;max_depth&#x27;: [2, 6, 12],\n",
       "                         &#x27;min_child_weight&#x27;: [0.5, 3, 5],\n",
       "                         &#x27;n_estimators&#x27;: [150, 200, 300]},\n",
       "             refit=&#x27;recall&#x27;,\n",
       "             scoring={&#x27;accuracy&#x27;: &#x27;accuracy&#x27;, &#x27;f1&#x27;: &#x27;f1&#x27;,\n",
       "                      &#x27;precision&#x27;: make_scorer(precision_score, response_method=&#x27;predict&#x27;, zero_division=0),\n",
       "                      &#x27;recall&#x27;: &#x27;recall&#x27;})</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item sk-dashed-wrapped\"><div class=\"sk-label-container\"><div class=\"sk-label fitted sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-4\" type=\"checkbox\" ><label for=\"sk-estimator-id-4\" class=\"sk-toggleable__label fitted sk-toggleable__label-arrow fitted\">&nbsp;&nbsp;GridSearchCV<a class=\"sk-estimator-doc-link fitted\" rel=\"noreferrer\" target=\"_blank\" href=\"https://scikit-learn.org/1.4/modules/generated/sklearn.model_selection.GridSearchCV.html\">?<span>Documentation for GridSearchCV</span></a><span class=\"sk-estimator-doc-link fitted\">i<span>Fitted</span></span></label><div class=\"sk-toggleable__content fitted\"><pre>GridSearchCV(cv=4,\n",
       "             estimator=XGBClassifier(base_score=None, booster=None,\n",
       "                                     callbacks=None, colsample_bylevel=None,\n",
       "                                     colsample_bynode=None,\n",
       "                                     colsample_bytree=None, device=None,\n",
       "                                     early_stopping_rounds=None,\n",
       "                                     enable_categorical=False, eval_metric=None,\n",
       "                                     feature_types=None, gamma=None,\n",
       "                                     grow_policy=None, importance_type=None,\n",
       "                                     interaction_constraints=None,\n",
       "                                     learning_rate=None,...\n",
       "                                     multi_strategy=None, n_estimators=None,\n",
       "                                     n_jobs=None, num_parallel_tree=None,\n",
       "                                     random_state=42, ...),\n",
       "             param_grid={&#x27;learning_rate&#x27;: [0.01, 0.1], &#x27;max_depth&#x27;: [2, 6, 12],\n",
       "                         &#x27;min_child_weight&#x27;: [0.5, 3, 5],\n",
       "                         &#x27;n_estimators&#x27;: [150, 200, 300]},\n",
       "             refit=&#x27;recall&#x27;,\n",
       "             scoring={&#x27;accuracy&#x27;: &#x27;accuracy&#x27;, &#x27;f1&#x27;: &#x27;f1&#x27;,\n",
       "                      &#x27;precision&#x27;: make_scorer(precision_score, response_method=&#x27;predict&#x27;, zero_division=0),\n",
       "                      &#x27;recall&#x27;: &#x27;recall&#x27;})</pre></div> </div></div><div class=\"sk-parallel\"><div class=\"sk-parallel-item\"><div class=\"sk-item\"><div class=\"sk-label-container\"><div class=\"sk-label fitted sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-5\" type=\"checkbox\" ><label for=\"sk-estimator-id-5\" class=\"sk-toggleable__label fitted sk-toggleable__label-arrow fitted\">estimator: XGBClassifier</label><div class=\"sk-toggleable__content fitted\"><pre>XGBClassifier(base_score=None, booster=None, callbacks=None,\n",
       "              colsample_bylevel=None, colsample_bynode=None,\n",
       "              colsample_bytree=None, device=None, early_stopping_rounds=None,\n",
       "              enable_categorical=False, eval_metric=None, feature_types=None,\n",
       "              gamma=None, grow_policy=None, importance_type=None,\n",
       "              interaction_constraints=None, learning_rate=None, max_bin=None,\n",
       "              max_cat_threshold=None, max_cat_to_onehot=None,\n",
       "              max_delta_step=None, max_depth=None, max_leaves=None,\n",
       "              min_child_weight=None, missing=nan, monotone_constraints=None,\n",
       "              multi_strategy=None, n_estimators=None, n_jobs=None,\n",
       "              num_parallel_tree=None, random_state=42, ...)</pre></div> </div></div><div class=\"sk-serial\"><div class=\"sk-item\"><div class=\"sk-estimator fitted sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-6\" type=\"checkbox\" ><label for=\"sk-estimator-id-6\" class=\"sk-toggleable__label fitted sk-toggleable__label-arrow fitted\">XGBClassifier</label><div class=\"sk-toggleable__content fitted\"><pre>XGBClassifier(base_score=None, booster=None, callbacks=None,\n",
       "              colsample_bylevel=None, colsample_bynode=None,\n",
       "              colsample_bytree=None, device=None, early_stopping_rounds=None,\n",
       "              enable_categorical=False, eval_metric=None, feature_types=None,\n",
       "              gamma=None, grow_policy=None, importance_type=None,\n",
       "              interaction_constraints=None, learning_rate=None, max_bin=None,\n",
       "              max_cat_threshold=None, max_cat_to_onehot=None,\n",
       "              max_delta_step=None, max_depth=None, max_leaves=None,\n",
       "              min_child_weight=None, missing=nan, monotone_constraints=None,\n",
       "              multi_strategy=None, n_estimators=None, n_jobs=None,\n",
       "              num_parallel_tree=None, random_state=42, ...)</pre></div> </div></div></div></div></div></div></div></div></div>"
      ],
      "text/plain": [
       "GridSearchCV(cv=4,\n",
       "             estimator=XGBClassifier(base_score=None, booster=None,\n",
       "                                     callbacks=None, colsample_bylevel=None,\n",
       "                                     colsample_bynode=None,\n",
       "                                     colsample_bytree=None, device=None,\n",
       "                                     early_stopping_rounds=None,\n",
       "                                     enable_categorical=False, eval_metric=None,\n",
       "                                     feature_types=None, gamma=None,\n",
       "                                     grow_policy=None, importance_type=None,\n",
       "                                     interaction_constraints=None,\n",
       "                                     learning_rate=None,...\n",
       "                                     multi_strategy=None, n_estimators=None,\n",
       "                                     n_jobs=None, num_parallel_tree=None,\n",
       "                                     random_state=42, ...),\n",
       "             param_grid={'learning_rate': [0.01, 0.1], 'max_depth': [2, 6, 12],\n",
       "                         'min_child_weight': [0.5, 3, 5],\n",
       "                         'n_estimators': [150, 200, 300]},\n",
       "             refit='recall',\n",
       "             scoring={'accuracy': 'accuracy', 'f1': 'f1',\n",
       "                      'precision': make_scorer(precision_score, response_method='predict', zero_division=0),\n",
       "                      'recall': 'recall'})"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Fit the model to the X_train and y_train data \n",
    "%time\n",
    "XGB_cv.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "id": "UFLTmIDm6805"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.1714843210388175"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Examine best score\n",
    "XGB_cv.best_score_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "id": "cdPUCuND6805"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'learning_rate': 0.1,\n",
       " 'max_depth': 12,\n",
       " 'min_child_weight': 5,\n",
       " 'n_estimators': 300}"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Examine best parameters\n",
    "XGB_cv.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "id": "QL19dH2h7KdD"
   },
   "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>Model</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>F1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>XGBoost cv</td>\n",
       "      <td>0.404893</td>\n",
       "      <td>0.171484</td>\n",
       "      <td>0.808135</td>\n",
       "      <td>0.24083</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        Model  Precision    Recall  Accuracy       F1\n",
       "0  XGBoost cv   0.404893  0.171484  0.808135  0.24083"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Pass the GridSearch object to the make_results() function.\n",
    "XGB_cv_results = make_results('XGBoost cv', XGB_cv, 'recall')\n",
    "XGB_cv_results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "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>Model</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>F1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Logistic Regression</td>\n",
       "      <td>0.535088</td>\n",
       "      <td>0.096215</td>\n",
       "      <td>0.824895</td>\n",
       "      <td>0.163102</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Random Forest cv</td>\n",
       "      <td>0.454665</td>\n",
       "      <td>0.128763</td>\n",
       "      <td>0.818160</td>\n",
       "      <td>0.200545</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>XGBoost cv</td>\n",
       "      <td>0.404893</td>\n",
       "      <td>0.171484</td>\n",
       "      <td>0.808135</td>\n",
       "      <td>0.240830</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 Model  Precision    Recall  Accuracy        F1\n",
       "0  Logistic Regression   0.535088  0.096215  0.824895  0.163102\n",
       "0     Random Forest cv   0.454665  0.128763  0.818160  0.200545\n",
       "0           XGBoost cv   0.404893  0.171484  0.808135  0.240830"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Concatentate decision tree model results table with master results table\n",
    "models_table = pd.concat([models_table, XGB_cv_results])\n",
    "models_table"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save the master results table\n",
    "models_table.to_csv('Retention_Results.csv', index=False);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "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>Model</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>F1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>XGBoost cv</td>\n",
       "      <td>0.404893</td>\n",
       "      <td>0.171484</td>\n",
       "      <td>0.808135</td>\n",
       "      <td>0.240830</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Random Forest cv</td>\n",
       "      <td>0.454665</td>\n",
       "      <td>0.128763</td>\n",
       "      <td>0.818160</td>\n",
       "      <td>0.200545</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Logistic Regression</td>\n",
       "      <td>0.535088</td>\n",
       "      <td>0.096215</td>\n",
       "      <td>0.824895</td>\n",
       "      <td>0.163102</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 Model  Precision    Recall  Accuracy        F1\n",
       "2           XGBoost cv   0.404893  0.171484  0.808135  0.240830\n",
       "1     Random Forest cv   0.454665  0.128763  0.818160  0.200545\n",
       "0  Logistic Regression   0.535088  0.096215  0.824895  0.163102"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# View the results\n",
    "models_table = pd.read_csv('Retention_Results.csv')\n",
    "models_table = models_table.sort_values(by=['Recall'], ascending=False)\n",
    "models_table.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Pickle the model (save)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Pickle the model\n",
    "with open('XGB_cv_WAZE.pickle', 'wb') as to_write: # wb = write binary\n",
    "    pickle.dump(XGB_cv, to_write)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Read in pickled model\n",
    "with open('XGB_cv_WAZE.pickle', 'rb') as to_read: # rb = read binary\n",
    "    XGB_cv = pickle.load(to_read)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Use models with validation set\n",
    "* Random forest model\n",
    "* XGBoost model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Random forest model on validation set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Use random forest model to predict on validation data with best tree\n",
    "RF_val_preds = RF_cv.best_estimator_.predict(X_val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Use get_test_scores funtion to create a table with test scores\n",
    "def get_test_scores(model_name:str, preds, y_test_data):   # model_name: name, preds: predictions, y_test_data: y_test data\n",
    "\n",
    "    accuracy = accuracy_score(y_test_data, preds)\n",
    "    precision = precision_score(y_test_data, preds)\n",
    "    recall = recall_score(y_test_data, preds)\n",
    "    f1 = f1_score(y_test_data, preds)\n",
    "    \n",
    "# Create table of best estimator row results\n",
    "    table = pd.DataFrame({'Model': [model_name],\n",
    "                          'Precision': [precision],\n",
    "                          'Recall': [recall],\n",
    "                          'Accuracy': [accuracy],\n",
    "                          'F1': [f1]\n",
    "                          },)\n",
    "\n",
    "    return table"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "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>Model</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>F1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Random Forest Val</td>\n",
       "      <td>0.423077</td>\n",
       "      <td>0.108481</td>\n",
       "      <td>0.815734</td>\n",
       "      <td>0.172684</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               Model  Precision    Recall  Accuracy        F1\n",
       "0  Random Forest Val   0.423077  0.108481  0.815734  0.172684"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Pass the scores get_test_scores function\n",
    "RF_cv_val = get_test_scores('Random Forest Val', RF_val_preds, y_val)\n",
    "RF_cv_val"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "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>Model</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>F1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>XGBoost cv</td>\n",
       "      <td>0.404893</td>\n",
       "      <td>0.171484</td>\n",
       "      <td>0.808135</td>\n",
       "      <td>0.240830</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Random Forest cv</td>\n",
       "      <td>0.454665</td>\n",
       "      <td>0.128763</td>\n",
       "      <td>0.818160</td>\n",
       "      <td>0.200545</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Logistic Regression</td>\n",
       "      <td>0.535088</td>\n",
       "      <td>0.096215</td>\n",
       "      <td>0.824895</td>\n",
       "      <td>0.163102</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Random Forest Val</td>\n",
       "      <td>0.423077</td>\n",
       "      <td>0.108481</td>\n",
       "      <td>0.815734</td>\n",
       "      <td>0.172684</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 Model  Precision    Recall  Accuracy        F1\n",
       "2           XGBoost cv   0.404893  0.171484  0.808135  0.240830\n",
       "1     Random Forest cv   0.454665  0.128763  0.818160  0.200545\n",
       "0  Logistic Regression   0.535088  0.096215  0.824895  0.163102\n",
       "0    Random Forest Val   0.423077  0.108481  0.815734  0.172684"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Concatentate decision tree model results table with master results table\n",
    "models_table = pd.concat([models_table, RF_cv_val])\n",
    "models_table"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save the master results table\n",
    "models_table.to_csv('Retention_Results.csv', index=False);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "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>Model</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>F1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>XGBoost cv</td>\n",
       "      <td>0.404893</td>\n",
       "      <td>0.171484</td>\n",
       "      <td>0.808135</td>\n",
       "      <td>0.240830</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Random Forest cv</td>\n",
       "      <td>0.454665</td>\n",
       "      <td>0.128763</td>\n",
       "      <td>0.818160</td>\n",
       "      <td>0.200545</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Random Forest Val</td>\n",
       "      <td>0.423077</td>\n",
       "      <td>0.108481</td>\n",
       "      <td>0.815734</td>\n",
       "      <td>0.172684</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Logistic Regression</td>\n",
       "      <td>0.535088</td>\n",
       "      <td>0.096215</td>\n",
       "      <td>0.824895</td>\n",
       "      <td>0.163102</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 Model  Precision    Recall  Accuracy        F1\n",
       "0           XGBoost cv   0.404893  0.171484  0.808135  0.240830\n",
       "1     Random Forest cv   0.454665  0.128763  0.818160  0.200545\n",
       "3    Random Forest Val   0.423077  0.108481  0.815734  0.172684\n",
       "2  Logistic Regression   0.535088  0.096215  0.824895  0.163102"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# View the results\n",
    "models_table = pd.read_csv('Retention_Results.csv')\n",
    "models_table = models_table.sort_values(by=['Recall'], ascending=False)\n",
    "models_table.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### XGBoost model on validation set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Use XGBoost model to predict on validation data with best tree\n",
    "XGB_val_preds = XGB_cv.best_estimator_.predict(X_val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "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>Model</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>F1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>XGBoost Val</td>\n",
       "      <td>0.412037</td>\n",
       "      <td>0.175542</td>\n",
       "      <td>0.809441</td>\n",
       "      <td>0.246196</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         Model  Precision    Recall  Accuracy        F1\n",
       "0  XGBoost Val   0.412037  0.175542  0.809441  0.246196"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Pass the scores get_test_scores function \n",
    "XGB_val = get_test_scores('XGBoost Val', XGB_val_preds, y_val)\n",
    "XGB_val"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Concatentate decision tree model results table with master results table\n",
    "models_table = pd.concat([models_table, XGB_val])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save the master results table\n",
    "models_table.to_csv('Retention_Results.csv', index=False);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "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>Model</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>F1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>XGBoost Val</td>\n",
       "      <td>0.412037</td>\n",
       "      <td>0.175542</td>\n",
       "      <td>0.809441</td>\n",
       "      <td>0.246196</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>XGBoost cv</td>\n",
       "      <td>0.404893</td>\n",
       "      <td>0.171484</td>\n",
       "      <td>0.808135</td>\n",
       "      <td>0.240830</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Random Forest cv</td>\n",
       "      <td>0.454665</td>\n",
       "      <td>0.128763</td>\n",
       "      <td>0.818160</td>\n",
       "      <td>0.200545</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Random Forest Val</td>\n",
       "      <td>0.423077</td>\n",
       "      <td>0.108481</td>\n",
       "      <td>0.815734</td>\n",
       "      <td>0.172684</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Logistic Regression</td>\n",
       "      <td>0.535088</td>\n",
       "      <td>0.096215</td>\n",
       "      <td>0.824895</td>\n",
       "      <td>0.163102</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 Model  Precision    Recall  Accuracy        F1\n",
       "4          XGBoost Val   0.412037  0.175542  0.809441  0.246196\n",
       "0           XGBoost cv   0.404893  0.171484  0.808135  0.240830\n",
       "1     Random Forest cv   0.454665  0.128763  0.818160  0.200545\n",
       "2    Random Forest Val   0.423077  0.108481  0.815734  0.172684\n",
       "3  Logistic Regression   0.535088  0.096215  0.824895  0.163102"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# View the results\n",
    "models_table = pd.read_csv('Retention_Results.csv')\n",
    "models_table = models_table.sort_values(by=['Recall'], ascending=False)\n",
    "models_table.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Use best model on test data\n",
    "The above table shows, when refitted to Recall, the XGB Boost model performed better than the random forest model on both the training and validation sets. It also performed better than the logistic regression model.  It will be used with the test data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "id": "7BkheTIsgU2b"
   },
   "outputs": [],
   "source": [
    "# Use XGBoost model to predict on test data with best tree\n",
    "XGB_test_preds = XGB_cv.best_estimator_.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "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>Model</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>F1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>XGBoost TEST</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.17357</td>\n",
       "      <td>0.807343</td>\n",
       "      <td>0.242091</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          Model  Precision   Recall  Accuracy        F1\n",
       "0  XGBoost TEST        0.4  0.17357  0.807343  0.242091"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Pass the scores get_test_scores function.\n",
    "XGB_test = get_test_scores('XGBoost TEST', XGB_test_preds, y_test)\n",
    "XGB_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Concatentate decision tree model results table with master results table\n",
    "models_table = pd.concat([models_table, XGB_test])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save the master results table\n",
    "models_table.to_csv('Retention_Results.csv', index=False);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "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>Model</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>F1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>XGBoost Val</td>\n",
       "      <td>0.412037</td>\n",
       "      <td>0.175542</td>\n",
       "      <td>0.809441</td>\n",
       "      <td>0.246196</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>XGBoost TEST</td>\n",
       "      <td>0.400000</td>\n",
       "      <td>0.173570</td>\n",
       "      <td>0.807343</td>\n",
       "      <td>0.242091</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>XGBoost cv</td>\n",
       "      <td>0.404893</td>\n",
       "      <td>0.171484</td>\n",
       "      <td>0.808135</td>\n",
       "      <td>0.240830</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Random Forest cv</td>\n",
       "      <td>0.454665</td>\n",
       "      <td>0.128763</td>\n",
       "      <td>0.818160</td>\n",
       "      <td>0.200545</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Random Forest Val</td>\n",
       "      <td>0.423077</td>\n",
       "      <td>0.108481</td>\n",
       "      <td>0.815734</td>\n",
       "      <td>0.172684</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               Model  Precision    Recall  Accuracy        F1\n",
       "0        XGBoost Val   0.412037  0.175542  0.809441  0.246196\n",
       "5       XGBoost TEST   0.400000  0.173570  0.807343  0.242091\n",
       "1         XGBoost cv   0.404893  0.171484  0.808135  0.240830\n",
       "2   Random Forest cv   0.454665  0.128763  0.818160  0.200545\n",
       "3  Random Forest Val   0.423077  0.108481  0.815734  0.172684"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# View the results\n",
    "models_table = pd.read_csv('Retention_Results.csv')\n",
    "models_table = models_table.sort_values(by=['Recall'], ascending=False)\n",
    "models_table.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "F8L_LyIbgV1I"
   },
   "source": [
    "#### Results\n",
    "All metrics for the XGBoost model fitted to the unseen test data decreased slightly, but this is to be expected. Recall score dropped ever so slightly to 0.1735 from 0.1755, Precision also decreased from 0.4120 to 0.4000 and Accuracy went from 0.8094 to 0.8073. Finally, F1 changed from 0.2461 to 0.2420.\n",
    "\n",
    "The XGBoost model also performed less well than the random forest in Precision and Accuracy; however, the goal was to boost the Recall score, where it performed several points higher. The combined F1 score demonstrates this."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "    retained       0.84      0.94      0.89      2353\n",
      "     churned       0.40      0.17      0.24       507\n",
      "\n",
      "    accuracy                           0.81      2860\n",
      "   macro avg       0.62      0.56      0.57      2860\n",
      "weighted avg       0.76      0.81      0.77      2860\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Create a classification report\n",
    "target_labels = ['retained', 'churned']\n",
    "print(classification_report(y_test, XGB_test_preds, target_names = target_labels))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "S5GNoz_QgWug"
   },
   "source": [
    "### Visualizations and evaluations on test model\n",
    "* plot a confusion matrix\n",
    "* get best features\n",
    "* plot ROC curve\n",
    "* commute AUC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "id": "WF3KErX8gXPc"
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAkgAAAHHCAYAAABEEKc/AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy81sbWrAAAACXBIWXMAAA9hAAAPYQGoP6dpAABjVUlEQVR4nO3deVgV1f8H8PdluewXRFkVATcEFUQsolxwA5evZWrmkkKilkGuqJkbaqlh5ZJb5d5PI9us3BL3DTUoXAkVQTRZVAQEZb3z+8OYHMArl3sR5b5fzzNPzJkzZ87crvLxc86ZkQmCIICIiIiIRHq13QEiIiKiZw0DJCIiIqJyGCARERERlcMAiYiIiKgcBkhERERE5TBAIiIiIiqHARIRERFROQyQiIiIiMphgERERERUDgMkIlJLREQEZDJZbXejVlR27y4uLggODtbaNYKDg+Hi4qK19oioehggEdWAjRs3QiaTiZuBgQEaNmyI4OBg/PPPP9Vq8+LFi4iIiEBKSop2O1uJ+/fvIyIiAocOHarxa6nj0c9UT08Pjo6OCAgIeOb6+SQ3b95EREQE4uPja7srRPQYBrXdAaK6bN68eXB1dUVBQQFOnjyJjRs34tixYzh//jyMjY3VauvixYuYO3cu/P39azzDcP/+fcydOxcA4O/vLzk2c+ZMfPDBBzV6fVV69OiBESNGQBAEJCcnY9WqVejatSt27tyJXr16PfX+JCYmQk9PvX9r3rx5E3PnzoWLiwvatm0rOfb1119DqVRqsYdEVB0MkIhqUK9evdC+fXsAwKhRo9CgQQN88skn+PXXXzFo0KBa7l31GBgYwMCg9v7qaNGiBd566y1x//XXX4enpyeWLl362ACpoKAAcrlc7UCmKoyMjLTanqGhoVbbI6Lq4RAb0VPUsWNHAEBSUpKk/O+//8bAgQNhbW0NY2NjtG/fHr/++qt4fOPGjXjjjTcAAF26dBGHmR4dWtq9ezc6duwIMzMzWFhYoE+fPrhw4YLkOsHBwTA3N8c///yDfv36wdzcHDY2NggPD0dpaSkAICUlBTY2NgCAuXPniteKiIgAUPk8nJKSEsyfPx9NmzaFkZERXFxc8OGHH6KwsFBSz8XFBf/73/9w7NgxvPjiizA2NkaTJk2wefPman6iQJs2bdCgQQMkJycDAA4dOgSZTIaoqCjMnDkTDRs2hKmpKXJzcwEAp06dQs+ePWFpaQlTU1N07twZx48fr9DusWPH8MILL8DY2BhNmzbFl19+Wen1K5uDlJ2djYkTJ8LFxQVGRkZo1KgRRowYgdu3b+PQoUN44YUXAABvv/22+Plu3LgRQOVzkPLz8zF58mQ4OTnByMgIbm5u+PTTTyEIgqSeTCZDWFgYtm/fjtatW8PIyAitWrXCnj17JPXu3buHCRMmiP2ztbVFjx498Oeff1bpMyfSBcwgET1FZfOH6tWrJ5ZduHABr7zyCho2bIgPPvgAZmZm2LZtG/r164cff/wRr7/+Ojp16oRx48Zh+fLl+PDDD+Hu7g4A4n+/+eYbBAUFITAwEJ988gnu37+P1atXo0OHDvjrr78kv3BLS0sRGBgIX19ffPrpp9i3bx8+++wzNG3aFGPHjoWNjQ1Wr16NsWPH4vXXX0f//v0BAJ6eno+9r1GjRmHTpk0YOHAgJk+ejFOnTmHhwoVISEjAzz//LKl75coVDBw4ECEhIQgKCsL69esRHBwMHx8ftGrVSu3P9O7du7h79y6aNWsmKZ8/fz7kcjnCw8NRWFgIuVyOAwcOoFevXvDx8cGcOXOgp6eHDRs2oGvXrjh69ChefPFFAMC5c+cQEBAAGxsbREREoKSkBHPmzIGdnd0T+5OXl4eOHTsiISEBI0eORLt27XD79m38+uuvuHHjBtzd3TFv3jzMnj0bY8aMEYPml19+udL2BEHAq6++ioMHDyIkJARt27bF77//jilTpuCff/7BkiVLJPWPHTuGn376Ce+99x4sLCywfPlyDBgwAKmpqahfvz4A4N1338UPP/yAsLAweHh44M6dOzh27BgSEhLQrl07tf8fENVJAhFp3YYNGwQAwr59+4Rbt24J169fF3744QfBxsZGMDIyEq5fvy7W7datm9CmTRuhoKBALFMqlcLLL78sNG/eXCz7/vvvBQDCwYMHJde6d++eYGVlJYwePVpSnp6eLlhaWkrKg4KCBADCvHnzJHW9vb0FHx8fcf/WrVsCAGHOnDkV7m3OnDnCo391xMfHCwCEUaNGSeqFh4cLAIQDBw6IZc7OzgIA4ciRI2JZZmamYGRkJEyePLnCtcoDIISEhAi3bt0SMjMzhVOnTgndunUTAAifffaZIAiCcPDgQQGA0KRJE+H+/fviuUqlUmjevLkQGBgoKJVKsfz+/fuCq6ur0KNHD7GsX79+grGxsXDt2jWx7OLFi4K+vr5Q/q9NZ2dnISgoSNyfPXu2AED46aefKvS/7Lp//PGHAEDYsGFDhTpBQUGCs7OzuL99+3YBgPDRRx9J6g0cOFCQyWTClStXJJ+PXC6XlJ05c0YAIHzxxRdimaWlpRAaGlrh2kT0Hw6xEdWg7t27w8bGBk5OThg4cCDMzMzw66+/olGjRgCArKwsHDhwAIMGDcK9e/dw+/Zt3L59G3fu3EFgYCAuX778xFVv0dHRyM7OxpAhQ8Tzb9++DX19ffj6+uLgwYMVznn33Xcl+x07dsTVq1erdY+7du0CAEyaNElSPnnyZADAzp07JeUeHh5i1gQAbGxs4ObmVuXrr1u3DjY2NrC1tYWvry+OHz+OSZMmYcKECZJ6QUFBMDExEffj4+Nx+fJlDB06FHfu3BE/p/z8fHTr1g1HjhyBUqlEaWkpfv/9d/Tr1w+NGzcWz3d3d0dgYOAT+/fjjz/Cy8sLr7/+eoVj1Xk8wq5du6Cvr49x48ZJyidPngxBELB7925Jeffu3dG0aVNx39PTEwqFQvL5WllZ4dSpU7h586ba/SHSFRxiI6pBK1euRIsWLZCTk4P169fjyJEjkkm9V65cgSAImDVrFmbNmlVpG5mZmWjYsOFjr3H58mUAQNeuXSs9rlAoJPvGxsbiHKMy9erVw927d6t0T+Vdu3YNenp6FYa47O3tYWVlhWvXrknKHw06qnP91157DWFhYZDJZLCwsECrVq1gZmZWoZ6rq6tkv+xzCgoKemzbOTk5KCwsxIMHD9C8efMKx93c3MSA8HGSkpIwYMCAqtxKlVy7dg2Ojo6wsLCQlJcNr1bn842MjERQUBCcnJzg4+OD3r17Y8SIEWjSpInW+k30vGOARFSDXnzxRXEVW79+/dChQwcMHToUiYmJMDc3F5dzh4eHPzY7UT7wKK+sjW+++Qb29vYVjpdfcaavr6/2fVRFVbMjj7u+UG7C8eM0atQI3bt3f2K9R7NHwH+f0+LFiyssrS9jbm5eYWL586Yqn++gQYPQsWNH/Pzzz9i7dy8WL16MTz75BD/99FOtPCqB6FnEAInoKdHX18fChQvRpUsXrFixAh988IH4L3ZDQ8Mn/tJ/XABSNpxia2tbpcChKtQZCnJ2doZSqcTly5fFrAYAZGRkIDs7G87Ozlrpk6bKPieFQqHyc7KxsYGJiYmYcXpUYmJila5z/vx5lXXU/Xz37duHe/fuSbJIf//9t3i8OhwcHPDee+/hvffeQ2ZmJtq1a4ePP/6YARLRvzgHiegp8vf3x4svvoilS5eioKAAtra28Pf3x5dffom0tLQK9W/duiX+XDaMlJ2dLakTGBgIhUKBBQsWoLi4WGUbVWVqalrptSrTu3dvAMDSpUsl5Z9//jkAoE+fPmpfvyb4+PigadOm+PTTT5GXl1fheNnnpK+vj8DAQGzfvh2pqani8YSEBPz+++9PvM6AAQNw5syZCqv3gP+yOI/7f1mZ3r17o7S0FCtWrJCUL1myBDKZTO2AprS0FDk5OZIyW1tbODo6PvfZMyJtYgaJ6CmbMmUK3njjDWzcuBHvvvsuVq5ciQ4dOqBNmzYYPXo0mjRpgoyMDMTExODGjRs4c+YMAKBt27bQ19fHJ598gpycHBgZGaFr166wtbXF6tWrMXz4cLRr1w6DBw+GjY0NUlNTsXPnTrzyyisVfrk+iYmJCTw8PPDdd9+hRYsWsLa2RuvWrdG6desKdb28vBAUFISvvvoK2dnZ6Ny5M06fPo1NmzahX79+6NKli1Y+N03p6elh7dq16NWrF1q1aoW3334bDRs2xD///IODBw9CoVDgt99+A/Dw+U979uxBx44d8d5776GkpARffPEFWrVqhbNnz6q8zpQpU/DDDz/gjTfewMiRI+Hj44OsrCz8+uuvWLNmDby8vNC0aVNYWVlhzZo1sLCwgJmZGXx9fSvMmwKAvn37okuXLpgxYwZSUlLg5eWFvXv34pdffsGECRMkE7Kr4t69e2jUqBEGDhwILy8vmJubY9++ffjjjz/w2WefqdUWUZ1Wq2voiOqosmX+f/zxR4VjpaWlQtOmTYWmTZsKJSUlgiAIQlJSkjBixAjB3t5eMDQ0FBo2bCj873//E3744QfJuV9//bXQpEkTcbn5o0v+Dx48KAQGBgqWlpaCsbGx0LRpUyE4OFiIjY0V6wQFBQlmZmYV+lR+6b4gCMKJEycEHx8fQS6XS5b8V1a3uLhYmDt3ruDq6ioYGhoKTk5OwvTp0yWPLhCEh0vi+/TpU+H6nTt3Fjp37lzxgywHwBOXp5ct8//+++8rPf7XX38J/fv3F+rXry8YGRkJzs7OwqBBg4T9+/dL6h0+fFi8/yZNmghr1qyp9N7LL/MXBEG4c+eOEBYWJjRs2FCQy+VCo0aNhKCgIOH27dtinV9++UXw8PAQDAwMJEv+yy/zF4SHj3KYOHGi4OjoKBgaGgrNmzcXFi9eLHlcgarP59E+FhYWClOmTBG8vLwECwsLwczMTPDy8hJWrVr1uI+USCfJBKGKMyOJiIiIdATnIBERERGVwwCJiIiIqBwGSERERETlMEAiIiIiKocBEhEREVE5DJCIiIiIyuGDIusYpVKJmzdvwsLColpvDiciotolCALu3bsHR0dH6OnVTB6joKAARUVFWmlLLpfD2NhYK209Sxgg1TE3b96Ek5NTbXeDiIg0dP36dTRq1Ejr7RYUFMDV2RzpmaVaac/e3h7Jycl1LkhigFTHlL3M8tqfLlCYcwSV6qb+bdrXdheIakyJUIyjRT9LXk6sTUVFRUjPLMW1OBcoLDT7PZF7TwlnnxQUFRUxQKJnW9mwmsJcT+MvPtGzykBmWNtdIKpxNT1NwtxCBnMLza6hRN2dysEAiYiISAeVCkqUaviysVJBqZ3OPIMYIBEREekgJQQooVmEpOn5zzKOwRARERGVwwwSERGRDlJCCU0HyDRv4dnFAImIiEgHlQoCSgXNhsg0Pf9ZxiE2IiIionKYQSIiItJBnKStGgMkIiIiHaSEgFIGSI/FITYiIiKicphBIiIi0kEcYlONARIREZEO4io21TjERkRERFQOM0hEREQ6SPnvpmkbdRUDJCIiIh1UqoVVbJqe/yxjgERERKSDSoWHm6Zt1FWcg0RERERUDjNIREREOohzkFRjgERERKSDlJChFDKN26irOMRGREREVA4zSERERDpIKTzcNG2jrmKAREREpINKtTDEpun5zzIOsRERERGVwwwSERGRDmIGSTUGSERERDpIKcigFDRcxabh+c8yDrERERERlcMMEhERkQ7iEJtqDJCIiIh0UCn0UKrhQFKplvryLGKAREREpIMELcxBEjgHiYiIiEh3MINERESkgzgHSTUGSERERDqoVNBDqaDhHKQ6/KoRDrERERHRU7Fw4UK88MILsLCwgK2tLfr164fExERJnYKCAoSGhqJ+/fowNzfHgAEDkJGRIamTmpqKPn36wNTUFLa2tpgyZQpKSkokdQ4dOoR27drByMgIzZo1w8aNG9XqKwMkIiIiHaSEDEroabipN8R2+PBhhIaG4uTJk4iOjkZxcTECAgKQn58v1pk4cSJ+++03fP/99zh8+DBu3ryJ/v37i8dLS0vRp08fFBUV4cSJE9i0aRM2btyI2bNni3WSk5PRp08fdOnSBfHx8ZgwYQJGjRqF33//vcp9lQmCUIcTZLonNzcXlpaWuHupCRQWjH+pburp6lvbXSCqMSVCMQ4WbkNOTg4UCoXW2y/7PfHr2aYws9DXqK38e6V41TOp2n29desWbG1tcfjwYXTq1Ak5OTmwsbHB1q1bMXDgQADA33//DXd3d8TExOCll17C7t278b///Q83b96EnZ0dAGDNmjWYNm0abt26BblcjmnTpmHnzp04f/68eK3BgwcjOzsbe/bsqVLf+BuUiIiINJKbmyvZCgsLq3ReTk4OAMDa2hoAEBcXh+LiYnTv3l2s07JlSzRu3BgxMTEAgJiYGLRp00YMjgAgMDAQubm5uHDhgljn0TbK6pS1URUMkIiIiHRQ2SRtTTcAcHJygqWlpbgtXLjwiddXKpWYMGECXnnlFbRu3RoAkJ6eDrlcDisrK0ldOzs7pKeni3UeDY7KjpcdU1UnNzcXDx48qNLnw1VsREREOujhHCQNX1b77/nXr1+XDLEZGRk98dzQ0FCcP38ex44d06gPNYUZJCIiItKIQqGQbE8KkMLCwrBjxw4cPHgQjRo1Esvt7e1RVFSE7OxsSf2MjAzY29uLdcqvaivbf1IdhUIBExOTKt0TAyQiIiIdpPz3XWyabEo1wwhBEBAWFoaff/4ZBw4cgKurq+S4j48PDA0NsX//frEsMTERqamp8PPzAwD4+fnh3LlzyMzMFOtER0dDoVDAw8NDrPNoG2V1ytqoCg6xERER6SDtPChSvYXwoaGh2Lp1K3755RdYWFiIc4YsLS1hYmICS0tLhISEYNKkSbC2toZCocD7778PPz8/vPTSSwCAgIAAeHh4YPjw4YiMjER6ejpmzpyJ0NBQMXP17rvvYsWKFZg6dSpGjhyJAwcOYNu2bdi5c2eV+8oAiYiISAcpq5EBqtiGegHS6tWrAQD+/v6S8g0bNiA4OBgAsGTJEujp6WHAgAEoLCxEYGAgVq1aJdbV19fHjh07MHbsWPj5+cHMzAxBQUGYN2+eWMfV1RU7d+7ExIkTsWzZMjRq1Ahr165FYGBglfvK5yDVMXwOEukCPgeJ6rKn9RykrfGtYarhc5Du3yvF0Lbna6yvtYkZJCIiIh1UKshQKmj4sloNz3+WMUAiIiLSQWUTrTVro+4OQnEMhoiIiKgcZpCIiIh0kFLQg1LDVWzKOjyNmQESERGRDuIQm2ocYiMiIiIqhxkkIiIiHaSE5qvQlNrpyjOJARIREZEO0s6DIuvuQFTdvTMiIiKiamIGiYiISAdp511sdTfPwgCJiIhIBykhgxKazkHik7SJiIioDmEGSbW6e2dERERE1cQMEhERkQ7SzoMi626ehQESERGRDlIKMig1fQ6Shuc/y+pu6EdERERUTcwgERER6SClFobY6vKDIhkgERER6SCloAelhqvQND3/WVZ374yIiIiomphBIiIi0kGlkKFUwwc9anr+s4wBEhERkQ7iEJtqdffOiIiIiKqJGSQiIiIdVArNh8hKtdOVZxIDJCIiIh3EITbVGCARERHpIL6sVrW6e2dERERE1cQMEhERkQ4SIINSwzlIApf5ExERUV3CITbV6u6dEREREVUTM0hEREQ6SCnIoBQ0GyLT9PxnGQMkIiIiHVQKPZRqOJCk6fnPsrp7Z0RERETVxACJiIhIB5UNsWm6qePIkSPo27cvHB0dIZPJsH37dslxmUxW6bZ48WKxjouLS4XjixYtkrRz9uxZdOzYEcbGxnByckJkZKTanw+H2IiIiHSQEnpQapgnUff8/Px8eHl5YeTIkejfv3+F42lpaZL93bt3IyQkBAMGDJCUz5s3D6NHjxb3LSwsxJ9zc3MREBCA7t27Y82aNTh37hxGjhwJKysrjBkzpsp9ZYBERERET0WvXr3Qq1evxx63t7eX7P/yyy/o0qULmjRpIim3sLCoULfMli1bUFRUhPXr10Mul6NVq1aIj4/H559/rlaAxCE2IiIiHVQqyLSy1ZSMjAzs3LkTISEhFY4tWrQI9evXh7e3NxYvXoySkhLxWExMDDp16gS5XC6WBQYGIjExEXfv3q3y9ZlBIiIi0kHaXOafm5srKTcyMoKRkZFGbW/atAkWFhYVhuLGjRuHdu3awdraGidOnMD06dORlpaGzz//HACQnp4OV1dXyTl2dnbisXr16lXp+gyQiIiIdJAg6EGp4ZOwhX/Pd3JykpTPmTMHERERGrW9fv16DBs2DMbGxpLySZMmiT97enpCLpfjnXfewcKFCzUOyh7FAImIiIg0cv36dSgUCnFf00Dl6NGjSExMxHfffffEur6+vigpKUFKSgrc3Nxgb2+PjIwMSZ2y/cfNW6oM5yARERHpoFLItLIBgEKhkGyaBkjr1q2Dj48PvLy8nlg3Pj4eenp6sLW1BQD4+fnhyJEjKC4uFutER0fDzc2tysNrAAMkIiIinaQUtPEsJPWumZeXh/j4eMTHxwMAkpOTER8fj9TUVLFObm4uvv/+e4waNarC+TExMVi6dCnOnDmDq1evYsuWLZg4cSLeeustMfgZOnQo5HI5QkJCcOHCBXz33XdYtmyZZGiuKjjERkRERE9FbGwsunTpIu6XBS1BQUHYuHEjACAqKgqCIGDIkCEVzjcyMkJUVBQiIiJQWFgIV1dXTJw4URL8WFpaYu/evQgNDYWPjw8aNGiA2bNnq7XEHwBkgiCoGf/Rsyw3NxeWlpa4e6kJFBZMEFZF1Be2OL7LCtevGEFurIRH+/sImXETTs0KAQC5d/Xxzaf2+POwBTJvymFpXYKXe+YgaGoazBRKAEDSBWNsW2GH86fNkHvXAHaNitBnxG28Puq2eJ07GQb4am5DXD5rgpvJRngt5DbGzvunVu75edfT1be2u/Dca/1iLgaOSUfz1vmob1eMuWOaIyb6v+GHt8bfQOe+WbBxKEJxsQxXzplh42eNkBhvDgCwa1iIoe//A6+Xc1HPphh3MuQ4sL0+olY6oqSYf/dookQoxsHCbcjJyZHM69GWst8TQQcHQ24uf/IJKhTlFWFTl6ga62tt0plvcXBwMPr161fj14mIiEDbtm0lZUVFRWjWrBlOnDhR5Xb27NmDtm3bQqlUarmHVN7ZGHP0Db6NpTsuY2FUEkpLgA+HNEXB/Yd/PLIyDHEnwxCjZ9/Elwf+RvjSVMQessDnkxuLbVw5awqrBiWYtuIavjr4N4aMz8CGBY74ZX0DsU5xkR6s6pdgyPgMNPF48NTvk+hRxiZKJCeYYuVs50qP30g2xqo5zni3Z2uEv+GOjH+MsGBTIiytH87raNT0AWR6wPIZrngnoA2++qgx+gzLRPCUG0/zNkgDSsi0stVVz90Qm7+/P9q2bYulS5eqdd6yZctQW8myNWvWwNXVFS+//LJYlpWVhffffx+//fYb9PT0MGDAACxbtgzm5g//ddazZ0/MmjULW7ZswfDhw2ul37piwdarkv3JS1PxZps2uHzWBG1eyodLywLMXpsiHnd0KULwtDREvu+M0hJA3wAIHJIlacPBuQgJsaY4vtsSr418mEWydyrC2PkPM0Z7o+rX7E0RPUHsYSvEHrZ67PFDvzaQ7H/1UWP0fPMWXFveR/wJS8QdsULckf/OT79ujEZNCtBnWCbWLmgMoufdM5VBKioqqrG2LS0tYWVlVWPtP44gCFixYkWFJ4EOGzYMFy5cQHR0NHbs2IEjR45UGB8NDg7G8uXLn2Z3CUB+rj4AwMKqVGUdU3Ml9FX8EyP/nr7KNoieFwaGSvQakom8XH1cTTB9bD0zi1Lcy9Z/ij0jTTzrT9KubbUaIPn7+yMsLAwTJkxAgwYNEBgYiPPnz6NXr14wNzeHnZ0dhg8fjtu3H/4LPDg4GIcPH8ayZcvEN/impKSgtLQUISEhcHV1hYmJCdzc3LBs2TLJtcoPsfn7+2PcuHGYOnUqrK2tYW9vX+GhVtnZ2Rg1ahRsbGygUCjQtWtXnDlzRlJn0aJFsLOzg4WFBUJCQlBQUCA5HhcXh6SkJPTp00csS0hIwJ49e7B27Vr4+vqiQ4cO+OKLLxAVFYWbN2+K9fr27YvY2FgkJSVp8jGTGpRKYM2chmj1Qh5cWhZUWifnjj62LrVHr7duV3ocAC78YYrDv9ZD72F3aqqrRDXuxa538fP5WPz6dyxeH5mOD4e7IfeuYaV1HZwL8OqIDOz61vYp95KqS/nvgyI13eqqWr+zTZs2QS6X4/jx41i0aBG6du0Kb29vxMbGYs+ePcjIyMCgQYMAPBwm8/Pzw+jRo5GWloa0tDQ4OTlBqVSiUaNG+P7773Hx4kXMnj0bH374IbZt2/bEa5uZmeHUqVOIjIzEvHnzEB0dLR5/4403kJmZid27dyMuLg7t2rVDt27dkJX1cDhl27ZtiIiIwIIFCxAbGwsHBwesWrVKco2jR4+iRYsWkjcNx8TEwMrKCu3btxfLunfvDj09PZw6dUosa9y4Mezs7HD06NHH3kNhYSFyc3MlG1Xfig8b4drfJpi++lqlx/Pv6WHWiCZo3KIAwyenV1on5W9jzH27Cd6alA4f/3s12V2iGnUmRoH3+rTGpAEeiDtsiQ9XXIFl/eIK9erbFeHjjYk4utsae6IYIFHdUOtzkJo3b47IyEgAwEcffQRvb28sWLBAPL5+/Xo4OTnh0qVLaNGiBeRyOUxNTSVPw9TX18fcuXPFfVdXV8TExGDbtm1icFUZT09PzJkzR+zHihUrsH//fvTo0QPHjh3D6dOnkZmZKT7w6tNPP8X27dvxww8/YMyYMVi6dClCQkLE4bOPPvoI+/btk2SRrl27BkdHR8l109PTxQdalTEwMIC1tTXS06W/dB0dHXHtWuW/rAFg4cKFknun6lvxYUOcilbgs5+vwMax4i+B+3l6mDG0KUzMlJizLhkGlfxD+tolI0wb1BS93rqNoRMyKlYgeo4UPtBH2jV9pF0D/o43x7oDZ9Bz0C18t/q/v9OsbYvwybcJuPinOZZNd6m9zpLalNDCu9jq8CTtWs8g+fj4iD+fOXMGBw8ehLm5ubi1bNkSAJ44zLRy5Ur4+PjAxsYG5ubm+OqrryQPnqqMp6enZN/BwQGZmZliX/Ly8lC/fn1Jf5KTk8W+JCQkwNdXutzYz89Psv/gwYMK75FRh4mJCe7fv//Y49OnT0dOTo64Xb9+vdrX0lWC8DA4OrHHEpHfX4F944pz4fLv6eHDIU1hKBcwd+NVyI0rTvhPSTTG1IHN0OONLLz9QeXZJaLnmUwPMJT/t7K2vl0RIqMScOWcGT6f0gRCHZ6PUhcJWljBJtThAKnWM0hmZmbiz3l5eejbty8++eSTCvUcHBwe20ZUVBTCw8Px2Wefwc/PDxYWFli8eLFkuKoyhobSFIBMJhOX1efl5cHBwQGHDh2qcJ46k70bNGiAc+fOScrs7e3FQKxMSUkJsrKyKrwnJisrCzY2No9tXxtvTNZ1Kz5shIM/10PEhqswMVciK/PhHwszi1IYmQhicFT4QA9Tv0jG/Tx93M97eK5l/RLo6z8cVpv6RlO097+H/u/cEtvQ0xdgVf+/idpJ500AAA/y9ZBzRx9J501gIFfCuUXh071p0nnGpqVwdP4v223vVIgm7vm4l2OA3LsGGBJ6Eyf31UPWLUMo6pWg7/AMNLAvwtFd1gD+DY6+TUDmP0b4ekFjcfk/ANy9rdmzdejpKHsatqZt1FW1HiA9ql27dvjxxx/h4uICA4PKuyaXy1FaKl0ZdPz4cbz88st47733xDJNJza3a9cO6enpMDAwgIuLS6V13N3dcerUKYwYMUIsO3nypKSOt7c3Vq9eDUEQIJM9/CL5+fkhOzsbcXFxYgbtwIEDUCqVkoxUQUEBkpKS4O3trdG9kGo7Nj1czjxlQHNJ+eQlqQh4MwtXzpni7z8fBvJvv+whqbPp1EXYOxXh6A4r5NwxxP4frbH/R2vxuF2jImw+fVHcfy/ATfz58llTHPzZukIdoqehRZt8REb9Le6/M+thxj36hwZYPsMFTk0L0H3AZSjqleBetgEunTVD+CB3XLv8cBVbuw45aOhaiIauhdhyMl7Sdk/XF5/afRDVlGcqQAoNDcXXX3+NIUOGiKvLrly5gqioKKxduxb6+vpwcXHBqVOnkJKSAnNzc1hbW6N58+bYvHkzfv/9d7i6uuKbb77BH3/8AVdX12r3pXv37vDz80O/fv0QGRmJFi1a4ObNm9i5cydef/11tG/fHuPHj0dwcDDat2+PV155BVu2bMGFCxfQpEkTsZ0uXbogLy8PFy5cQOvWrQE8DKx69uyJ0aNHY82aNSguLkZYWBgGDx4sma908uRJGBkZVRi2I+36/Wa8yuNeL+c9sc7w8HQMD3/ysNqT2iF6Ws6eUqgMZOaPbf7YYwAQ/aMNon98fHabnn3aWIXGVWxPiaOjI44fP47S0lIEBASgTZs2mDBhAqysrKCn97Cr4eHh0NfXh4eHB2xsbJCamop33nkH/fv3x5tvvglfX1/cuXNHkk2qDplMhl27dqFTp054++230aJFCwwePBjXrl2DnZ0dAODNN9/ErFmzMHXqVPj4+ODatWsYO3aspJ369evj9ddfx5YtWyTlW7ZsQcuWLdGtWzf07t0bHTp0wFdffSWp8+2332LYsGEwNX38c0eIiIiqQ/MX1Wo+RPcs47vYnoKzZ8+iR48eSEpKEp+U/SS3b9+Gm5sbYmNj1cqE8V1spAv4Ljaqy57Wu9he2zsShmaazRcrzi/CLwHr+S42qh5PT0988sknSE5OrvI5KSkpWLVqlUbDhERERI/Dd7Gp9kzNQarLgoOD1arfvn17yYMkiYiItImr2FRjBomIiIioHGaQiIiIdBAzSKoxQCIiItJBDJBU4xAbERERUTnMIBEREekgZpBUY4BERESkgwRA42X6dflBigyQiIiIdBAzSKpxDhIRERFROcwgERER6SBmkFRjgERERKSDGCCpxiE2IiIionKYQSIiItJBzCCpxgCJiIhIBwmCDIKGAY6m5z/LOMRGREREVA4zSERERDpICZnGD4rU9PxnGQMkIiIiHcQ5SKpxiI2IiIioHGaQiIiIdBAnaavGAImIiEgHcYhNNQ6xERER6aCyDJKmmzqOHDmCvn37wtHRETKZDNu3b5ccDw4Ohkwmk2w9e/aU1MnKysKwYcOgUChgZWWFkJAQ5OXlSeqcPXsWHTt2hLGxMZycnBAZGan258MAiYiIiJ6K/Px8eHl5YeXKlY+t07NnT6SlpYnbt99+Kzk+bNgwXLhwAdHR0dixYweOHDmCMWPGiMdzc3MREBAAZ2dnxMXFYfHixYiIiMBXX32lVl85xEZERKSDBC0MsambQerVqxd69eqlso6RkRHs7e0rPZaQkIA9e/bgjz/+QPv27QEAX3zxBXr37o1PP/0Ujo6O2LJlC4qKirB+/XrI5XK0atUK8fHx+PzzzyWB1JMwg0RERKSDBACCoOH2b1u5ubmSrbCwsNr9OnToEGxtbeHm5oaxY8fizp074rGYmBhYWVmJwREAdO/eHXp6ejh16pRYp1OnTpDL5WKdwMBAJCYm4u7du1XuBwMkIiIi0oiTkxMsLS3FbeHChdVqp2fPnti8eTP279+PTz75BIcPH0avXr1QWloKAEhPT4etra3kHAMDA1hbWyM9PV2sY2dnJ6lTtl9Wpyo4xEZERKSDlJBBpqUnaV+/fh0KhUIsNzIyqlZ7gwcPFn9u06YNPD090bRpUxw6dAjdunXTqK/qYgaJiIhIB2lzFZtCoZBs1Q2QymvSpAkaNGiAK1euAADs7e2RmZkpqVNSUoKsrCxx3pK9vT0yMjIkdcr2Hze3qTIMkIiIiOiZdOPGDdy5cwcODg4AAD8/P2RnZyMuLk6sc+DAASiVSvj6+op1jhw5guLiYrFOdHQ03NzcUK9evSpfmwESERGRDip7UKSmmzry8vIQHx+P+Ph4AEBycjLi4+ORmpqKvLw8TJkyBSdPnkRKSgr279+P1157Dc2aNUNgYCAAwN3dHT179sTo0aNx+vRpHD9+HGFhYRg8eDAcHR0BAEOHDoVcLkdISAguXLiA7777DsuWLcOkSZPU6ivnIBEREemgspVomrahjtjYWHTp0kXcLwtagoKCsHr1apw9exabNm1CdnY2HB0dERAQgPnz50uG7LZs2YKwsDB069YNenp6GDBgAJYvXy4et7S0xN69exEaGgofHx80aNAAs2fPVmuJP8AAiYiIiJ4Sf39/CCqiqt9///2JbVhbW2Pr1q0q63h6euLo0aNq9+9RDJCIiIh0EF9WqxoDJCIiIh3EAEk1BkhEREQ6SCnIINMwwNH0VSXPMq5iIyIiIiqHGSQiIiIdVBur2J4nDJCIiIh00MMASdM5SFrqzDOIQ2xERERE5TCDREREpIO4ik01BkhEREQ6SPh307SNuopDbERERETlMINERESkgzjEphoDJCIiIl3EMTaVGCARERHpIi1kkFCHM0icg0RERERUDjNIREREOohP0laNARIREZEO4iRt1TjERkRERFQOM0hERES6SJBpPsm6DmeQGCARERHpIM5BUo1DbERERETlMINERESki/igSJWqFCD9+uuvVW7w1VdfrXZniIiI6OngKjbVqhQg9evXr0qNyWQylJaWatIfIiIiolpXpQBJqVTWdD+IiIjoaavDQ2Sa0mgOUkFBAYyNjbXVFyIiInpKOMSmmtqr2EpLSzF//nw0bNgQ5ubmuHr1KgBg1qxZWLdundY7SERERDVA0NJWR6kdIH388cfYuHEjIiMjIZfLxfLWrVtj7dq1Wu0cERERUW1QO0DavHkzvvrqKwwbNgz6+vpiuZeXF/7++2+tdo6IiIhqikxLW92k9hykf/75B82aNatQrlQqUVxcrJVOERERUQ3jc5BUUjuD5OHhgaNHj1Yo/+GHH+Dt7a2VThERERHVJrUzSLNnz0ZQUBD++ecfKJVK/PTTT0hMTMTmzZuxY8eOmugjERERaRszSCqpnUF67bXX8Ntvv2Hfvn0wMzPD7NmzkZCQgN9++w09evSoiT4SERGRtgky7Wx1VLVeVtuxY0dER0cjMzMT9+/fx7FjxxAQEKDtvhEREVEdcuTIEfTt2xeOjo6QyWTYvn27eKy4uBjTpk1DmzZtYGZmBkdHR4wYMQI3b96UtOHi4gKZTCbZFi1aJKlz9uxZdOzYEcbGxnByckJkZKTafa32gyJjY2ORkJAA4OG8JB8fn+o2RURERE+ZIDzcNG1DHfn5+fDy8sLIkSPRv39/ybH79+/jzz//xKxZs+Dl5YW7d+9i/PjxePXVVxEbGyupO2/ePIwePVrct7CwEH/Ozc1FQEAAunfvjjVr1uDcuXMYOXIkrKysMGbMmCr3Ve0A6caNGxgyZAiOHz8OKysrAEB2djZefvllREVFoVGjRuo2SURERE9bLcxB6tWrF3r16lXpMUtLS0RHR0vKVqxYgRdffBGpqalo3LixWG5hYQF7e/tK29myZQuKioqwfv16yOVytGrVCvHx8fj888/VCpDUHmIbNWoUiouLkZCQgKysLGRlZSEhIQFKpRKjRo1StzkiIiJ6zuXm5kq2wsJCrbSbk5MDmUwmJmTKLFq0CPXr14e3tzcWL16MkpIS8VhMTAw6deokeZh1YGAgEhMTcffu3SpfW+0M0uHDh3HixAm4ubmJZW5ubvjiiy/QsWNHdZsjIiKi2qCNSdb/nu/k5CQpnjNnDiIiIjRquqCgANOmTcOQIUOgUCjE8nHjxqFdu3awtrbGiRMnMH36dKSlpeHzzz8HAKSnp8PV1VXSlp2dnXisXr16Vbq+2gGSk5NTpQ+ELC0thaOjo7rNERERUS2QCQ83TdsAgOvXr0uCGCMjI43aLS4uxqBBgyAIAlavXi05NmnSJPFnT09PyOVyvPPOO1i4cKHG132U2kNsixcvxvvvvy+ZMBUbG4vx48fj008/1VrHiIiIqAZp8WW1CoVCsmkSqJQFR9euXUN0dLQk8KqMr68vSkpKkJKSAgCwt7dHRkaGpE7Z/uPmLVWmShmkevXqQSb7Lw2Xn58PX19fGBg8PL2kpAQGBgYYOXIk+vXrV+WLExEREZUpC44uX76MgwcPon79+k88Jz4+Hnp6erC1tQUA+Pn5YcaMGSguLoahoSEAIDo6Gm5ublUeXgOqGCAtXbq0yg0SERHRc0CLc5CqKi8vD1euXBH3k5OTER8fD2trazg4OGDgwIH4888/sWPHDpSWliI9PR0AYG1tDblcjpiYGJw6dQpdunSBhYUFYmJiMHHiRLz11lti8DN06FDMnTsXISEhmDZtGs6fP49ly5ZhyZIlavW1SgFSUFCQWo0SERHRM64WlvnHxsaiS5cu4n7ZfKKgoCBERETg119/BQC0bdtWct7Bgwfh7+8PIyMjREVFISIiAoWFhXB1dcXEiRMl85IsLS2xd+9ehIaGwsfHBw0aNMDs2bPVWuIPaPCgSODhDPOioiJJ2ZPGComIiEg3+fv7Q1DxdElVxwCgXbt2OHny5BOv4+npiaNHj6rdv0epPUk7Pz8fYWFhsLW1hZmZGerVqyfZiIiI6DmgxUnadZHaAdLUqVNx4MABrF69GkZGRli7di3mzp0LR0dHbN68uSb6SERERNrGAEkltYfYfvvtN2zevBn+/v54++230bFjRzRr1gzOzs7YsmULhg0bVhP9JCIiInpq1M4gZWVloUmTJgAezjfKysoCAHTo0AFHjhzRbu+IiIioZpStYtN0q6PUDpCaNGmC5ORkAEDLli2xbds2AA8zS+XflUJERETPprInaWu61VVqB0hvv/02zpw5AwD44IMPsHLlShgbG2PixImYMmWK1jtIRERE9LSpPQdp4sSJ4s/du3fH33//jbi4ODRr1gyenp5a7RwRERHVkFp4DtLzRKPnIAGAs7MznJ2dtdEXIiIiomdClQKk5cuXV7nBcePGVbszRERE9HTIoPkcoro7RbuKAVJV318ik8kYIBEREdFzr0oBUtmqNXp+9B88CAb6xrXdDaIaIRSer+0uENUYQSh+Shd6+i+rfZ5oPAeJiIiInkOcpK2S2sv8iYiIiOo6ZpCIiIh0ETNIKjFAIiIi0kHaeBI2n6RNREREpEOqFSAdPXoUb731Fvz8/PDPP/8AAL755hscO3ZMq50jIiKiGiJoaauj1A6QfvzxRwQGBsLExAR//fUXCgsLAQA5OTlYsGCB1jtIRERENYABkkpqB0gfffQR1qxZg6+//hqGhoZi+SuvvII///xTq50jIiIiqg1qT9JOTExEp06dKpRbWloiOztbG30iIiKiGsZJ2qqpnUGyt7fHlStXKpQfO3YMTZo00UqniIiIqIaVPUlb062OUjtAGj16NMaPH49Tp05BJpPh5s2b2LJlC8LDwzF27Nia6CMRERFpG+cgqaT2ENsHH3wApVKJbt264f79++jUqROMjIwQHh6O999/vyb6SERERPRUqR0gyWQyzJgxA1OmTMGVK1eQl5cHDw8PmJub10T/iIiIqAZwDpJq1X6Stlwuh4eHhzb7QkRERE8LXzWiktoBUpcuXSCTPX5S1oEDBzTqEBEREVFtUztAatu2rWS/uLgY8fHxOH/+PIKCgrTVLyIiIqpJWhhiYwbpEUuWLKm0PCIiAnl5eRp3iIiIiJ4CDrGppLWX1b711ltYv369tpojIiIiqjXVnqRdXkxMDIyNjbXVHBEREdUkZpBUUjtA6t+/v2RfEASkpaUhNjYWs2bN0lrHiIiIqOZwmb9qagdIlpaWkn09PT24ublh3rx5CAgI0FrHiIiIiGqLWnOQSktL8fbbb+Pzzz/Hhg0bsGHDBqxbtw6LFi1icEREREQqHTlyBH379oWjoyNkMhm2b98uOS4IAmbPng0HBweYmJige/fuuHz5sqROVlYWhg0bBoVCASsrK4SEhFRYJHb27Fl07NgRxsbGcHJyQmRkpNp9VStA0tfXR0BAALKzs9W+EBERET1DauFdbPn5+fDy8sLKlSsrPR4ZGYnly5djzZo1OHXqFMzMzBAYGIiCggKxzrBhw3DhwgVER0djx44dOHLkCMaMGSMez83NRUBAAJydnREXF4fFixcjIiICX331lVp9VXuIrXXr1rh69SpcXV3VPZWIiIieEbUxB6lXr17o1atXpccEQcDSpUsxc+ZMvPbaawCAzZs3w87ODtu3b8fgwYORkJCAPXv24I8//kD79u0BAF988QV69+6NTz/9FI6OjtiyZQuKioqwfv16yOVytGrVCvHx8fj8888lgdSTqL3M/6OPPkJ4eDh27NiBtLQ05ObmSjYiIiIidSUnJyM9PR3du3cXyywtLeHr64uYmBgAD1fMW1lZicERAHTv3h16eno4deqUWKdTp06Qy+VincDAQCQmJuLu3btV7k+VM0jz5s3D5MmT0bt3bwDAq6++KnnliCAIkMlkKC0trfLFiYiIqBZpaRVa+QSJkZERjIyM1GojPT0dAGBnZycpt7OzE4+lp6fD1tZWctzAwADW1taSOuVHucraTE9PR7169arUnyoHSHPnzsW7776LgwcPVvUUIiIielZp8TlITk5OkuI5c+YgIiJCw8ZrV5UDJEF4+Cl07ty5xjpDREREz5/r169DoVCI++pmjwDA3t4eAJCRkQEHBwexPCMjQ3wPrL29PTIzMyXnlZSUICsrSzzf3t4eGRkZkjpl+2V1qkKtOUiPDqkRERHR86tskramGwAoFArJVp0AydXVFfb29ti/f79Ylpubi1OnTsHPzw8A4Ofnh+zsbMTFxYl1Dhw4AKVSCV9fX7HOkSNHUFxcLNaJjo6Gm5tblYfXADVXsbVo0eKJQVJWVpY6TRIREVFtqIVXjeTl5eHKlSvifnJyMuLj42FtbY3GjRtjwoQJ+Oijj9C8eXO4urpi1qxZcHR0RL9+/QAA7u7u6NmzJ0aPHo01a9aguLgYYWFhGDx4MBwdHQEAQ4cOxdy5cxESEoJp06bh/PnzWLZsGZYsWaJWX9UKkObOnVvhSdpEREREVREbG4suXbqI+5MmTQIABAUFYePGjZg6dSry8/MxZswYZGdno0OHDtizZ4/kXa9btmxBWFgYunXrBj09PQwYMADLly8Xj1taWmLv3r0IDQ2Fj48PGjRogNmzZ6u1xB8AZELZ5KIn0NPTq3T2OD1bcnNzYWlpiS7tPoCBPl8eTHWTEHu+trtAVGNKhGIcwi/IycmRzOvRlrLfEy3CF0DfSLPfE6WFBbj06Yc11tfaVOUMEucfERER1SG1MMT2PKnyJO0qJpqIiIiInntVziAplcqa7AcRERE9TcwgqaT2u9iIiIjo+Vcb72J7njBAIiIi0kXMIKmk9stqiYiIiOo6ZpCIiIh0ETNIKjFAIiIi0kGcg6Qah9iIiIiIymEGiYiISBdxiE0lBkhEREQ6iENsqnGIjYiIiKgcZpCIiIh0EYfYVGKAREREpIsYIKnEITYiIiKicphBIiIi0kGyfzdN26irGCARERHpIg6xqcQAiYiISAdxmb9qnINEREREVA4zSERERLqIQ2wqMUAiIiLSVXU4wNEUh9iIiIiIymEGiYiISAdxkrZqDJCIiIh0EecgqcQhNiIiIqJymEEiIiLSQRxiU40BEhERkS7iEJtKHGIjIiIiKocZJCIiIh3EITbVGCARERHpIg6xqcQAiYiISBcxQFKJc5CIiIiIymEGiYiISAdxDpJqzCARERHpIkFLmxpcXFwgk8kqbKGhoQAAf3//CsfeffddSRupqano06cPTE1NYWtriylTpqCkpKSaH8LjMYNERERET8Uff/yB0tJScf/8+fPo0aMH3njjDbFs9OjRmDdvnrhvamoq/lxaWoo+ffrA3t4eJ06cQFpaGkaMGAFDQ0MsWLBAq31lgERERKSDZIIAmaDZGJm659vY2Ej2Fy1ahKZNm6Jz585imampKezt7Ss9f+/evbh48SL27dsHOzs7tG3bFvPnz8e0adMQEREBuVyu/k08BofYiIiIdJEWh9hyc3MlW2Fh4RMvX1RUhP/7v//DyJEjIZPJxPItW7agQYMGaN26NaZPn4779++Lx2JiYtCmTRvY2dmJZYGBgcjNzcWFCxeq/VFUhhkkIiIi0oiTk5Nkf86cOYiIiFB5zvbt25GdnY3g4GCxbOjQoXB2doajoyPOnj2LadOmITExET/99BMAID09XRIcARD309PTNb+RRzBAIiIi0kHaXMV2/fp1KBQKsdzIyOiJ565btw69evWCo6OjWDZmzBjx5zZt2sDBwQHdunVDUlISmjZtqlln1cQhNiIiIl2kxSE2hUIh2Z4UIF27dg379u3DqFGjVNbz9fUFAFy5cgUAYG9vj4yMDEmdsv3HzVuqLgZIRERE9FRt2LABtra26NOnj8p68fHxAAAHBwcAgJ+fH86dO4fMzEyxTnR0NBQKBTw8PLTaRw6xERER6aDaelCkUqnEhg0bEBQUBAOD/8KQpKQkbN26Fb1790b9+vVx9uxZTJw4EZ06dYKnpycAICAgAB4eHhg+fDgiIyORnp6OmTNnIjQ0tErDeupggERERKSLauldbPv27UNqaipGjhwpKZfL5di3bx+WLl2K/Px8ODk5YcCAAZg5c6ZYR19fHzt27MDYsWPh5+cHMzMzBAUFSZ6bpC0MkIiIiHRQbWWQAgICIFTy/CQnJyccPnz4iec7Oztj165d6l9YTZyDRERERFQOM0hERES6qJaG2J4XDJCIiIh0lKZDbHUZh9iIiIiIymEGiYiISBcJwsNN0zbqKAZIREREOqi2VrE9LzjERkRERFQOM0hERES6iKvYVGKAREREpINkyoebpm3UVRxiIyIiIiqHGSSiJxg04AJGBsXj51/d8OXa9gCAXoGX0aVTCpo2zYKZaQkGDHkD+flyyXnNmmRhZPBfaNHsDpRKGY7FNMZX69qhoMCwNm6D6LH09AS8NTkd3QZko55NMe5kGCJ6mzW2LrUFIAMAGJuWImRGGvwCc6GoV4L063L8sq4Bdn7ToHY7T9XHITaVnukMUkpKCmQyGeLj42u7K2pxcXHB0qVLJWX79++Hu7s7SktLq9zOBx98gPfff1/LvSN1tGh2B717XsbVZCtJuZFRKWL/dMR337eu9Dxr6/tYOH8/bqZZYMKUnpgZ0RXOjbMxeXzMU+g1kXoGhWbif0F3sHJGQ4zu3BLrPnbAG+9l4rWQ22KddyJuor3/PUS+3xijO7fEz1/bIPTjf/BSQE4t9pw0UbaKTdOtrnqmA6S6ZOrUqZg5cyb09fUBAGlpaRg6dChatGgBPT09TJgwocI54eHh2LRpE65evfqUe0sAYGxcjKmTj2PZCl/k5UmzQ9t/bYltP7bC34mV/+vZ94V/UFKqh5VrXsCNfxS4dKU+vlj1Ijq+ch0ODveeRveJqsyjfT5ifrfE6f0KZNyQ49hOK/x52AJube8/Uuc+or+3xtkYc2TckGP3lvq4etFEUoeeM2XPQdJ0q6N0MkAqKip6qtc7duwYkpKSMGDAALGssLAQNjY2mDlzJry8vCo9r0GDBggMDMTq1aufVlfpEaHv/oHTsQ3x1xkHtc81NFCipFgPgiATywqLHo5ot3bP1FofibThYqwZ2na4h4ZNCgEATTweoNWL+fjjgOKROqZ4KSAH9e2LAQjwejkPDZsUIu6wRS31mqhmPRMBklKpRGRkJJo1awYjIyM0btwYH3/8sXj86tWr6NKlC0xNTeHl5YWYmP+GKSIiItC2bVtJe0uXLoWLi4u4HxwcjH79+uHjjz+Go6Mj3NzcxOG7n3766bFtAw+Dm44dO8LExAROTk4YN24c8vPzxeOZmZno27cvTExM4Orqii1btlS4v6ioKPTo0QPGxsZimYuLC5YtW4YRI0bA0tLysZ9N3759ERUV9djjhYWFyM3NlWykuc4dU9CsSRY2bG5brfPPnLVDvXoPMPD1izAwKIW5WSFGjvgLAGBt/UCLPSXS3HcrbHH4FyusPfI3dl47g5V7L+Hnrxvg4M/1xDqrZjZE6iVjbP3zInZeO4uPtlzFyg8b4vwp81rsOWmCQ2yqPRMB0vTp07Fo0SLMmjULFy9exNatW2FnZycenzFjBsLDwxEfH48WLVpgyJAhKCkpUesa+/fvR2JiIqKjo7Fjx44qtZ2UlISePXtiwIABOHv2LL777jscO3YMYWFh4vnBwcG4fv06Dh48iB9++AGrVq1CZqY0Q3D06FG0b9++Oh8NXnzxRdy4cQMpKSmVHl+4cCEsLS3FzcnJqVrXof80aJCPd0fHIfLzV1BcrF+tNq5dt8KnS/3Qv18Cfvn+O2zd/BMyMsyRddcYSqXsyQ0QPUWdXs1G1/7ZWBTaGKGBLfDpeCcMfPcWur+RJdZ5beRttPS5j9lBLgjr2QJfz3NE6IJ/4N2RQ8bPLUFLWx1V66vY7t27h2XLlmHFihUICgoCADRt2hQdOnQQg4Lw8HD06dMHADB37ly0atUKV65cQcuWLat8HTMzM6xduxZy+cO5JFVpe+HChRg2bJg4P6h58+ZYvnw5OnfujNWrVyM1NRW7d+/G6dOn8cILLwAA1q1bB3d3d8m1r127BkdHx2p9PmXnXbt2TZIVKzN9+nRMmjRJ3M/NzWWQpKHmTbNQz6oAK5bsFsv09QW0bpWJV/tcQt8Bg6FUPvnfFoeOuOLQEVdYWT1AQYEBBEGG11/7G+kZ/Bc3PVtGz0r7N4v0MGOU8rcJbBsVY/D7mdj3vTXkxkoEf5COeSEuOL3/4bBbcoIJmrR6gIHv3sJfRznMRnVPrQdICQkJKCwsRLdu3R5bx9PTU/zZweHhfJDMzEy1AqQ2bdqIwVFV2z5z5gzOnj0rGTYTBAFKpRLJycm4dOkSDAwM4OPjIx5v2bIlrKysJNd48OCBZHhNHSYmJgCA+/crnwhpZGQEIyOjarVNlYs/a493wvpIyiaPj8H1Gwps+7FVlYKjR2VnP/x/GNA9CcXFevgzXv05TUQ1ychYCaHcA/+UpYDs3/ETAwMBhnIBysrq6NXhFEIdx3exqVbrAVJZAKCKoeF/z42RyR4OTyj//ZOqp6cHodws+uLi4gptmJmZqd12Xl4e3nnnHYwbN67CeY0bN8alS5ee2Hfg4WTru3fvVqlueVlZD1PcNjY21Tqf1PfggSGupVpJygoKDJB7z0gsr2f1APXqPYDjvyvSXJyz8eCBATJvmSEv72HA2rdPIhISbPCgwADt2qYh5O2/sGFT2wrPSyKqbSejFRg8LhOZ/8hxLdEYTVs/QP93bmFvlDUA4H6ePs6cMMPoWWkoKtBDxg1DePrlo/vAu/hqbvWy4/QM0MYqtDq8iq3WA6TmzZvDxMQE+/fvx6hRo9Q+38bGBunp6RAEQQxwtPXcpHbt2uHixYto1qxZpcdbtmyJkpISxMXFiUNsiYmJyM7OltTz9vbGxYsXq9WH8+fPw9DQEK1atarW+VQz+vS6jLeGnBP3P1sU/fC/S19C9IGmAAC35ncwfMhZGJuU4MYNBb5Y+SL2H2pSK/0lUmXVzIYImpqOsIU3YFW/BHcyDLHrm/rYsuS/uaALxzpj5IdpmLbiGiysSpH5jxwbP3HAjs31a7HnRDWn1gMkY2NjTJs2DVOnToVcLscrr7yCW7du4cKFCyqH3cr4+/vj1q1biIyMxMCBA7Fnzx7s3r0bCoXiiec+ybRp0/DSSy8hLCwMo0aNgpmZGS5evIjo6GisWLECbm5u6NmzJ9555x2sXr0aBgYGmDBhQoWsWGBgIDZt2lSh/bJALi8vD7du3UJ8fDzkcjk8PDzEOkePHhVX0VHtmTqjh2T//771xP996/mY2g99uvTlmuwSkdY8yNfHmjkNsWZOw8fWuXvLEJ9NbPwUe0U1jUNsqj0Tq9hmzZqFyZMnY/bs2XB3d8ebb75ZYSXY47i7u2PVqlVYuXIlvLy8cPr0aYSHh2ulX56enjh8+DAuXbqEjh07wtvbG7Nnz5ZMuN6wYQMcHR3RuXNn9O/fH2PGjIGtra2knWHDhuHChQtITEyUlHt7e8Pb2xtxcXHYunUrvL290bt3b0mdqKgojB49Wiv3Q0REJOIqNpVkQvkJPFQjpkyZgtzcXHz55ZdVPmf37t2YPHkyzp49CwODqiX7cnNzYWlpiS7tPoCBfvUmhhM964TY87XdBaIaUyIU4xB+QU5OjlZGQ8or+z3h13MeDAw1+z1RUlyAmD2za6yvtemZyCDpghkzZsDZ2VmcAF4V+fn52LBhQ5WDIyIioqrigyJV42/ep8TKygoffvihWucMHDiwhnpDREQ6Tyk83DRto45igERERKSLtDGHqO7GRxxiIyIiIiqPGSQiIiIdJIMWlvlrpSfPJgZIREREuohP0laJQ2xERERE5TCDREREpIP4JG3VGCARERHpIq5iU4lDbERERPRUREREQCaTSbaWLVuKxwsKChAaGor69evD3NwcAwYMQEZGhqSN1NRU9OnTB6amprC1tcWUKVNQUlKi9b4yg0RERKSDZIIAmYaTrKtzfqtWrbBv3z5x/9G3RUycOBE7d+7E999/D0tLS4SFhaF///44fvw4AKC0tBR9+vSBvb09Tpw4gbS0NIwYMQKGhoZYsGCBRvdSHgMkIiIiXaT8d9O0DTUZGBjA3t6+QnlOTg7WrVuHrVu3omvXrgAevhDe3d0dJ0+exEsvvYS9e/fi4sWL2LdvH+zs7NC2bVvMnz8f06ZNQ0REBORyuYY39B8OsREREZFGcnNzJVthYeFj616+fBmOjo5o0qQJhg0bhtTUVABAXFwciouL0b17d7Fuy5Yt0bhxY8TExAAAYmJi0KZNG9jZ2Yl1AgMDkZubiwsXLmj1nhggERER6aCyITZNNwBwcnKCpaWluC1cuLDSa/r6+mLjxo3Ys2cPVq9ejeTkZHTs2BH37t1Deno65HI5rKysJOfY2dkhPT0dAJCeni4JjsqOlx3TJg6xERER6SItrmK7fv06FAqFWGxkZFRp9V69eok/e3p6wtfXF87Ozti2bRtMTEw07Ix2MYNERESki8qepK3pBkChUEi2xwVI5VlZWaFFixa4cuUK7O3tUVRUhOzsbEmdjIwMcc6Svb19hVVtZfuVzWvSBAMkIiIiqhV5eXlISkqCg4MDfHx8YGhoiP3794vHExMTkZqaCj8/PwCAn58fzp07h8zMTLFOdHQ0FAoFPDw8tNo3DrERERHpoNp4knZ4eDj69u0LZ2dn3Lx5E3PmzIG+vj6GDBkCS0tLhISEYNKkSbC2toZCocD7778PPz8/vPTSSwCAgIAAeHh4YPjw4YiMjER6ejpmzpyJ0NDQKmetqooBEhERkS6qhZfV3rhxA0OGDMGdO3dgY2ODDh064OTJk7CxsQEALFmyBHp6ehgwYAAKCwsRGBiIVatWiefr6+tjx44dGDt2LPz8/GBmZoagoCDMmzdPs/uoBAMkIiIieiqioqJUHjc2NsbKlSuxcuXKx9ZxdnbGrl27tN21ChggERER6SCZ8uGmaRt1FQMkIiIiXVQLQ2zPE65iIyIiIiqHGSQiIiJdpMUHRdZFDJCIiIh00KOvCtGkjbqKQ2xERERE5TCDREREpIs4SVslBkhERES6SACg6TL9uhsfMUAiIiLSRZyDpBrnIBERERGVwwwSERGRLhKghTlIWunJM4kBEhERkS7iJG2VOMRGREREVA4zSERERLpICUCmhTbqKAZIREREOoir2FTjEBsRERFROcwgERER6SJO0laJARIREZEuYoCkEofYiIiIiMphBomIiEgXMYOkEgMkIiIiXcRl/ioxQCIiItJBXOavGucgEREREZXDDBIREZEu4hwklRggERER6SKlAMg0DHCUdTdA4hAbERERUTnMIBEREekiDrGpxACJiIhIJ2khQELdDZA4xEZERERUDjNIREREuohDbCoxQCIiItJFSgEaD5FxFRsRERGR7mCAREREpIsEpXY2NSxcuBAvvPACLCwsYGtri379+iExMVFSx9/fHzKZTLK9++67kjqpqano06cPTE1NYWtriylTpqCkpETjj+RRHGIjIiLSRbUwB+nw4cMIDQ3FCy+8gJKSEnz44YcICAjAxYsXYWZmJtYbPXo05s2bJ+6bmpqKP5eWlqJPnz6wt7fHiRMnkJaWhhEjRsDQ0BALFizQ7H4ewQCJiIhIF9XCHKQ9e/ZI9jdu3AhbW1vExcWhU6dOYrmpqSns7e0rbWPv3r24ePEi9u3bBzs7O7Rt2xbz58/HtGnTEBERAblcrv59VIJDbERERFQrcnJyAADW1taS8i1btqBBgwZo3bo1pk+fjvv374vHYmJi0KZNG9jZ2YllgYGByM3NxYULF7TWN2aQiIiIdJEWh9hyc3MlxUZGRjAyMlJ5qlKpxIQJE/DKK6+gdevWYvnQoUPh7OwMR0dHnD17FtOmTUNiYiJ++uknAEB6erokOAIg7qenp2t2P49ggERERKSLBGghQHr4HycnJ0nxnDlzEBERofLU0NBQnD9/HseOHZOUjxkzRvy5TZs2cHBwQLdu3ZCUlISmTZtq1l81MEAiIiIijVy/fh0KhULcf1L2KCwsDDt27MCRI0fQqFEjlXV9fX0BAFeuXEHTpk1hb2+P06dPS+pkZGQAwGPnLVUH5yARERHporIhNk03AAqFQrI9LkASBAFhYWH4+eefceDAAbi6uj6xm/Hx8QAABwcHAICfnx/OnTuHzMxMsU50dDQUCgU8PDw0/FD+wwwSERGRLlIqAaj3HKPK26i60NBQbN26Fb/88gssLCzEOUOWlpYwMTFBUlIStm7dit69e6N+/fo4e/YsJk6ciE6dOsHT0xMAEBAQAA8PDwwfPhyRkZFIT0/HzJkzERoa+sTMlTqYQSIiIqKnYvXq1cjJyYG/vz8cHBzE7bvvvgMAyOVy7Nu3DwEBAWjZsiUmT56MAQMG4LfffhPb0NfXx44dO6Cvrw8/Pz+89dZbGDFihOS5SdrADBIREZEuqoUHRQpPqO/k5ITDhw8/sR1nZ2fs2rVLrWuriwESERGRLqqFAOl5wiE2IiIionKYQSIiItJFtfCqkecJAyQiIiIdJAhKCIJmq9g0Pf9ZxgCJiIhIFwmC5hkgzkEiIiIi0h3MIBEREekiQQtzkOpwBokBEhERkS5SKgGZhnOI6vAcJA6xEREREZXDDBIREZEu4hCbSgyQiIiIdJCgVELQcIitLi/z5xAbERERUTnMIBEREekiDrGpxACJiIhIFykFQMYA6XE4xEZERERUDjNIREREukgQAGj6HKS6m0FigERERKSDBKUAQcMhNoEBEhEREdUpghKaZ5C4zJ+IiIhIZzCDREREpIM4xKYaAyQiIiJdxCE2lRgg1TFl0XxJaWEt94So5ghCcW13gajGlODh97umszMlKNb4OZFlfa2LGCDVMffu3QMAHD2zpJZ7QkREmrh37x4sLS213q5cLoe9vT2Ope/SSnv29vaQy+VaaetZIhPq8gCiDlIqlbh58yYsLCwgk8lquzt1Xm5uLpycnHD9+nUoFIra7g6R1vE7/vQJgoB79+7B0dEReno1s5aqoKAARUVFWmlLLpfD2NhYK209S5hBqmP09PTQqFGj2u6GzlEoFPzlQXUav+NPV01kjh5lbGxcJ4MabeIyfyIiIqJyGCARERERlcMAiUgDRkZGmDNnDoyMjGq7K0Q1gt9x0lWcpE1ERERUDjNIREREROUwQCIiIiIqhwESERERUTkMkIgeIzg4GP369avx60RERKBt27aSsqKiIjRr1gwnTpyocjt79uxB27ZtoVTW3Xcj6bqUlBTIZDLEx8fXdlfU4uLigqVLl0rK9u/fD3d3d5SWlla5nQ8++ADvv/++lntHVDkGSFTn+fv7Y8KECWqft2zZMmzcuFHr/amKNWvWwNXVFS+//LJYlpWVhWHDhkGhUMDKygohISHIy8sTj/fs2ROGhobYsmVLbXSZSC1Tp07FzJkzoa+vDwBIS0vD0KFD0aJFC+jp6VX6ZzY8PBybNm3C1atXn3JvSRcxQKLnmrYelV8ZS0tLWFlZ1Vj7jyMIAlasWIGQkBBJ+bBhw3DhwgVER0djx44dOHLkCMaMGSOpExwcjOXLlz/N7lIdUJN/jipz7NgxJCUlYcCAAWJZYWEhbGxsMHPmTHh5eVV6XoMGDRAYGIjVq1c/ra6SDmOARM8Vf39/hIWFYcKECeJflufPn0evXr1gbm4OOzs7DB8+HLdv3wbwMGA4fPgwli1bBplMBplMhpSUFJSWliIkJASurq4wMTGBm5sbli1bJrlW+SE2f39/jBs3DlOnToW1tTXs7e0REREhOSc7OxujRo2CjY0NFAoFunbtijNnzkjqLFq0CHZ2drCwsEBISAgKCgokx+Pi4pCUlIQ+ffqIZQkJCdizZw/Wrl0LX19fdOjQAV988QWioqJw8+ZNsV7fvn0RGxuLpKQkTT5mqmVKpRKRkZFo1qwZjIyM0LhxY3z88cfi8atXr6JLly4wNTWFl5cXYmJixGOVDdkuXboULi4u4n7Zd/vjjz+Go6Mj3NzcxOG7n3766bFtAw+Dm44dO8LExAROTk4YN24c8vPzxeOZmZno27cvTExM4OrqWmlGMyoqCj169JC86sLFxQXLli3DiBEjVL5mo2/fvoiKinriZ0ikKQZI9NzZtGkT5HI5jh8/jkWLFqFr167w9vZGbGws9uzZg4yMDAwaNAjAw2EyPz8/jB49GmlpaUhLS4OTkxOUSiUaNWqE77//HhcvXsTs2bPx4YcfYtu2bU+8tpmZGU6dOoXIyEjMmzcP0dHR4vE33ngDmZmZ2L17N+Li4tCuXTt069YNWVlZAIBt27YhIiICCxYsQGxsLBwcHLBq1SrJNY4ePYoWLVrAwsJCLIuJiYGVlRXat28vlnXv3h16eno4deqUWNa4cWPY2dnh6NGj1f+AqdZNnz4dixYtwqxZs3Dx4kVs3boVdnZ24vEZM2YgPDwc8fHxaNGiBYYMGYKSkhK1rrF//34kJiaKGcmqtJ2UlISePXtiwIABOHv2LL777jscO3YMYWFh4vnBwcG4fv06Dh48iB9++AGrVq1CZmam5NpHjx6VfJfV8eKLL+LGjRtISUmp1vlEVSYQPUc6d+4seHt7i/vz588XAgICJHWuX78uABASExPFc8aPH//EtkNDQ4UBAwaI+0FBQcJrr70muXaHDh0k57zwwgvCtGnTBEEQhKNHjwoKhUIoKCiQ1GnatKnw5ZdfCoIgCH5+fsJ7770nOe7r6yt4eXmJ++PHjxe6du0qqfPxxx8LLVq0qNBnGxsbYdWqVZIyb29vISIi4gl3S8+q3NxcwcjISPj6668rHEtOThYACGvXrhXLLly4IAAQEhISBEEQhDlz5ki+T4IgCEuWLBGcnZ3F/aCgIMHOzk4oLCxUq+2QkBBhzJgxkraPHj0q6OnpCQ8ePBASExMFAMLp06fF4wkJCQIAYcmSJWKZpaWlsHnz5sd+Bqr+zObk5AgAhEOHDj32fCJtYAaJnjs+Pj7iz2fOnMHBgwdhbm4ubi1btgSAJw4zrVy5Ej4+PrCxsYG5uTm++uorpKamqjzH09NTsu/g4CD+6/jMmTPIy8tD/fr1Jf1JTk4W+5KQkABfX19JG35+fpL9Bw8eaPSWbRMTE9y/f7/a51PtSkhIQGFhIbp16/bYOo9+Dx0cHACgQpbmSdq0aQO5XK5W22fOnMHGjRsl3+/AwEAolUokJycjISEBBgYGkj+jLVu2rDCXT5PvuImJCQDwO041zqC2O0CkLjMzM/HnvLw89O3bF5988kmFemV/uVcmKioK4eHh+Oyzz+Dn5wcLCwssXrxYMlxVGUNDQ8m+TCYTl9Xn5eXBwcEBhw4dqnCeOpO9GzRogHPnzknK7O3tK/wCLCkpQVZWFuzt7SXlWVlZsLGxqfL16NlSFgCo8uj3UCaTAYD4PdTT04NQ7g1SxcXFFdp49M9RVdvOy8vDO++8g3HjxlU4r3Hjxrh06dIT+w48/I7fvXu3SnXLKxuu5necahoDJHqutWvXDj/++CNcXFxgYFD511kul1d41srx48fx8ssv47333hPLNJ3Y3K5dO6Snp8PAwEAyIfZR7u7uOHXqFEaMGCGWnTx5UlLH29sbq1evhiAI4i8oPz8/ZGdnIy4uTvzX+YEDB6BUKiUZqYKCAiQlJcHb21uje6Ha07x5c5iYmGD//v0YNWqU2ufb2NggPT1d8v3R1nOT2rVrh4sXL6JZs2aVHm/ZsiVKSkoQFxeHF154AQCQmJiI7OxsST1vb29cvHixWn04f/48DA0N0apVq2qdT1RVHGKj51poaCiysrIwZMgQ/PHHH0hKSsLvv/+Ot99+WwyKXFxccOrUKaSkpOD27dtQKpVo3rw5YmNj8fvvv+PSpUuYNWsW/vjjD4360r17d/j5+aFfv37Yu3cvUlJScOLECcyYMQOxsbEAgPHjx2P9+vXYsGEDLl26hDlz5uDChQuSdrp06YK8vDxJubu7O3r27InRo0fj9OnTOH78OMLCwjB48GA4OjqK9U6ePAkjI6MKw3b0/DA2Nsa0adMwdepUbN68GUlJSTh58iTWrVtXpfP9/f1x69YtREZGIikpCStXrsTu3bu10rdp06bhxIkTCAsLQ3x8PC5fvoxffvlFnKTt5uaGnj174p133sGpU6cQFxeHUaNGVciKBQYG4tixYxXaj4+PR3x8PPLy8nDr1i3Ex8dXCKSOHj0qrqIjqkkMkOi55ujoiOPHj6O0tBQBAQFo06YNJkyYACsrK+jpPfx6h4eHQ19fHx4eHrCxsUFqaireeecd9O/fH2+++SZ8fX1x584dSTapOmQyGXbt2oVOnTrh7bffRosWLTB48GBcu3ZNXIH05ptvYtasWZg6dSp8fHxw7do1jB07VtJO/fr18frrr1dYHr1lyxa0bNkS3bp1Q+/evdGhQwd89dVXkjrffvsthg0bBlNTU43uhWrXrFmzMHnyZMyePRvu7u548803qzzHyN3dHatWrcLKlSvh5eWF06dPIzw8XCv98vT0xOHDh3Hp0iV07NgR3t7emD17tiRI37BhAxwdHdG5c2f0798fY8aMga2traSdsmd6JSYmSsq9vb3h7e2NuLg4bN26Fd7e3ujdu7ekTlRUFEaPHq2V+yFSRSaUH6wmolp39uxZ9OjRA0lJSTA3N6/SObdv34abmxtiY2Ph6upawz0k0syUKVOQm5uLL7/8ssrn7N69G5MnT8bZs2cfO6ROpC3MIBE9gzw9PfHJJ58gOTm5yuekpKRg1apVDI7ouTBjxgw4Ozur9e7A/Px8bNiwgcERPRXMIBERERGVwwwSERERUTkMkIiIiIjKYYBEREREVA4DJCIiIqJyGCARERERlcMAiYi0Ljg4GP369RP3/f39MWHChKfej0OHDkEmk1V41cWjZDIZtm/fXuU2IyIi0LZtW436lZKSAplMprVXgBCR9jFAItIRwcHBkMlkkMlkkMvlaNasGebNm4eSkpIav/ZPP/2E+fPnV6luVYIaIqKaxqdtEemQnj17YsOGDSgsLMSuXbsQGhoKQ0NDTJ8+vULdoqIiyOVyrVzX2tpaK+0QET0tzCAR6RAjIyPY29vD2dkZY8eORffu3fHrr78C+G9Y7OOPP4ajoyPc3NwAANevX8egQYNgZWUFa2trvPbaa0hJSRHbLC0txaRJk2BlZYX69etj6tSpKP/82fJDbIWFhZg2bRqcnJxgZGSEZs2aYd26dUhJSUGXLl0AAPXq1YNMJkNwcDAAQKlUYuHChXB1dYWJiQm8vLzwww8/SK6za9cutGjRAiYmJujSpYukn1U1bdo0tGjRAqampmjSpAlmzZqF4uLiCvW+/PJLODk5wdTUFIMGDUJOTo7k+Nq1a+Hu7g5jY2O0bNkSq1atUrsvRFR7GCAR6TATExMUFRWJ+/v370diYiKio6OxY8cOFBcXIzAwEBYWFjh69CiOHz8Oc3Nz9OzZUzzvs88+w8aNG7F+/XocO3YMWVlZ+Pnnn1Ved8SIEfj222+xfPlyJCQk4Msvv4S5uTmcnJzw448/AgASExORlpaGZcuWAQAWLlyIzZs3Y82aNbhw4QImTpyIt956C4cPHwbwMJDr378/+vbti/j4eIwaNQoffPCB2p+JhYUFNm7ciIsXL2LZsmX4+uuvsWTJEkmdK1euYNu2bfjtt9+wZ88e/PXXX5KXHW/ZsgWzZ8/Gxx9/jISEBCxYsACzZs3Cpk2b1O4PEdUSgYh0QlBQkPDaa68JgiAISqVSiI6OFoyMjITw8HDxuJ2dnVBYWCie88033whubm6CUqkUywoLCwUTExPh999/FwRBEBwcHITIyEjxeHFxsdCoUSPxWoIgCJ07dxbGjx8vCIIgJCYmCgCE6OjoSvt58OBBAYBw9+5dsaygoEAwNTUVTpw4IakbEhIiDBkyRBAEQZg+fbrg4eEhOT5t2rQKbZUHQPj5558fe3zx4sWCj4+PuD9nzhxBX19fuHHjhli2e/duQU9PT0hLSxMEQRCaNm0qbN26VdLO/PnzBT8/P0EQBCE5OVkAIPz111+PvS4R1S7OQSLSITt27IC5uTmKi4uhVCoxdOhQREREiMfbtGkjmXd05swZXLlyBRYWFpJ2CgoKkJSUhJycHKSlpcHX11c8ZmBggPbt21cYZisTHx8PfX19dO7cucr9vnLlCu7fv48ePXpIyouKiuDt7Q0ASEhIkPQDAPz8/Kp8jTLfffcdli9fjqSkJOTl5aGkpAQKhUJSp3HjxmjYsKHkOkqlEomJibCwsEBSUhJCQkIwevRosU5JSQksLS3V7g8R1Q4GSEQ6pEuXLli9ejXkcjkcHR0rvBXdzMxMsp+XlwcfHx9s2bKlQls2NjbV6oOJiYna5+Tl5QEAdu7cKQlMgIfzqrQlJiYGw4YNw9y5cxEYGAhLS0tERUXhs88+U7uvX3/9dYWATV9fX2t9JaKaxQCJSIeYmZmhWbNmVa7frl07fPfdd7C1ta2QRSnj4OCAU6dOoVOnTgAeZkri4uLQrl27Suu3adMGSqUShw8fRvfu3SscL8tglZaWimUeHh4wMjJCamrqYzNP7u7u4oTzMidPnnzyTT7ixIkTcHZ2xowZM8Sya9euVaiXmpqKmzdvwtHRUbyOnp4e3NzcYGdnB0dHR1y9ehXDhg1T6/pE9OzgJG0ieqxhw4ahQYMGeO2113D06FEkJyfj0KFDGDduHG7cuAEAGD9+PBYtWoTt27fj77//xnvvvafyGUYuLi4ICgrCyJEjsX37drHNbdu2AQCcnZ0hk8mwY8cO3Lp1C3l5ebCwsEB4eDgmTpyITZs2ISkpCX/++Se++OILceLzu+++i8uXL2PKlClITEzE1q1bsXHjRrXut3nz5khNTUVUVBSSkpKwfPnySiecGxsbIygoCGfOnMHRo0cxbtw4DBo0CPb29gCAuXPnYuHChVi+fDkuXbqEc+fOYcOGDfj888/V6g8R1R4GSET0WKampjhy5AgaN26M/v37w93dHSEhISgoKBAzSpMnT8bw4cMRFBQEPz8/WFhY4PXXX1fZ7urVqzFw4EC89957aNmyJUaPHo38/HwAQMOGDTF37lx88MEHsLOzQ1hYGABg/vz5mDVrFhYuXAh3d3f07NkTO3fuhKurK4CH84J+/PFHbN++HV5eXlizZg0WLFig1v2++uqrmDhxIsLCwtC2bVucOHECs2bNqlCvWbNm6N+/P3r37o2AgAB4enpKlvGPGjUKa9euxYYNG9CmTRt07twZGzduFPtKRM8+mfC4mZREREREOooZJCIiIqJyGCARERERlcMAiYiIiKgcBkhERERE5TBAIiIiIiqHARIRERFROQyQiIiIiMphgERERERUDgMkIiIionIYIBERERGVwwCJiIiIqBwGSERERETl/D9eMa+EkyWGjQAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "cm = confusion_matrix(y_test, XGB_test_preds, labels=XGB_cv.classes_)\n",
    "\n",
    "# Plot confusion matrix\n",
    "XGB_disp = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=['retained(0)', 'churned(1)'])\n",
    "XGB_disp.plot()\n",
    "plt.title('Retention Predictions');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "*Legend*\n",
    "* 0 = the person was labeled as retained for that observation\n",
    "* 1 = a person was labeled as churned for that observation\n",
    "\n",
    "*Translation*<br>\n",
    "`True negatives`(UL): # of people that were retained(0) that the model accurately predicted as retained(0) = 2221   \n",
    "`True positives`(LR): # of people that churned(1) that the model accurately predicted churned(1) = 88   \n",
    "`False positives`(UR)/ Type I Errors: # of people that were retained(0) that the model inaccurately predicted churned(1) = 132   \n",
    "`False negatives`(LL)/Type II Errors: # of people that churned(1) that the model inaccurately predicted were retained(0) = 419   \n",
    "\n",
    "The XGBoost model both predicted more users who would churn and missed more users who would churn (Type I Errors) than the logistic regression model. It also correctly predicted fewer users who would stay, while at the same time had fewer false identifications of those who would leave."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Plot ROC and AOC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot ROC curve\n",
    "RocCurveDisplay.from_predictions(y_test, XGB_test_preds)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.5587357110944021"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Compute the AUC\n",
    "roc_auc_score(y_test, XGB_test_preds)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Results    \n",
    "The champion XGBoost model performed slightly better (0.54 > 0.56) across all possible classification thresholds (classification decisions) than the logistic regression model. \n",
    "\n",
    "With an AUC score of ~56%, it performs ~6% better than random classification (50%)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Plot feature importance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_importance(XGB_cv.best_estimator_);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "EU3GIZNrga5z"
   },
   "source": [
    "#### Results    \n",
    "According to the feature importance plot, the rate at which a driver drove (`km_per_hour`) and whether or not they took advantage of the favorite navigations feature (`total_navigations_fav1`) were the most predictive features in the data set. Second were `percent_sessions_in_month` and `n_days_after_onboarding`. The created feature `professional_driver` had little to no impact on predictability. Also, the device a user used (`device2`) was not relevant for the model's calculations.\n",
    "\n",
    "However, relying on the results of a feature importance rank in an unbalanced data such as this one is not wise, since it can be biased towards the majority class. This is especially true when it is the minority class that is the target. Feature importance may prioritize features that are more prevalent in the majority class, thereby missing those features crucial for predicting the minority class."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Plot precision-recall curve"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import PrecisionRecallDisplay"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "id": "kVgiwBWtSxq4"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Compute precision-recall pairs for different probability thresholds\n",
    "display = PrecisionRecallDisplay.from_estimator(XGB_cv.best_estimator_, X_test, y_test, name='XGBoost')\n",
    "plt.title('XGBoost model Precision-recall curve');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "wxyauTCNSyND"
   },
   "source": [
    "Results    \n",
    "Precision is a metric that quantifies the number of correct positive predictions made. \\\n",
    "Recall is a metric that quantifies the number of correct positive predictions made out of all positive predictions that could have been made.\n",
    "\n",
    "As recall increases, precision decreases. \n",
    "\n",
    "The precision-recall curve plots precision (y-axis) and recall (x-axis) for different probability thresholds. In a balanced dataset this is at 50% by default. The \"high\" rating of the ROC curve reflects the majority class. Here, the curve is more \"honest,\" with a 32% score.\n",
    "\n",
    "Since a false positive is less of a concern for this data set because the goal is to predicate users who churn (the minority) accidently reaching out to a user would wouldn't even without contact would likely not have major negative consequences = low risk. The 50% threshold should be lowered."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Create new threshold**\n",
    "* the 1st # in the row is the probability of belonging to the negative class\n",
    "* the 2nd # in the row is the probability of belonging to the positive class\n",
    "* together they will = 1\n",
    "\n",
    "Changing the threshold of a positive response where any users who have a value ≥ 0.4 (as opposed to ≥ 0.5) in the 2nd column will get assigned a prediction of `1` = they churned."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "id": "iiKAYfRwS1bW"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.97620225, 0.02379777],\n",
       "       [0.7518184 , 0.24818155],\n",
       "       [0.9936058 , 0.00639419],\n",
       "       ...,\n",
       "       [0.8098501 , 0.19014993],\n",
       "       [0.9927123 , 0.00728769],\n",
       "       [0.7999854 , 0.20001458]], dtype=float32)"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Get predicted probabilities on the test data to return a 2-D array of probabilities where each row represents a user\n",
    "predicted_probabilities = XGB_cv.best_estimator_.predict_proba(X_test)\n",
    "predicted_probabilities"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {
    "id": "ynzlIds4S1xi"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 0, 0, ..., 0, 0, 0])"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Create a probs list for the 2nd column\n",
    "probs = [x[1] for x in predicted_probabilities]\n",
    "\n",
    "# Create an array of new predictions that assigns a 1 to any value >= 0.4\n",
    "new_preds = np.array([1 if x >= 0.4 else 0 for x in probs])\n",
    "new_preds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {
    "id": "pYAKZbeqS5Vr"
   },
   "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>Model</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>F1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>XGB, threshold = 0.4</td>\n",
       "      <td>0.36858</td>\n",
       "      <td>0.240631</td>\n",
       "      <td>0.792308</td>\n",
       "      <td>0.291169</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  Model  Precision    Recall  Accuracy        F1\n",
       "0  XGB, threshold = 0.4    0.36858  0.240631  0.792308  0.291169"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Get evaluation metrics for new threshold\n",
    "get_test_scores('XGB, threshold = 0.4', new_preds, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {
    "id": "i-tsYPiJTzdn"
   },
   "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>Model</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>F1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>XGBoost Val</td>\n",
       "      <td>0.412037</td>\n",
       "      <td>0.175542</td>\n",
       "      <td>0.809441</td>\n",
       "      <td>0.246196</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>XGBoost TEST</td>\n",
       "      <td>0.400000</td>\n",
       "      <td>0.173570</td>\n",
       "      <td>0.807343</td>\n",
       "      <td>0.242091</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>XGBoost cv</td>\n",
       "      <td>0.404893</td>\n",
       "      <td>0.171484</td>\n",
       "      <td>0.808135</td>\n",
       "      <td>0.240830</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Random Forest cv</td>\n",
       "      <td>0.454665</td>\n",
       "      <td>0.128763</td>\n",
       "      <td>0.818160</td>\n",
       "      <td>0.200545</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Random Forest Val</td>\n",
       "      <td>0.423077</td>\n",
       "      <td>0.108481</td>\n",
       "      <td>0.815734</td>\n",
       "      <td>0.172684</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               Model  Precision    Recall  Accuracy        F1\n",
       "0        XGBoost Val   0.412037  0.175542  0.809441  0.246196\n",
       "5       XGBoost TEST   0.400000  0.173570  0.807343  0.242091\n",
       "1         XGBoost cv   0.404893  0.171484  0.808135  0.240830\n",
       "2   Random Forest cv   0.454665  0.128763  0.818160  0.200545\n",
       "3  Random Forest Val   0.423077  0.108481  0.815734  0.172684"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# View results table\n",
    "models_table = pd.read_csv('Retention_Results.csv')\n",
    "models_table = models_table.sort_values(by=['Recall'], ascending=False)\n",
    "models_table.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "F_Hl_g6rTz5o"
   },
   "source": [
    "Result \\\n",
    "Recall went up significantly as did the combined F1 score. In spite of the fact that Precision went down, Accuracy still held at ~80%."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {
    "id": "ahSvceazUsnP"
   },
   "outputs": [],
   "source": [
    "# Create function to arrive at perfect threshold for desired Recall score\n",
    "def threshold_finder(y_test_data, probabilities, desired_recall): # y_test_data: Array of true y values, probabilities:  results of `predict_proba()`\n",
    "\n",
    "    probs = [x[1] for x in probabilities]  # Isolate 2nd column of `probabilities`\n",
    "    thresholds = np.arange(0, 1, 0.001)    # Set a grid of 1,000 thresholds to test\n",
    "\n",
    "    scores = []\n",
    "    for threshold in thresholds:\n",
    "        preds = np.array([1 if x >= threshold else 0 for x in probs]) # new array of {0, 1} predictions based on new threshold\n",
    "         \n",
    "        recall = recall_score(y_test_data, preds) # Calculate recall score for that threshold\n",
    "       \n",
    "    # Append the threshold and its corresponding recall score as a tuple to `scores`\n",
    "        scores.append((threshold, recall))\n",
    "        distances = []\n",
    "        for idx, score in enumerate(scores):\n",
    "            distance = abs(score[1] - desired_recall) # Calculates how close each actual score is to the desired score\n",
    "            distances.append((idx, distance)) # Appends the (index#, distance) tuple to `distances`\n",
    "\n",
    "    sorted_distances = sorted(distances, key=lambda x: x[1], reverse=False) # Sorts `distances` by 2nd value in each of its tuples (< to >)\n",
    "   \n",
    "    best = sorted_distances[0] # Identify the tuple with actual recall closest to desired recall\n",
    "    \n",
    "    best_idx = best[0] # Isolate index of threshold with closest recall score\n",
    "    \n",
    "    threshold, recall = scores[best_idx]  # Retrieve threshold and actual recall score closest to desired recall\n",
    "\n",
    "    return threshold, recall"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {
    "id": "rVlW592vT_kT"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.134, 0.5009861932938856)"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Get the predicted probabilities best XGBoost model\n",
    "probabilities = XGB_cv.best_estimator_.predict_proba(X_test)\n",
    "\n",
    "# Call threshold_finder function\n",
    "threshold_finder(y_test, probabilities, 0.5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "gZkp1FY3Uw22"
   },
   "source": [
    "Setting a threshold of 0.134 will result in a recall of 0.50."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {
    "id": "2PuIr8byUx8y"
   },
   "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>Model</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Accuracy</th>\n",
       "      <th>F1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>XGB, threshold = 0.134</td>\n",
       "      <td>0.297773</td>\n",
       "      <td>0.500986</td>\n",
       "      <td>0.702098</td>\n",
       "      <td>0.373529</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                    Model  Precision    Recall  Accuracy        F1\n",
       "0  XGB, threshold = 0.134   0.297773  0.500986  0.702098  0.373529"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Create an array of new predictions that assigns a 1 to any value >= 0.124\n",
    "probs = [x[1] for x in probabilities]\n",
    "new_preds = np.array([1 if x >= 0.134 else 0 for x in probs])\n",
    "\n",
    "# Get evaluation metrics for when the threshold is 0.124\n",
    "get_test_scores('XGB, threshold = 0.134', new_preds, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "cm = confusion_matrix(y_test, new_preds, labels=XGB_cv.classes_)\n",
    "\n",
    "# Plot confusion matrix\n",
    "XGB_disp = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=['retained(0)', 'churned(1)'])\n",
    "XGB_disp.plot()\n",
    "plt.title('Retention Predictions');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Results    \n",
    "*Metrics* \\\n",
    "With the new threshold, Precision drops precipitously, however, depending on the cost expended to reach ot to users who might churn this could be an acceptable outcome. Accuracy is still relatively high at 70%.\n",
    "\n",
    "*Confusion matrix* \\\n",
    "True negatives(UL): # of people that were retained(0) that the model accurately predicted as retained(0) = 1774 (< 2221)\\\n",
    "True positives(LR): # of people that churned(1) that the model accurately predicted churned(1) = 254 (> 88)\\\n",
    "False positives(UR)/ Type I Errors: # of people that were retained(0) that the model inaccurately predicted churned(1) = 599 (> 132)\\\n",
    "False negatives(LL)/Type II Errors: # of people that churned(1) that the model inaccurately predicted were retained(0) = 253 (< 419)\n",
    "\n",
    "With the new threshold, 166 more churned users were identified, although the number of incorrectly users predicted to churn rose significantly (Type I Error); however, the number of Type II Errors decreased from 419 to 253, meaning that with the lower threshold, the model predicts significantly better.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Results of the winning XGBoost model on the test data set \n",
    "Out of the 2,860 users in the test data set, 507 churned. Of these, the XGBoost model captured 88, thereby achieving an ~18% Recall. The test data set also included 2353 remaining users. Of these, the model captured 2221 – missing only 132. By refitting the model to Recall, this also drove the F1 up from the logistic regression model to ~25%. This metric balances the scores the model received in predicting either churned or retained users.\n",
    " \n",
    "Since the goal of this project is to build a model which can predict whether a user will churn using the available variables, this model’s performance is less than stellar. However, these results are somewhat expected in a data set as unbalanced as this one is. \n",
    "\n",
    "*Boosting results further* \\\n",
    "In order to account for the imbalance in the dataset, the threshold at which the model classified a user as likely to churn was adjusted. Targeting a 50% Recall required the threshold be set to 0.134 and this provided far more satisfactory results. With this new threshold setting, out of the 507 churned users in the test data the model now captured 254 (166 more users) and the number of Type II errors decreased from 419 to 253, meaning that with the lower threshold the model predicts significantly better. This altered threshold did mean that the number of users incorrectly predicted to churn rose significantly (Type I error), lowering Precision from ~41% to ~30%. In an effort to identify more customers, the model will become more sensitive to finding churned users and will become prone to incorrectly categorizing people as churned (false-positives/Type-I error) when they remained. This then lowers the Precision score.\n",
    "\n",
    "Finally, although Accuracy also decreased to 70% (from ~80%), F1 increased substantially from ~25% to ~37%. This means that on the whole the model was far better at achieving the desired result: locating users who would churn.\n",
    "\n",
    "*Most predictive features* \\\n",
    "According to the feature importance plot, the rate at which a driver drove (`km_per_hour`) and whether or not they took advantage of the favorite navigations feature (`total_navigations_fav1`) were the most predictive features in the data set. Second were `percent_sessions_in_month` and `n_days_after_onboarding`. The created feature `professional_driver` had little to no impact on predictability. Also, the device a user used (`device2`) was not relevant for the model's calculations.\n",
    " \n",
    "However, relying on the results of a feature importance rank in an unbalanced data set such as this one is not wise, since it can be biased towards the majority class. This is especially true when it is the minority class that is the target. Feature importance may prioritize features that are more prevalent in the majority class, thereby missing those features crucial for predicting the minority class.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Going Forward\n",
    "\n",
    "As is, this XGBoost model with the revised threshold could be used to predict users who churn, if the company’s retention strategy is cost productive per user. This is because the model will make more Type-I errors. It will classify users as likely to churn when they would not, regardless of intervention. \n",
    "\n",
    "* Alternatively, the data could be scaled so that the minority class (churned users) is more rep-resented or majority class deemphasized.\n",
    "* Or, new models could be built using more or different features. For example, the data set does not contain true records of total drives.\n",
    "* Finally, it might be helpful to learn why so many sessions seem to have taken place in the last month."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "provenance": [
    {
     "file_id": "1DHsmIEwaXUmfVT4tFzyOwyyfXAX0v6IF",
     "timestamp": 1675262571681
    },
    {
     "file_id": "1oNheYh5WbljxkvoK_BMkQTey2DWnFXMs",
     "timestamp": 1674856595373
    }
   ]
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}