online_course_pred_rxu17.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Is student X going to complete their online course?\n",
    "\n",
    "### Description: For this project, we are going to predict whether a student will complete (be certified) in their course depending on a number of factors. \n",
    "### How to Run: Click run in jupyter notebook\n",
    "### Author(s): rxu17\n",
    "### Team Name (Kaggle): RXU\n",
    "### Date: 08/01/21\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Import important required libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "# import general libraries\n",
    "import os\n",
    "import sklearn\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "# for feature engineering\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "from pandas.api.types import is_numeric_dtype\n",
    "from sklearn.impute import KNNImputer\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "\n",
    "\n",
    "# for splitting data\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "# for our various model trainings\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.ensemble import RandomForestClassifier, AdaBoostClassifier\n",
    "\n",
    "# scoring\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Read in our training data and test data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "# input in filepath of input data\n",
    "input_path = os.getcwd()\n",
    "\n",
    "# validate path\n",
    "assert os.path.isdir(input_path),\\\n",
    "    \"{} is not a valid filepath\".format(input_path)\n",
    "\n",
    "# read in our data\n",
    "train_and_validate_data = pd.read_csv(\"{}/edx_su2021_train.csv\".format(input_path))\n",
    "test_data = pd.read_csv(\"{}/edx_su2021_test.csv\".format(input_path))\n",
    "\n",
    "assert not train_and_validate_data.empty and not test_data.empty,\\\n",
    "    \"training and/or test data is empty!\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Splitting our data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "# our features and descriptions of each\n",
    "features = ['registered', # 0/1; registered for course\n",
    "            'viewed', # 0/1; anyone who accessed the ‘Courseware’ tab\n",
    "            'explored',# 0/1; anyone who accessed at least half of the chapters in the courseware\n",
    "            'LoE_DI', # highest level of education completed\n",
    "            'YoB', # year of birth\n",
    "            'gender', # Possible values: m (male), f (female) and o (other)\n",
    "            'grade', #  final grade in the course, ranges from 0 to 1\n",
    "            'start_time_DI', # date of course registration\n",
    "            'last_event_DI', # date of last interaction with course\n",
    "            'nevents', # number of interactions with the course\n",
    "            'ndays_act', # number of unique days student interacted with course\n",
    "            'nplay_video', # number of play video events within the course\n",
    "            'nchapters', # number of chapters (within the Courseware) with which the student interacted.\n",
    "            'nforum_posts' # number of posts to the Discussion Forum\n",
    "           ]\n",
    "\n",
    "target = \"certified\"\n",
    "# split our data into 80% training, 20% validation for now\n",
    "validate_size = 0.2\n",
    "train_data, validation_data = \\\n",
    "    train_test_split(train_and_validate_data, test_size=validate_size, random_state=6)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Descriptive Statistics"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here we start looking into specific aspects of our dataset such as datatypes of the columns, range of values in each column, shape of our dataset, and a subset of our dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "      registered  viewed  explored      LoE_DI     YoB gender grade  \\\n",
      "2030           1       1         0   Secondary  1988.0      m     0   \n",
      "5247           1       1         1    Master's  1981.0      f  0.95   \n",
      "3480           1       1         0  Bachelor's  1992.0      m  0.31   \n",
      "7076           1       0         0  Bachelor's  1981.0      m     0   \n",
      "8640           1       1         0   Secondary  1991.0      m     0   \n",
      "\n",
      "     start_time_DI last_event_DI  nevents  ndays_act  nplay_video  nchapters  \\\n",
      "2030      11/30/12       4/22/13      8.0        1.0          NaN        1.0   \n",
      "5247      10/24/12       1/16/13   5255.0       48.0        445.0       16.0   \n",
      "3480      11/19/12       1/15/13    278.0        3.0          NaN        3.0   \n",
      "7076       8/27/12           NaN      NaN        NaN          NaN        NaN   \n",
      "8640      10/17/12       11/3/12    148.0        5.0         25.0        1.0   \n",
      "\n",
      "      nforum_posts  \n",
      "2030             0  \n",
      "5247             0  \n",
      "3480             0  \n",
      "7076             0  \n",
      "8640             0  \n"
     ]
    }
   ],
   "source": [
    "# sneak peak of our training data\n",
    "print(train_data[features].head())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "registered         int64\n",
      "viewed             int64\n",
      "explored           int64\n",
      "LoE_DI            object\n",
      "YoB              float64\n",
      "gender            object\n",
      "grade             object\n",
      "start_time_DI     object\n",
      "last_event_DI     object\n",
      "nevents          float64\n",
      "ndays_act        float64\n",
      "nplay_video      float64\n",
      "nchapters        float64\n",
      "nforum_posts       int64\n",
      "dtype: object\n"
     ]
    }
   ],
   "source": [
    "# data types of the columns\n",
    "print(train_data[features].dtypes)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note that grade is shown as having decimal values but is an object/string column later. We'll have to add this to the features we will engineer later. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "registered                                                     [1]\n",
      "viewed                                                      [1, 0]\n",
      "explored                                                    [0, 1]\n",
      "LoE_DI           [Secondary, Master's, Bachelor's, Less than Se...\n",
      "YoB              [1988.0, 1981.0, 1992.0, 1991.0, 1990.0, 1984....\n",
      "gender                                                 [m, f, nan]\n",
      "grade            [0, 0.95, 0.31, 0.6, 1, 0.92, 0.75, 0.89, 0.86...\n",
      "start_time_DI    [11/30/12, 10/24/12, 11/19/12, 8/27/12, 10/17/...\n",
      "last_event_DI    [4/22/13, 1/16/13, 1/15/13, nan, 11/3/12, 10/2...\n",
      "nevents          [8.0, 5255.0, 278.0, nan, 148.0, 1.0, 1743.0, ...\n",
      "ndays_act        [1.0, 48.0, 3.0, nan, 5.0, 22.0, 51.0, 34.0, 2...\n",
      "nplay_video      [nan, 445.0, 25.0, 129.0, 2.0, 270.0, 1828.0, ...\n",
      "nchapters        [1.0, 16.0, 3.0, nan, 9.0, 12.0, 2.0, 28.0, 6....\n",
      "nforum_posts                              [0, 1, 5, 3, 2, 6, 4, 7]\n",
      "dtype: object\n"
     ]
    }
   ],
   "source": [
    "# print the unique values in each columns\n",
    "print(train_data[features].apply(lambda col: col.unique()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(7006, 14)\n"
     ]
    }
   ],
   "source": [
    "# shape of our dataset\n",
    "print(train_data[features].shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "registered        0.000000\n",
      "viewed            0.000000\n",
      "explored          0.000000\n",
      "LoE_DI           10.862118\n",
      "YoB              10.362546\n",
      "gender            8.792464\n",
      "grade             4.125036\n",
      "start_time_DI     0.000000\n",
      "last_event_DI    17.028261\n",
      "nevents          18.512703\n",
      "ndays_act        18.512703\n",
      "nplay_video      76.862689\n",
      "nchapters        29.146446\n",
      "nforum_posts      0.000000\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "# see proportion of nans in each columns\n",
    "print(train_data[features].apply(lambda col: col.isna().sum()*100/len(train_data)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "nplay_video is full of nas > 70% missing here. We'll see if the data has any correlation with our target var otherwise, it sounds worthwhile to drop this variable as a predictor. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number certified: 3143 (44.86%)\n",
      "Number non-certified: 3863 (55.14%)\n"
     ]
    }
   ],
   "source": [
    "# See distribution of certified values\n",
    "cert = train_data[train_data['certified'] == 1]\n",
    "not_cert = train_data[train_data['certified'] ==0]\n",
    "\n",
    "print(f'Number certified: {len(cert)} ({len(cert) * 100.0 / len(train_data):.2f}%)')\n",
    "print(f'Number non-certified: {len(not_cert)} ({len(not_cert) * 100.0 / len(train_data):.2f}%)')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We have a pretty good balance between certified and non-certified numbers here"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['registered', 'viewed', 'explored', 'YoB', 'nevents', 'ndays_act', 'nplay_video', 'nchapters', 'nforum_posts']\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# only float or int columns and produce a heatmap\n",
    "num_cols = train_data[features].select_dtypes(include=np.number).columns.tolist()\n",
    "print(num_cols)\n",
    "sns.heatmap(train_data[features][num_cols].corr(), annot=True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "scrolled": true
   },
   "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>registered</th>\n",
       "      <th>viewed</th>\n",
       "      <th>explored</th>\n",
       "      <th>YoB</th>\n",
       "      <th>nevents</th>\n",
       "      <th>ndays_act</th>\n",
       "      <th>nplay_video</th>\n",
       "      <th>nchapters</th>\n",
       "      <th>nforum_posts</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>registered</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>viewed</th>\n",
       "      <td>NaN</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.564746</td>\n",
       "      <td>-0.014121</td>\n",
       "      <td>0.216304</td>\n",
       "      <td>0.335358</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.039953</td>\n",
       "      <td>0.092296</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>explored</th>\n",
       "      <td>NaN</td>\n",
       "      <td>0.564746</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.059479</td>\n",
       "      <td>0.477613</td>\n",
       "      <td>0.710598</td>\n",
       "      <td>0.321807</td>\n",
       "      <td>0.710468</td>\n",
       "      <td>0.133175</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>YoB</th>\n",
       "      <td>NaN</td>\n",
       "      <td>-0.014121</td>\n",
       "      <td>-0.059479</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.108710</td>\n",
       "      <td>-0.134370</td>\n",
       "      <td>-0.088240</td>\n",
       "      <td>-0.076464</td>\n",
       "      <td>0.035443</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>nevents</th>\n",
       "      <td>NaN</td>\n",
       "      <td>0.216304</td>\n",
       "      <td>0.477613</td>\n",
       "      <td>-0.108710</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.693119</td>\n",
       "      <td>0.837960</td>\n",
       "      <td>0.289292</td>\n",
       "      <td>0.092803</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ndays_act</th>\n",
       "      <td>NaN</td>\n",
       "      <td>0.335358</td>\n",
       "      <td>0.710598</td>\n",
       "      <td>-0.134370</td>\n",
       "      <td>0.693119</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.613114</td>\n",
       "      <td>0.490699</td>\n",
       "      <td>0.119665</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>nplay_video</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.321807</td>\n",
       "      <td>-0.088240</td>\n",
       "      <td>0.837960</td>\n",
       "      <td>0.613114</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.442048</td>\n",
       "      <td>0.036803</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>nchapters</th>\n",
       "      <td>NaN</td>\n",
       "      <td>0.039953</td>\n",
       "      <td>0.710468</td>\n",
       "      <td>-0.076464</td>\n",
       "      <td>0.289292</td>\n",
       "      <td>0.490699</td>\n",
       "      <td>0.442048</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.165796</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>nforum_posts</th>\n",
       "      <td>NaN</td>\n",
       "      <td>0.092296</td>\n",
       "      <td>0.133175</td>\n",
       "      <td>0.035443</td>\n",
       "      <td>0.092803</td>\n",
       "      <td>0.119665</td>\n",
       "      <td>0.036803</td>\n",
       "      <td>0.165796</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              registered    viewed  explored       YoB   nevents  ndays_act  \\\n",
       "registered           NaN       NaN       NaN       NaN       NaN        NaN   \n",
       "viewed               NaN  1.000000  0.564746 -0.014121  0.216304   0.335358   \n",
       "explored             NaN  0.564746  1.000000 -0.059479  0.477613   0.710598   \n",
       "YoB                  NaN -0.014121 -0.059479  1.000000 -0.108710  -0.134370   \n",
       "nevents              NaN  0.216304  0.477613 -0.108710  1.000000   0.693119   \n",
       "ndays_act            NaN  0.335358  0.710598 -0.134370  0.693119   1.000000   \n",
       "nplay_video          NaN       NaN  0.321807 -0.088240  0.837960   0.613114   \n",
       "nchapters            NaN  0.039953  0.710468 -0.076464  0.289292   0.490699   \n",
       "nforum_posts         NaN  0.092296  0.133175  0.035443  0.092803   0.119665   \n",
       "\n",
       "              nplay_video  nchapters  nforum_posts  \n",
       "registered            NaN        NaN           NaN  \n",
       "viewed                NaN   0.039953      0.092296  \n",
       "explored         0.321807   0.710468      0.133175  \n",
       "YoB             -0.088240  -0.076464      0.035443  \n",
       "nevents          0.837960   0.289292      0.092803  \n",
       "ndays_act        0.613114   0.490699      0.119665  \n",
       "nplay_video      1.000000   0.442048      0.036803  \n",
       "nchapters        0.442048   1.000000      0.165796  \n",
       "nforum_posts     0.036803   0.165796      1.000000  "
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cor_mat = train_data[features].corr()\n",
    "cor_mat"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here we're going to investigate into specific features of interest starting with our last_event_DI"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "scrolled": true
   },
   "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>registered</th>\n",
       "      <th>viewed</th>\n",
       "      <th>explored</th>\n",
       "      <th>LoE_DI</th>\n",
       "      <th>YoB</th>\n",
       "      <th>gender</th>\n",
       "      <th>grade</th>\n",
       "      <th>start_time_DI</th>\n",
       "      <th>last_event_DI</th>\n",
       "      <th>nevents</th>\n",
       "      <th>ndays_act</th>\n",
       "      <th>nplay_video</th>\n",
       "      <th>nchapters</th>\n",
       "      <th>nforum_posts</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>7076</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>Bachelor's</td>\n",
       "      <td>1981.0</td>\n",
       "      <td>m</td>\n",
       "      <td>0</td>\n",
       "      <td>8/27/12</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4877</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>Bachelor's</td>\n",
       "      <td>1989.0</td>\n",
       "      <td>m</td>\n",
       "      <td>0</td>\n",
       "      <td>12/24/12</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4749</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>Secondary</td>\n",
       "      <td>1991.0</td>\n",
       "      <td>m</td>\n",
       "      <td>0</td>\n",
       "      <td>8/7/12</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>783</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>Secondary</td>\n",
       "      <td>1991.0</td>\n",
       "      <td>f</td>\n",
       "      <td>NaN</td>\n",
       "      <td>7/26/13</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>712</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>Secondary</td>\n",
       "      <td>1995.0</td>\n",
       "      <td>f</td>\n",
       "      <td>0</td>\n",
       "      <td>12/21/12</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3610</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>Secondary</td>\n",
       "      <td>1993.0</td>\n",
       "      <td>m</td>\n",
       "      <td>0</td>\n",
       "      <td>10/9/12</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7452</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>10/1/12</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3266</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>12/21/12</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7045</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>Doctorate</td>\n",
       "      <td>1949.0</td>\n",
       "      <td>f</td>\n",
       "      <td>0</td>\n",
       "      <td>1/17/13</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6462</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>Master's</td>\n",
       "      <td>1976.0</td>\n",
       "      <td>m</td>\n",
       "      <td>0</td>\n",
       "      <td>10/6/12</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1193 rows × 14 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      registered  viewed  explored      LoE_DI     YoB gender grade  \\\n",
       "7076           1       0         0  Bachelor's  1981.0      m     0   \n",
       "4877           1       0         0  Bachelor's  1989.0      m     0   \n",
       "4749           1       0         0   Secondary  1991.0      m     0   \n",
       "783            1       0         0   Secondary  1991.0      f   NaN   \n",
       "712            1       0         0   Secondary  1995.0      f     0   \n",
       "...          ...     ...       ...         ...     ...    ...   ...   \n",
       "3610           1       0         0   Secondary  1993.0      m     0   \n",
       "7452           1       0         0         NaN     NaN    NaN     0   \n",
       "3266           1       0         0         NaN     NaN    NaN     0   \n",
       "7045           1       0         0   Doctorate  1949.0      f     0   \n",
       "6462           1       0         0    Master's  1976.0      m     0   \n",
       "\n",
       "     start_time_DI last_event_DI  nevents  ndays_act  nplay_video  nchapters  \\\n",
       "7076       8/27/12           NaN      NaN        NaN          NaN        NaN   \n",
       "4877      12/24/12           NaN      NaN        NaN          NaN        NaN   \n",
       "4749        8/7/12           NaN      NaN        NaN          NaN        NaN   \n",
       "783        7/26/13           NaN      NaN        NaN          NaN        NaN   \n",
       "712       12/21/12           NaN      NaN        NaN          NaN        NaN   \n",
       "...            ...           ...      ...        ...          ...        ...   \n",
       "3610       10/9/12           NaN      NaN        NaN          NaN        NaN   \n",
       "7452       10/1/12           NaN      NaN        NaN          NaN        NaN   \n",
       "3266      12/21/12           NaN      NaN        NaN          NaN        NaN   \n",
       "7045       1/17/13           NaN      NaN        NaN          NaN        NaN   \n",
       "6462       10/6/12           NaN      NaN        NaN          NaN        NaN   \n",
       "\n",
       "      nforum_posts  \n",
       "7076             0  \n",
       "4877             0  \n",
       "4749             0  \n",
       "783              0  \n",
       "712              0  \n",
       "...            ...  \n",
       "3610             0  \n",
       "7452             0  \n",
       "3266             0  \n",
       "7045             0  \n",
       "6462             0  \n",
       "\n",
       "[1193 rows x 14 columns]"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data[features].loc[(train_data['last_event_DI'].isna()) & (train_data['certified'].eq(0))]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "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>registered</th>\n",
       "      <th>viewed</th>\n",
       "      <th>explored</th>\n",
       "      <th>LoE_DI</th>\n",
       "      <th>YoB</th>\n",
       "      <th>gender</th>\n",
       "      <th>grade</th>\n",
       "      <th>start_time_DI</th>\n",
       "      <th>last_event_DI</th>\n",
       "      <th>nevents</th>\n",
       "      <th>ndays_act</th>\n",
       "      <th>nplay_video</th>\n",
       "      <th>nchapters</th>\n",
       "      <th>nforum_posts</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Empty DataFrame\n",
       "Columns: [registered, viewed, explored, LoE_DI, YoB, gender, grade, start_time_DI, last_event_DI, nevents, ndays_act, nplay_video, nchapters, nforum_posts]\n",
       "Index: []"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data[features].loc[(train_data['last_event_DI'].isna()) & ~(train_data['certified'].eq(0))]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here we can presume that since last event DI is blank/missing, the student didn't interact with the course at all, therefore the nevents, ndays_act, nplay_video and nchapters are 0, (since viewed, and explored are 0 as well). It also stated in the description of the nevents that it's the number of interactions with the course, recorded in the tracking logs; blank if no interactions beyond registration. We will feature engineer this to be 0s for those variables related to the course interactions. \n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "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>registered</th>\n",
       "      <th>viewed</th>\n",
       "      <th>explored</th>\n",
       "      <th>LoE_DI</th>\n",
       "      <th>YoB</th>\n",
       "      <th>gender</th>\n",
       "      <th>grade</th>\n",
       "      <th>start_time_DI</th>\n",
       "      <th>last_event_DI</th>\n",
       "      <th>nevents</th>\n",
       "      <th>ndays_act</th>\n",
       "      <th>nplay_video</th>\n",
       "      <th>nchapters</th>\n",
       "      <th>nforum_posts</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Empty DataFrame\n",
       "Columns: [registered, viewed, explored, LoE_DI, YoB, gender, grade, start_time_DI, last_event_DI, nevents, ndays_act, nplay_video, nchapters, nforum_posts]\n",
       "Index: []"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data[features].loc[(train_data['start_time_DI'].isna())]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Looking at histogram of features by whether they are certified"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "doesn't work for LoE_DI\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "doesn't work for gender\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEQCAYAAABfiGi4AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8vihELAAAACXBIWXMAAAsTAAALEwEAmpwYAAAW3ElEQVR4nO3dfbRddX3n8fdHfGy1CCHByINBS1RwFJ2UsdKpKFhRWhFXUVgum3HopDNFrbbjI2ups1zMsKY+1PFpJj5GFoJQoGJxAYKK7QwUA0UeDAISgpGUYKSKreOU8J0/9s7mcHNyc5Pcc/a5975fa92Vc377PHwIK/dzzt6/vX+pKiRJAnhU3wEkSZPDUpAkdSwFSVLHUpAkdSwFSVLn0X0H2BP77bdfLVu2rO8YkjSnXHfddT+uqsXDts3pUli2bBlr167tO4YkzSlJNuxom7uPJEkdS0GS1LEUJEkdS0GS1LEUJEkdS0GS1LEUJEkdS0GS1LEUJEmdOX1G83xy/IknsWnzlqHbli5ZxCUXnT/mRJIWIkthQmzavIXlK88Yuu22NaePOY2khcrdR5KkjqUgSepYCpKkjqUgSepYCpKkjqUgSepYCpKkjqUgSepYCpKkjqUgSepYCpKkjqUgSepYCpKkjldJHaPpLo+9fsMGlo85jyRNZSmM0XSXx779vaeMOY0kbW9ku4+SHJTkm0nWJbklyZ+04+9P8qMkN7Q/rxx4zruT3JHk+0lePqpskqThRvlN4UHgz6rq+iRPAq5L8vV220eq6oODD05yGHAycDjwVOCKJMurausIM0qSBozsm0JVbaqq69vbDwDrgAOmecoJwLlV9cuqWg/cARw5qnySpO2NZfZRkmXA84G/a4felOTGJJ9Lsk87dgDww4GnbWRIiSRZlWRtkrX33XffKGNL0oIz8lJI8kTgAuCtVfUz4FPAM4AjgE3Ah7Y9dMjTa7uBqtVVtaKqVixevHg0oSVpgRppKSR5DE0hnF1VFwJU1b1VtbWqHgI+zcO7iDYCBw08/UDgnlHmkyQ90ihnHwX4LLCuqj48ML504GEnAje3ty8GTk7yuCSHAIcC144qnyRpe6OcfXQU8AbgpiQ3tGPvAU5JcgTNrqG7gD8CqKpbkpwHfI9m5tJpzjySpPEaWSlU1d8y/DjB16Z5zhnA8LO7JEkj57WPJEkdS0GS1LEUJEkdS0GS1LEUJEkdS0GS1LEUJEkdS0GS1LEUJEkdS0GS1HGN5nns+BNPYtPmLUO3LV2yiEsuOn/MiSRNOkthHtu0eQvLVw6/lNRta04fcxpJc4G7jyRJHUtBktSxFCRJHUtBktSxFCRJHUtBktRxSuocN925COs3bGD5mPNImtsshTluunMRbn/vKWNOI2muc/eRJKljKUiSOpaCJKljKUiSOpaCJKnj7CNtx0tuSwuXpaDteMltaeGyFOaAO+/8AS846qVDt3mCmqTZNLJSSHIQ8EXgKcBDwOqq+miSfYEvA8uAu4DXVtX97XPeDZwKbAXeUlWXjSrfXLK14glqksZilAeaHwT+rKqeDbwQOC3JYcC7gCur6lDgyvY+7baTgcOB44BPJtlrhPkkSVOMrBSqalNVXd/efgBYBxwAnACsaR+2Bnh1e/sE4Nyq+mVVrQfuAI4cVT5J0vbGMiU1yTLg+cDfAftX1SZoigNY0j7sAOCHA0/b2I5Nfa1VSdYmWXvfffeNNLckLTQjL4UkTwQuAN5aVT+b7qFDxmq7garVVbWiqlYsXrx4tmJKkhjx7KMkj6EphLOr6sJ2+N4kS6tqU5KlwOZ2fCNw0MDTDwTuGWW+3eU8fknz1ShnHwX4LLCuqj48sOliYCVwZvvnVwbGv5Tkw8BTgUOBa0eVb084j1/SfDXKbwpHAW8AbkpyQzv2HpoyOC/JqcDdwEkAVXVLkvOA79HMXDqtqraOMJ8kaYqRlUJV/S3DjxMAHLOD55wBDP8Irokw3Yl07jqT5j7PaNYume5EOnedSXOfV0mVJHUsBUlSx1KQJHU8pjDLvKKppLnMUphlXtFU0lzm7iNJUsdSkCR1LAVJUsdSkCR1LAVJUsdSkCR1LAVJUsdSkCR1LAVJUsdSkCR1LAVJUsdSkCR1LAVJUsdSkCR1LAVJUsf1FBYoFwOSNIylsEC5GJCkYWa0+yjJUTMZkyTNbTM9pvCxGY5JkuawaXcfJflN4EXA4iR/OrDp14C9RhlMkjR+Ozum8Fjgie3jnjQw/jPg90cVSpLUj2lLoaquAq5K8oWq2jCmTJKknsz0mMLjkqxOcnmSb2z7me4JST6XZHOSmwfG3p/kR0luaH9eObDt3UnuSPL9JC/fzf8eSdIemOmU1POB/wl8Btg6w+d8Afg48MUp4x+pqg8ODiQ5DDgZOBx4KnBFkuVVNdP3mnXHn3gSmzZvGbrNefyS5quZlsKDVfWpXXnhqvp2kmUzfPgJwLlV9UtgfZI7gCOBq3flPWfTps1bnMcvacGZ6e6jryb54yRLk+y77Wc33/NNSW5sdy/t044dAPxw4DEb2zFJ0hjN9JvCyvbPtw+MFfD0XXy/TwEfaJ/7AeBDwL8HMuSxNewFkqwCVgEcfPDBu/j26tN0u+SWLlnEJRedP+ZEkqaaUSlU1SGz8WZVde+220k+Dfx1e3cjcNDAQw8E7tnBa6wGVgOsWLFiaHFoMk23S+62NaePOY2kYWZUCkn+YNh4VU09iLyz11laVZvauycC22YmXQx8KcmHaQ40HwpcuyuvLUnaczPdffQbA7cfDxwDXM/2M4s6Sc4Bjgb2S7IReB9wdJIjaHYN3QX8EUBV3ZLkPOB7wIPAaX3OPJKkhWqmu4/ePHg/yd7AWTt5zrApOp+d5vFnAMP3LUiSxmJ3F9n5Z5pdPJKkeWSmxxS+ysOzgfYCng2cN6pQkqR+zPSYwuAZyA8CG6pq4wjySJJ6NKPdR+2F8W6luVLqPsD/G2UoSVI/Zrr76LXAnwPfojnR7GNJ3l5VfznCbCPn9Y0k6ZFmuvvodOA3qmozQJLFwBXAnC4Fr28kSY8009lHj9pWCK0tu/BcSdIcMdNvCpcmuQw4p73/OuBro4kkSerLztZo/nVg/6p6e5LXAL9Fc0zhauDsMeSTJI3RznYB/QXwAEBVXVhVf1pVb6P5lvAXo40mSRq3nZXCsqq6cepgVa0Flo0kkSSpNzsrhcdPs+0JsxlEktS/nZXCd5L8h6mDSU4FrhtNJElSX3Y2++itwEVJXs/DJbACeCzNegiSpHlk2lJoV0p7UZKXAM9phy+pqm+MPJkkaexmup7CN4FvjjiLJKlnnpUsSepYCpKkjqUgSepYCpKkjqUgSepYCpKkjqUgSepYCpKkjqUgSepYCpKkzkyX45RG6s47f8ALjnrp0G1LlyzikovOH3MiaWGyFDQRtlZYvvKModtuW3P6mNNIC5eloFkz3ad9gPUbNrB8jHkk7bqRlUKSzwG/C2yuque0Y/sCX6ZZyvMu4LVVdX+77d3AqcBW4C1Vddmosmk0pvu0D3D7e08ZYxpJu2OUB5q/ABw3ZexdwJVVdShwZXufJIcBJwOHt8/5ZJK9RphNkjTEyEqhqr4N/GTK8AnAmvb2GuDVA+PnVtUvq2o9cAdw5KiySZKGG/eU1P2rahNA++eSdvwA4IcDj9vYjm0nyaoka5Osve+++0YaVpIWmkk5TyFDxmrYA6tqdVWtqKoVixcvHnEsSVpYxl0K9yZZCtD+ubkd3wgcNPC4A4F7xpxNkha8cZfCxcDK9vZK4CsD4ycneVySQ4BDgWvHnE2SFrxRTkk9Bzga2C/JRuB9wJnAeUlOBe4GTgKoqluSnAd8D3gQOK2qto4qmyRpuJGVQlXtaFL6MTt4/BnAjie5S5JGblIONEuSJoClIEnqWAqSpI6lIEnqWAqSpI6lIEnqWAqSpI6lIEnqWAqSpI6lIEnquEazJt50az8vXbKISy46f8yJpPnLUtDEm27t59vWnD7mNNL85u4jSVLHUpAkdSwFSVLHYwqa0zwILc0uS0Fzmgehpdnl7iNJUsdSkCR1LAVJUsdSkCR1LAVJUsdSkCR1LAVJUsdSkCR1LAVJUsdSkCR1LAVJUqeXax8luQt4ANgKPFhVK5LsC3wZWAbcBby2qu7vI58kLVR9flN4SVUdUVUr2vvvAq6sqkOBK9v7kqQxmqTdRycAa9rba4BX9xdFkhamvi6dXcDlSQr4X1W1Gti/qjYBVNWmJEuGPTHJKmAVwMEHHzyuvJqDXGtB2nV9lcJRVXVP+4v/60lunekT2wJZDbBixYoaVUDNfa61IO26XnYfVdU97Z+bgYuAI4F7kywFaP/c3Ec2SVrIxl4KSX41yZO23QZ+B7gZuBhY2T5sJfCVcWeTpIWuj91H+wMXJdn2/l+qqkuTfAc4L8mpwN3AST1kk6QFbeylUFV3As8bMr4FOGbceSRJD5ukKamSpJ5ZCpKkjqUgSepYCpKkjqUgSepYCpKkTl+XuZB65XWRpOEsBS1I010X6dL3nWxhaMGyFKQpvJCeFjKPKUiSOpaCJKljKUiSOh5TkHaBs5Y031kK0i7wILTmO3cfSZI6loIkqWMpSJI6loIkqWMpSJI6loIkqWMpSJI6loIkqePJa9IYHH/iSWzavGXoNs+E1iSxFKQx2LR5i2dCa06wFKSeeT0lTRJLQZol0/1yX79hA8t38Dyvp6RJYilIs2S6X+63v/eUMaeRdo+lIM1R0x28Bnc9afdYCtIcNd3Ba9j9XU/OlFrYJq4UkhwHfBTYC/hMVZ3ZcySpN7t7nGJPOFNqYZuoUkiyF/AJ4GXARuA7SS6uqu/1m0zqx54cpxjFrCZnSs1/E1UKwJHAHVV1J0CSc4ETAEtB2kWjmNU03Wte+r6Td1gYP9p4NwccePDQbdOVyXS7snb3Nfuwu7vk+tiVl6qa9RfdXUl+Hziuqv6wvf8G4N9U1ZsGHrMKWNXefSbw/d14q/2AH+9h3FGY1FwwudkmNRdMbrZJzQWTm21Sc8HuZXtaVS0etmHSvilkyNgjWquqVgOr9+hNkrVVtWJPXmMUJjUXTG62Sc0Fk5ttUnPB5Gab1Fww+9km7YJ4G4GDBu4fCNzTUxZJWnAmrRS+Axya5JAkjwVOBi7uOZMkLRgTtfuoqh5M8ibgMpopqZ+rqltG8FZ7tPtphCY1F0xutknNBZObbVJzweRmm9RcMMvZJupAsySpX5O2+0iS1CNLQZLUsRQkSZ2JOtA8CkmeRXNW9AE05zzcA1xcVet6DSZJE2hef1NI8k7gXJqT4q6lmfIa4Jwk7+o5295Jzkxya5It7c+6duzJ5po72SY1l7Q75vXsoyS3AYdX1b9MGX8scEtVHdpPMkhyGfANYE1V/UM79hRgJXBsVb3MXHMj26TmGsgXmuuKDX5bvrYm4B//pGab1Fww+mzzvRRuBV5eVRumjD8NuLyqntlPMkjy/R29/3TbRm1Sc+3s/f07Gy7J7wCfBG4HftQOHwj8OvDHVXW52eZGrnFlm+/HFN4KXJnkduCH7djBNH+Bb9rRk8ZkQ5J30Hy6vBcgyf7Av+PhrOZ6pEnNNqm5oFmb5NiqumtwMMkhwNeAZ/cRqjWp2SY1F4wh27w+plBVlwLLgf9Cc5b05cD7gWe22/r0OmARcFWS+5P8BPgWsC/w2gnKdX+ba1HPuWDu/J1NSi5oPvhtHDL+I+AxY84y1aRmm9RcMIZs8/2bAlX1EHBN3zmmqqr7k3we+DpwTVX9fNu2dvW5Xkqrqu4H3tn+bMtzVlW9o488UywH/mtVvTPJrwDvAl7QbtvaXyz+iWbNj69X1RVJXg+8CLgbeKDHXACfo1ms6lwe/tZyEM11xT7bW6rGpGYblutgmvKfxL+zWc02r48pTLIkbwFOA9YBRwB/UlVfabddX1UvmObpo8w17AKEL6U5kEpVvWq8iR6W5Bbgee01slbT/DK+ADimHX9NT7nOpvmA9QTgp8CvAhe1uVJVK/vItU2Sw4BX0RyYDM0nzYlY0XBSsyV5Ng9PZZ+YXDD6bJZCT5LcBPxmVf08yTLgL4GzquqjSf6+qp7fU67raT71foZmZkOAc2g+vVFVV/WRq822rqqevS3nYHEmuaGqjugp141V9dwkj6b5Gv/UqtrazhL5blU9t49cmj1JllTV5r5zDJNkUVUNX55tN8zrYwoTbq9tu4zag0ZHA69I8mGGLzY0LiuA64DTgZ9W1beAX1TVVX0WQuvmJG9sb383yQqAJMuBf9nx00buUe005ycBvwLs3Y4/jp73QU/yORRJfi3Jf0tyVpJTpmz7ZI+59p36A1ybZJ/2dm/a/2/7tbf/dZI7gWuSbEjy4tl4D0uhP/+Q5Ihtd9qC+F2apfX+VV+hquqhqvoI8Ebg9CQfZ3KOPf0h8OIkPwAOA65u/1F8ut3Wl88CtwI30JTp+Uk+TXOy5Lk95gI4D7gfOLqqFlXVIuAlwD8CfS9i/HmaD0AXAKckuSDJ49ptL+wvFj+m+WA0+HMAcD2wtsdcAMdX1balNz8IvK493+plwIdm4w3cfdSTJAcCD2472WnKtqOq6n/3EGs7SY4Hjqqq9/SdZZskTwKeTjsTY9s00D4leSpAVd3TfgI/Fri7qq7tOdckn0PxiF1+SU4HXklzjOHrPR5X+880///eXlU3tWPrq+qQPvIMSnPu1XPa42rXVNULB7bdVFV7/IHSUpDmsSSXA1cw/ByKl1XVsT1mW0dzxYGHBsZWAu8AnlhVT+sx24HAR2hm+LyP5tjQ0/vKs02SNwO/B5wJ/DbwZOBCmkkNT6+qN+zxe1gK0vyVZB+aqbsnAEva4Xtplrk9s52C3Fe2/05zZYErpowfB3ysz8vQDGT5PZpdgsuq6il95wFIcjTwn2imaD+aprj+imalygf3+PUtBWlhSvLGqvp83zmGmaRsSZ4APKOqbp6kXFPNVjZLQVqgktxdVQf3nWOYSc02qblg9rJNyqwSSSOQ5MYdbQL2H2eW7QJMaLZJzQXjyWYpSPPb/sDLaaalDgrwf8Yf5xEmNduk5oIxZLMUpPntr2lm8twwdUOSb409zSNNarZJzQVjyOYxBUlSxzOaJUkdS0GS1LEUJEkdS0HagSTLktzcd45BSY5O8qK+c2j+shSkueVomlXdpJGwFLQgtd8C1iX5dJJbklye5AntNeq/m+RqmpXxBh//N0mub39e1I6fleSEgcedneRVSQ5Pcm2SG5LcmGSH1/FJ8ldJrmtzrBoYP659r+8muTLNYkz/EXhb+7r/dhR/N1rYnJKqBan9BXsHsKKqbkhyHs1F4t4BvLmqrkry58Arquo5adaEfqiq/m/7C/6cqlrRLmzytqp6dZK9adZUOJTmCpvXVNXZ7QI8e1XVL3aQZd+q+kl7jZ3vAC+m+cB2PfDbVbV+4DHvB35eVR8c2V+OFjRPXtNCtn7gJKDrgGcATx5YYe4s4BXt7ccAH28XRtpKc4VK2vL4RJIlwGuAC9pr3V9Ns0jRgcCFVXX7NDnekuTE9vZBNKWyGPh2Va1v3+cne/6fK+2cu4+0kP1y4PZWYB+adamHeRvNJaefR7Nk6WMHtp0FvJ5mtbrPA1TVl2gWi/kFcFmSlw570fYyyMfSrNf9PODvgcfTXLbAr/EaO0tBetg/Aj9N8lvt/dcPbNsb2NQuCPMGYK+BbV8A3gpQVbcAJHk6cGdV/Q+a3VLP3cF77g3cX1X/nORZPLwM5dU0S48e0r7etrWBH6BZC1oaCUtBeqQ3Ap9od/8MHgP4JLAyyTU0u47+aduGdkWzdbTfElqvA25OcgPwLOCLO3i/S4FHt1e//ABwTfua9wGrgAuTfBf4cvv4rwIneqBZo+KBZmkPtQehbwJeUFU/7TuPtCf8piDtgSTHArfSLB9pIWjO85uCNAZJFgFXDtl0TFVtGXceaUcsBUlSx91HkqSOpSBJ6lgKkqSOpSBJ6vx/kcwoRsToSzAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "for feat in features:\n",
    "    try:\n",
    "        sns.histplot(data=train_data[features].loc[train_data.certified == 1], x=feat)\n",
    "        plt.xticks(rotation=90)\n",
    "        plt.show()\n",
    "    except:\n",
    "        print(\"doesn't work for {}\".format(feat))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1])"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data[features].loc[train_data.certified == 1, 'viewed'].unique()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Or not certified"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "doesn't work for LoE_DI\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "doesn't work for gender\n",
      "doesn't work for grade\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "doesn't work for last_event_DI\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "for feat in features:\n",
    "    try:\n",
    "        sns.histplot(data=train_data[features].loc[train_data.certified == 0], x=feat)\n",
    "        plt.xticks(rotation=90)\n",
    "        plt.show()\n",
    "    except:\n",
    "        print(\"doesn't work for {}\".format(feat))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature Engineering"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here we try some feature engineering techniques on our data. Some things to try is creating age variable from our year of birth data, creating an education level based on increasing education (less than secondary to doctorate), as well as dummy variables for gender and country where participant is from. \n",
    "\n",
    "We also try some base level imputation techniques like using the mode, mean, and/or median to fill in the NAs in the data. Special consideration was for a subset of our data where last_event_DI was blank, and it was decided to fill in \"nevents\", \"ndays_act\", \"nplay_video\", \"nchapters\" with 0s as we can presume that if they didn't access the course content (viewed, etc) they didn't have anything for any of those variables. \n",
    "\n",
    "Didn't get enough time to try out the KNN imputation technique but for future modeling improvement, might be improvement for the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "def feature_engineer(input_df : pd.DataFrame, features_to_update : list = None, nan_method : str = None,\n",
    "                    feat_method : str = None, n_neighbors : int = 5) -> pd.DataFrame:\n",
    "    ''' Parameters\n",
    "        ================\n",
    "        input_data - our input training data\n",
    "        features_to_update - our list of features to process nan/feat method on\n",
    "        nan_method - what to replace nan with, Allowed values: ['mode', 'mean', 'median', 'knn']\n",
    "        feat_method - what to process for data, Allowed values: ['impute', 'encode']\n",
    "        n_neighbors - if knn in nan_method specified, how many neighbors to use\n",
    "    '''\n",
    "    input_data = input_df.copy()\n",
    "    if features_to_update is None:\n",
    "        features_to_update = list(input_data.columns)\n",
    "        \n",
    "    for feat in features_to_update:\n",
    "        if feat_method == \"encode\":\n",
    "            if feat == \"YoB\":\n",
    "                # create age variable\n",
    "                input_data['age'] = (2020 - input_data[feat]) + 1\n",
    "            elif feat == \"LoE_DI\":\n",
    "                # create variable where education level is encoded\n",
    "                edu_lvls = {\"Master's\":4, \"Bachelor's\" :3, 'Secondary':2, 'Doctorate':5,\n",
    "                           'Less than Secondary':1}\n",
    "                input_data['edu_lvl'] = input_data[feat].map(edu_lvls)\n",
    "            elif feat in ['gender', 'final_cc_cname_DI']:\n",
    "                feat_encoded = pd.get_dummies(input_data[feat])\n",
    "                input_data = input_data.drop(feat,axis = 1)\n",
    "                input_data = input_data.join(feat_encoded)\n",
    "                \n",
    "        elif feat_method == \"impute\":\n",
    "            # fill in NAs\n",
    "            input_data[feat] = input_data[feat].replace('nan', np.nan)\n",
    "            \n",
    "            # replace blanks in grade with nan values\n",
    "            if feat == \"grade\":\n",
    "                input_data.loc[input_data[feat] == \" \", feat] = np.nan\n",
    "                input_data[feat] = input_data[feat].astype(float)\n",
    "            elif feat == \"last_event_DI\":\n",
    "                replace_cols = [\"nevents\", \"ndays_act\", \"nplay_video\", \"nchapters\"]\n",
    "                train_data.loc[(train_data['last_event_DI'].isna()), replace_cols] = 0\n",
    "                \n",
    "            # replace blankk/nas with set nan method\n",
    "            if nan_method == \"mean\" and is_numeric_dtype(input_data[feat]):\n",
    "                nan_method_val = input_data[feat].mean()\n",
    "                input_data.loc[input_data[feat].isna(), feat] = nan_method_val.values[0]\n",
    "            elif nan_method == \"mode\":\n",
    "                nan_method_val = input_data[feat].mode()\n",
    "                input_data.loc[input_data[feat].isna(), feat] = nan_method_val.values[0]\n",
    "            elif nan_method == \"median\" and is_numeric_dtype(input_data[feat]):\n",
    "                nan_method_val = input_data[feat].median()\n",
    "                input_data.loc[input_data[feat].isna(), feat] = nan_method_val.values[0]\n",
    "            elif nan_method == \"knn\" and is_numeric_dtype(input_data[feat]):\n",
    "                scaler = MinMaxScaler()\n",
    "                input_data = pd.DataFrame(scaler.fit_transform(input_data), columns = input_data.columns)\n",
    "                imputer = KNNImputer(n_neighbors=n_neighbors)\n",
    "                input_data = pd.DataFrame(imputer.fit_transform(input_data),columns = input_data.columns)\n",
    "    \n",
    "    return(input_data)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_data = train_data[features + [target]]\n",
    "# let's impute and encode our training data\n",
    "train_stage1 = feature_engineer(input_df = train_data, \n",
    "                               features_to_update = ['gender', 'LoE_DI', 'YoB', 'grade', 'nevents', \n",
    "                                                     'ndays_act', 'nplay_video', 'nchapters', 'last_event_DI'],\n",
    "                        nan_method = \"mode\", feat_method = \"impute\")\n",
    "\n",
    "train_stage2 = feature_engineer(input_df = train_stage1, \n",
    "                        features_to_update = ['gender', 'LoE_DI', 'YoB', 'grade'],\n",
    "                        feat_method = \"encode\")\n",
    "\n",
    "# do the same for our validation data\n",
    "valid_stage1 = feature_engineer(input_df = validation_data, \n",
    "                               features_to_update = ['gender', 'LoE_DI', 'YoB', 'grade', 'nevents', \n",
    "                                                     'ndays_act', 'nplay_video', 'nchapters', 'last_event_DI'],\n",
    "                        nan_method = \"mode\", feat_method = \"impute\")\n",
    "\n",
    "valid_stage2 = feature_engineer(input_df = valid_stage1, \n",
    "                        features_to_update = ['gender', 'LoE_DI', 'YoB', 'grade'],\n",
    "                        feat_method = \"encode\")\n",
    "\n",
    "\n",
    "# do the same for our test data\n",
    "test_data_tran = feature_engineer(input_df =  test_data,\n",
    "                                  features_to_update = ['gender', 'LoE_DI', 'YoB', 'grade', 'nevents', \n",
    "                                                     'ndays_act', 'nplay_video', 'nchapters', 'last_event_DI'],\n",
    "                                  nan_method = \"mode\", feat_method = \"impute\")\n",
    "\n",
    "test_data_tran = feature_engineer(input_df = test_data_tran,\n",
    "                                  features_to_update = ['gender', 'LoE_DI', 'YoB', 'grade'],\n",
    "                                  feat_method = \"encode\")\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "registered       0.0\n",
      "viewed           0.0\n",
      "explored         0.0\n",
      "LoE_DI           0.0\n",
      "YoB              0.0\n",
      "grade            0.0\n",
      "start_time_DI    0.0\n",
      "last_event_DI    0.0\n",
      "nevents          0.0\n",
      "ndays_act        0.0\n",
      "nplay_video      0.0\n",
      "nchapters        0.0\n",
      "nforum_posts     0.0\n",
      "certified        0.0\n",
      "f                0.0\n",
      "m                0.0\n",
      "edu_lvl          0.0\n",
      "age              0.0\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "# how many NAs are in our training data\n",
    "print(train_stage2.apply(lambda col: col.isna().sum()*100/len(train_stage2)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "course_id            0.0\n",
      "userid_DI            0.0\n",
      "registered           0.0\n",
      "viewed               0.0\n",
      "explored             0.0\n",
      "final_cc_cname_DI    0.0\n",
      "LoE_DI               0.0\n",
      "YoB                  0.0\n",
      "grade                0.0\n",
      "start_time_DI        0.0\n",
      "last_event_DI        0.0\n",
      "nevents              0.0\n",
      "ndays_act            0.0\n",
      "nplay_video          0.0\n",
      "nchapters            0.0\n",
      "nforum_posts         0.0\n",
      "f                    0.0\n",
      "m                    0.0\n",
      "edu_lvl              0.0\n",
      "age                  0.0\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "# how many NAs are left in our testing data\n",
    "print(test_data_tran.apply(lambda col: col.isna().sum()*100/len(test_data_tran)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "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>grade</th>\n",
       "      <th>nevents</th>\n",
       "      <th>ndays_act</th>\n",
       "      <th>nplay_video</th>\n",
       "      <th>nchapters</th>\n",
       "      <th>certified</th>\n",
       "      <th>YoB</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>grade</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.503464</td>\n",
       "      <td>0.741085</td>\n",
       "      <td>0.483352</td>\n",
       "      <td>0.699449</td>\n",
       "      <td>0.976876</td>\n",
       "      <td>-0.019294</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>nevents</th>\n",
       "      <td>0.503464</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.711719</td>\n",
       "      <td>0.852091</td>\n",
       "      <td>0.371307</td>\n",
       "      <td>0.502443</td>\n",
       "      <td>-0.104133</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ndays_act</th>\n",
       "      <td>0.741085</td>\n",
       "      <td>0.711719</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.669505</td>\n",
       "      <td>0.592679</td>\n",
       "      <td>0.729028</td>\n",
       "      <td>-0.126029</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>nplay_video</th>\n",
       "      <td>0.483352</td>\n",
       "      <td>0.852091</td>\n",
       "      <td>0.669505</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.539414</td>\n",
       "      <td>0.458148</td>\n",
       "      <td>-0.096089</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>nchapters</th>\n",
       "      <td>0.699449</td>\n",
       "      <td>0.371307</td>\n",
       "      <td>0.592679</td>\n",
       "      <td>0.539414</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.756906</td>\n",
       "      <td>-0.075893</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>certified</th>\n",
       "      <td>0.976876</td>\n",
       "      <td>0.502443</td>\n",
       "      <td>0.729028</td>\n",
       "      <td>0.458148</td>\n",
       "      <td>0.756906</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.019057</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>YoB</th>\n",
       "      <td>-0.019294</td>\n",
       "      <td>-0.104133</td>\n",
       "      <td>-0.126029</td>\n",
       "      <td>-0.096089</td>\n",
       "      <td>-0.075893</td>\n",
       "      <td>-0.019057</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                grade   nevents  ndays_act  nplay_video  nchapters  certified  \\\n",
       "grade        1.000000  0.503464   0.741085     0.483352   0.699449   0.976876   \n",
       "nevents      0.503464  1.000000   0.711719     0.852091   0.371307   0.502443   \n",
       "ndays_act    0.741085  0.711719   1.000000     0.669505   0.592679   0.729028   \n",
       "nplay_video  0.483352  0.852091   0.669505     1.000000   0.539414   0.458148   \n",
       "nchapters    0.699449  0.371307   0.592679     0.539414   1.000000   0.756906   \n",
       "certified    0.976876  0.502443   0.729028     0.458148   0.756906   1.000000   \n",
       "YoB         -0.019294 -0.104133  -0.126029    -0.096089  -0.075893  -0.019057   \n",
       "\n",
       "                  YoB  \n",
       "grade       -0.019294  \n",
       "nevents     -0.104133  \n",
       "ndays_act   -0.126029  \n",
       "nplay_video -0.096089  \n",
       "nchapters   -0.075893  \n",
       "certified   -0.019057  \n",
       "YoB          1.000000  "
      ]
     },
     "execution_count": 68,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# update grade here in train_data to see corr before and after imputation for grade\n",
    "# looking at the correlation of the updated features in our original training data\n",
    "\n",
    "train_data.loc[train_data[\"grade\"] == \" \", \"grade\"] = np.nan\n",
    "train_data[\"grade\"] = train_data[\"grade\"].astype(float)\n",
    "cor_mat = train_data[['grade', 'nevents', \n",
    "                      'ndays_act', 'nplay_video', 'nchapters', 'certified','LoE_DI', 'YoB']].corr()\n",
    "cor_mat"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "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>grade</th>\n",
       "      <th>nevents</th>\n",
       "      <th>ndays_act</th>\n",
       "      <th>nplay_video</th>\n",
       "      <th>nchapters</th>\n",
       "      <th>certified</th>\n",
       "      <th>f</th>\n",
       "      <th>m</th>\n",
       "      <th>edu_lvl</th>\n",
       "      <th>age</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>grade</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.513379</td>\n",
       "      <td>0.746225</td>\n",
       "      <td>0.281298</td>\n",
       "      <td>0.518284</td>\n",
       "      <td>0.977762</td>\n",
       "      <td>0.016508</td>\n",
       "      <td>-0.016508</td>\n",
       "      <td>0.100343</td>\n",
       "      <td>0.017345</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>nevents</th>\n",
       "      <td>0.513379</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.713385</td>\n",
       "      <td>0.832869</td>\n",
       "      <td>0.284028</td>\n",
       "      <td>0.505109</td>\n",
       "      <td>0.132697</td>\n",
       "      <td>-0.132697</td>\n",
       "      <td>0.239268</td>\n",
       "      <td>0.101976</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ndays_act</th>\n",
       "      <td>0.746225</td>\n",
       "      <td>0.713385</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.463999</td>\n",
       "      <td>0.446094</td>\n",
       "      <td>0.728639</td>\n",
       "      <td>0.017794</td>\n",
       "      <td>-0.017794</td>\n",
       "      <td>0.166075</td>\n",
       "      <td>0.113770</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>nplay_video</th>\n",
       "      <td>0.281298</td>\n",
       "      <td>0.832869</td>\n",
       "      <td>0.463999</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.059245</td>\n",
       "      <td>0.258762</td>\n",
       "      <td>0.088414</td>\n",
       "      <td>-0.088414</td>\n",
       "      <td>0.171522</td>\n",
       "      <td>0.064334</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>nchapters</th>\n",
       "      <td>0.518284</td>\n",
       "      <td>0.284028</td>\n",
       "      <td>0.446094</td>\n",
       "      <td>0.059245</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.572220</td>\n",
       "      <td>0.022212</td>\n",
       "      <td>-0.022212</td>\n",
       "      <td>0.054735</td>\n",
       "      <td>0.055748</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>certified</th>\n",
       "      <td>0.977762</td>\n",
       "      <td>0.505109</td>\n",
       "      <td>0.728639</td>\n",
       "      <td>0.258762</td>\n",
       "      <td>0.572220</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.028872</td>\n",
       "      <td>-0.028872</td>\n",
       "      <td>0.095819</td>\n",
       "      <td>0.017509</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>f</th>\n",
       "      <td>0.016508</td>\n",
       "      <td>0.132697</td>\n",
       "      <td>0.017794</td>\n",
       "      <td>0.088414</td>\n",
       "      <td>0.022212</td>\n",
       "      <td>0.028872</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-1.000000</td>\n",
       "      <td>0.092995</td>\n",
       "      <td>0.045318</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>m</th>\n",
       "      <td>-0.016508</td>\n",
       "      <td>-0.132697</td>\n",
       "      <td>-0.017794</td>\n",
       "      <td>-0.088414</td>\n",
       "      <td>-0.022212</td>\n",
       "      <td>-0.028872</td>\n",
       "      <td>-1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.092995</td>\n",
       "      <td>-0.045318</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>edu_lvl</th>\n",
       "      <td>0.100343</td>\n",
       "      <td>0.239268</td>\n",
       "      <td>0.166075</td>\n",
       "      <td>0.171522</td>\n",
       "      <td>0.054735</td>\n",
       "      <td>0.095819</td>\n",
       "      <td>0.092995</td>\n",
       "      <td>-0.092995</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.517775</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>age</th>\n",
       "      <td>0.017345</td>\n",
       "      <td>0.101976</td>\n",
       "      <td>0.113770</td>\n",
       "      <td>0.064334</td>\n",
       "      <td>0.055748</td>\n",
       "      <td>0.017509</td>\n",
       "      <td>0.045318</td>\n",
       "      <td>-0.045318</td>\n",
       "      <td>0.517775</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                grade   nevents  ndays_act  nplay_video  nchapters  certified  \\\n",
       "grade        1.000000  0.513379   0.746225     0.281298   0.518284   0.977762   \n",
       "nevents      0.513379  1.000000   0.713385     0.832869   0.284028   0.505109   \n",
       "ndays_act    0.746225  0.713385   1.000000     0.463999   0.446094   0.728639   \n",
       "nplay_video  0.281298  0.832869   0.463999     1.000000   0.059245   0.258762   \n",
       "nchapters    0.518284  0.284028   0.446094     0.059245   1.000000   0.572220   \n",
       "certified    0.977762  0.505109   0.728639     0.258762   0.572220   1.000000   \n",
       "f            0.016508  0.132697   0.017794     0.088414   0.022212   0.028872   \n",
       "m           -0.016508 -0.132697  -0.017794    -0.088414  -0.022212  -0.028872   \n",
       "edu_lvl      0.100343  0.239268   0.166075     0.171522   0.054735   0.095819   \n",
       "age          0.017345  0.101976   0.113770     0.064334   0.055748   0.017509   \n",
       "\n",
       "                    f         m   edu_lvl       age  \n",
       "grade        0.016508 -0.016508  0.100343  0.017345  \n",
       "nevents      0.132697 -0.132697  0.239268  0.101976  \n",
       "ndays_act    0.017794 -0.017794  0.166075  0.113770  \n",
       "nplay_video  0.088414 -0.088414  0.171522  0.064334  \n",
       "nchapters    0.022212 -0.022212  0.054735  0.055748  \n",
       "certified    0.028872 -0.028872  0.095819  0.017509  \n",
       "f            1.000000 -1.000000  0.092995  0.045318  \n",
       "m           -1.000000  1.000000 -0.092995 -0.045318  \n",
       "edu_lvl      0.092995 -0.092995  1.000000  0.517775  \n",
       "age          0.045318 -0.045318  0.517775  1.000000  "
      ]
     },
     "execution_count": 69,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# looking at the correlation of the updated features in our training data\n",
    "cor_mat = train_stage2[['LoE_DI', 'grade', 'nevents', \n",
    "                      'ndays_act', 'nplay_video', 'nchapters', 'certified', 'f', 'm', 'edu_lvl', 'age']].corr()\n",
    "cor_mat"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As a result of imputation of NAs using mode method, it looks like the correlations with certified (our target variable) haven't really improved much and the correlation with nplay_video decreased. nchapters seemed to have the greatest improvement after imputation. Adding age didn't seemed to add much explantary value to the data. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Feature Selection"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here we use a base level feature selection method called recursive feature elimination.\n",
    "\n",
    "It's given an external estimator that assigns weights to features, and the goal of recursive feature elimination (RFE) is to select features by recursively considering smaller and smaller sets of features. Here we use Random Forest as our external estimator to assign weights to features. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [],
   "source": [
    "# here are the best features we have decided to use\n",
    "best_features = [\"viewed\", \n",
    "                 \"explored\",\n",
    "                 \"nevents\",\n",
    "                 #\"nplay_video\",\n",
    "                 \"grade\",\n",
    "                 #\"ndays_act\",\n",
    "                 \"nchapters\",\n",
    "                 \"nforum_posts\",\n",
    "                 #\"age\",\n",
    "                 #\"f\",\n",
    "                 #\"m\",\n",
    "                 #'edu_lvl'\n",
    "                ]\n",
    "\n",
    "target = 'certified'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Reasoning for the features selected: \n",
    "\n",
    "#### viewed: Generally we saw a correlation between someone having viewed the page\n",
    "#### explored: Generally saw a trend with more exploration as the grade goes up and with certified vs non-certified people\n",
    "#### nevents: Generally saw a trend with more events as the grade goes up and with certified vs non-certified people\n",
    "#### nplay_video: We removed this because we believe that there were too many missing data points to keep this (> 70% of the training data missing)\n",
    "#### grade: Grade of someone in the class has the strongest correlation/explanation for whether someone is certified as seen by the histograms and the correlation matrix\n",
    "#### ndays_act: Generally saw a trend with more days using the site as the grade goes up and with certified vs non-certified people\n",
    "#### nchapters: Generally saw a trend with more chapters viewed as the grade goes up and with certified vs non-certified people\n",
    "#### nforum_posts: Generally saw a trend with more forum posts as the grade goes up and with certified vs non-certified people\n",
    "\n",
    "We removed some of the more repetitive variables like ndays act. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn import decomposition\n",
    "\n",
    "num_cols_new = train_stage2.select_dtypes(include=np.number).columns.tolist()\n",
    "pca = decomposition.PCA(n_components=len(num_cols_new))\n",
    "pca.fit(train_stage2[num_cols_new].to_numpy())\n",
    "Z = pca.transform(train_stage2[num_cols_new].to_numpy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[9.84351404e-01 1.55898674e-02 2.72943919e-05 2.34629228e-05\n",
      " 7.76122477e-06 7.57080960e-08 6.14235257e-08 4.26010021e-08\n",
      " 1.36977327e-08 1.11538384e-08 5.22946500e-09 6.04396841e-10\n",
      " 5.93942093e-35 3.48608726e-39 0.00000000e+00]\n"
     ]
    }
   ],
   "source": [
    "print(pca.explained_variance_ratio_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Conclusion from above: majority of the variation in our data can be explained by just two variables (98.4% from just one (looks like it would be the grade variable))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[0.000e+00 8.000e+00 0.000e+00]\n",
      " [1.000e+00 5.255e+03 9.500e-01]\n",
      " [0.000e+00 2.780e+02 3.100e-01]\n",
      " ...\n",
      " [1.000e+00 3.934e+03 9.400e-01]\n",
      " [0.000e+00 1.520e+02 0.000e+00]\n",
      " [1.000e+00 4.190e+03 7.300e-01]] [[1.000e+00 7.510e+02 8.900e-01]\n",
      " [1.000e+00 1.214e+03 6.000e-01]\n",
      " [0.000e+00 7.450e+02 3.000e-02]\n",
      " ...\n",
      " [1.000e+00 2.028e+03 8.200e-01]\n",
      " [0.000e+00 2.390e+02 1.000e-02]\n",
      " [0.000e+00 6.100e+01 0.000e+00]] [[1.000e+00 1.320e+02 1.000e+00]\n",
      " [0.000e+00 1.000e+00 0.000e+00]\n",
      " [1.000e+00 2.650e+03 8.100e-01]\n",
      " ...\n",
      " [0.000e+00 1.198e+03 2.000e-02]\n",
      " [0.000e+00 1.900e+02 7.300e-01]\n",
      " [0.000e+00 3.400e+01 0.000e+00]]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.feature_selection import RFE\n",
    "\n",
    "def feature_selection(input_data : pd.DataFrame, n_feat : int, n_estimators : int,\n",
    "                     features : list, target : str, test_data : pd.DataFrame = None, use_test : bool = False):\n",
    "    ''' Parameters:\n",
    "        =====================\n",
    "        train_data - input training data\n",
    "        n_feat - number of features to select as best\n",
    "        n_estimators - using random forest as our model to select features from\n",
    "        features - our features to select from\n",
    "        target - our target variable\n",
    "        test_data - our test data to transform\n",
    "        use_test - whether to use our test data or not\n",
    "        \n",
    "        Returns: predictor data with features selected\n",
    "    '''\n",
    "    \n",
    "    #use the random forest as the model\n",
    "\n",
    "    model = RandomForestClassifier(random_state=6, n_estimators=n_estimators)\n",
    "    rfe_tree = RFE(estimator=model, n_features_to_select=n_feat, step=1)\n",
    "    X_rfe_tree = rfe_tree.fit_transform(input_data[features].to_numpy(), \n",
    "                                                  input_data[target].to_numpy())\n",
    "    if use_test:\n",
    "        Y_test_rfe_tree = rfe_tree.transform(test_data[features].to_numpy())\n",
    "        return({\"input_data\":X_rfe_tree, \"test_data\":Y_test_rfe_tree})\n",
    "    else:\n",
    "        return(X_rfe_tree)\n",
    "\n",
    "\n",
    "# use for later model training/predicting\n",
    "X_train_w_select = feature_selection(input_data = train_stage2, n_feat = 3, n_estimators = 50,\n",
    "                            features = best_features, target = \"certified\")\n",
    "\n",
    "X_valid_w_select = feature_selection(input_data = valid_stage2, n_feat = 3, n_estimators = 50,\n",
    "                            features = best_features, target = \"certified\")\n",
    "    \n",
    "X_test_w_select = feature_selection(input_data = train_stage2, n_feat = 3, n_estimators = 50,\n",
    "                                    features = best_features, target = \"certified\", test_data = test_data_tran,\n",
    "                                    use_test = True)['test_data']\n",
    "\n",
    "print(X_train_w_select, X_valid_w_select, X_test_w_select)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training models"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Parameter Selection by Model\n",
    "Here we use GridSearch to find the best params by model using a set amount of parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": [
    "# search best params\n",
    "\n",
    "\n",
    "def find_best_model_params(model : object, train_feat : pd.DataFrame,\n",
    "                           train_target : pd.DataFrame, hyperparams : dict,\n",
    "                           cv : int) -> dict:\n",
    "    ''' Parameters:\n",
    "        ================\n",
    "        model : selected model\n",
    "        train_feat: features for use\n",
    "        train_target: our target var\n",
    "        hyper_params : dictionary of the parameters with possible values by model\n",
    "        cv : parameter of Grid Search\n",
    "        \n",
    "        Returns: model parameters\n",
    "    '''\n",
    "    if isinstance(train_feat, pd.DataFrame):\n",
    "        train_df = train_feat.to_numpy()\n",
    "    else:\n",
    "        train_df = train_feat.copy()\n",
    "    search = GridSearchCV(estimator = model,\n",
    "                          param_grid=hyperparams,\n",
    "                          return_train_score=True,\n",
    "                          cv = cv).fit(X = train_df, \n",
    "                                      y = train_target.to_numpy())\n",
    "    return(search.best_params_)\n",
    "\n",
    "\n",
    "decision_params = {\"min_samples_leaf\": [1, 10, 50, 100, 200, 300], \n",
    "                   \"max_depth\":[1, 5, 10, 15, 20]}\n",
    "\n",
    "rand_params = {\"min_samples_leaf\": [1, 10, 50, 100, 200, 300], \n",
    "                   \"max_depth\":[1, 5, 10, 15, 20],\n",
    "                   \"n_estimators\":[10, 20, 50, 100]}\n",
    "\n",
    "ada_params = {\"base_estimator\": [DecisionTreeClassifier()], \n",
    "                   \"n_estimators\":[10, 20, 50, 100],\n",
    "                   \"learning_rate\":[1, 2, 3, 4, 5]}\n",
    "\n",
    "knn_params = {\"n_neighbors\":[1,2,3,4,5,6,7,8,9,10]}\n",
    "\n",
    "\n",
    "\n",
    "# without feature selection\n",
    "best_params_decision = find_best_model_params(model = DecisionTreeClassifier(), \n",
    "                                              train_feat=train_stage2[best_features], \n",
    "                                     train_target = train_stage2['certified'], \n",
    "                                     hyperparams = decision_params, cv = 6)\n",
    "\n",
    "best_params_rand = find_best_model_params(model = RandomForestClassifier(), \n",
    "                                          train_feat=train_stage2[best_features], \n",
    "                                     train_target = train_stage2['certified'], \n",
    "                                     hyperparams = rand_params, cv = 6)\n",
    "\n",
    "best_params_ada = find_best_model_params(model = AdaBoostClassifier(), \n",
    "                                         train_feat=train_stage2[best_features], \n",
    "                                     train_target = train_stage2['certified'], \n",
    "                                     hyperparams = ada_params, cv = 6)\n",
    "\n",
    "best_params_knn = find_best_model_params(model = KNeighborsClassifier(), \n",
    "                                         train_feat=train_stage2[best_features], \n",
    "                                     train_target = train_stage2['certified'], \n",
    "                                     hyperparams = knn_params, cv = 6)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [],
   "source": [
    "# for the with feature selection\n",
    "best_params_decision_w_select = find_best_model_params(model = DecisionTreeClassifier(), \n",
    "                                              train_feat=X_train_w_select, \n",
    "                                     train_target = train_stage2['certified'], \n",
    "                                     hyperparams = decision_params, cv = 6)\n",
    "\n",
    "best_params_rand_w_select = find_best_model_params(model = RandomForestClassifier(), \n",
    "                                          train_feat=X_train_w_select, \n",
    "                                     train_target = train_stage2['certified'], \n",
    "                                     hyperparams = rand_params, cv = 6)\n",
    "\n",
    "best_params_ada_w_select = find_best_model_params(model = AdaBoostClassifier(), \n",
    "                                         train_feat=X_train_w_select, \n",
    "                                     train_target = train_stage2['certified'], \n",
    "                                     hyperparams = ada_params, cv = 6)\n",
    "\n",
    "best_params_knn_w_select = find_best_model_params(model = KNeighborsClassifier(), \n",
    "                                         train_feat=X_train_w_select, \n",
    "                                     train_target = train_stage2['certified'], \n",
    "                                     hyperparams = knn_params, cv = 6)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Model Evaluation and Results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train_and_evaluate_model(model : object,\n",
    "                             model_type : str,\n",
    "                             X_train : pd.DataFrame, y_train : pd.DataFrame, \n",
    "                             X_valid : pd.DataFrame, y_valid : pd.DataFrame,\n",
    "                             model_params : dict,\n",
    "                             run_test : bool = False, \n",
    "                             X_test : pd.DataFrame = None, \n",
    "                             y_test : pd.DataFrame = None,\n",
    "                             return_pred : bool = False) -> pd.DataFrame:\n",
    "    ''' Parameters:\n",
    "        =============\n",
    "        model_type - Allowed values: ['decision', \"randomforest\",'adaboost']\n",
    "        model_params - input parameters for specific model of your choice\n",
    "        X_train - training input data with selected features\n",
    "        y_train - training data with target variable\n",
    "        X_valid - validation input data with selected features\n",
    "        y_valid - validation data with target variable\n",
    "        run_test - whether to run model prediction for testing data, Allowed Values: [True, False]\n",
    "        X_test - testing input data with selected features, Optional if run_test is False\n",
    "        y_test - testing data with target variable, Optional if run_test is False\n",
    "        \n",
    "        Returns: root mean square error of the predicted target values with the \n",
    "                actual target values (depends on rmse_type)\n",
    "    '''\n",
    "    assert model_type in ['decision', \"randomforest\",'adaboost', 'knn'], \\\n",
    "        \"model_type can only be a value in ['decision', 'randomforest','adaboost', 'knn']\"\n",
    "\n",
    "    assert len(model_params) > 0, \\\n",
    "        \"Empty input parameter list\"\n",
    "    \n",
    "    try:\n",
    "        assert not any(x.empty for x in [X_train, y_train, X_valid, y_valid]),\\\n",
    "            \"Empty input data out of X_train, y_train, X_valid, and/or y_valid\"\n",
    "    except:\n",
    "        pass\n",
    "\n",
    "    model_data = pd.DataFrame()\n",
    "    \n",
    "    model_eval_dict = {'model_type':[], \n",
    "                       'model':[], 'train_score':[], \n",
    "                       'validation_score':[]}\n",
    "\n",
    "    if run_test:\n",
    "        try:\n",
    "            assert not any(x.empty for x in [X_test, y_test]),\\\n",
    "                \"Empty input data out of X_test, y_test\"\n",
    "        except:\n",
    "            pass\n",
    "\n",
    "    # helper function for validating classification accuracy\n",
    "    def _calculate_accuracy(model, model_dict : dict, valid_type : str, \n",
    "                        X_eval : pd.DataFrame, y_eval : pd.DataFrame) -> dict:\n",
    "        ''' Parameters:\n",
    "        ===============\n",
    "        model - input model, can be of sklearn object type Lasso or Ridge\n",
    "        model_dict - input dict of model param and evaluations\n",
    "        valid_type - allows different rmse to be calculated, Allowed values: ['test', 'train', 'validation]\n",
    "        X_eval - input data with selected features to predict with\n",
    "        y_eval - data with target variable to evaluate predicted values against\n",
    "        '''\n",
    "        assert valid_type in ['test', 'train', 'validation'], \\\n",
    "        \"valid_type can only be a value in ['test', 'train', 'validation']\"\n",
    "        \n",
    "        if isinstance(X_eval, pd.DataFrame):\n",
    "            y_pred = model.predict(X_eval.to_numpy())\n",
    "            y_pred_prob = model.predict_proba(X_eval.to_numpy())\n",
    "        else:\n",
    "            y_pred = model.predict(X_eval)\n",
    "            y_pred_prob = model.predict_proba(X_eval)\n",
    "            \n",
    "        if valid_type == \"test\":\n",
    "            model_dict['{}_score'.format(valid_type)] = \"TBD\"\n",
    "        else:\n",
    "            accuracy = accuracy_score(y_true = y_eval.to_numpy(), y_pred = y_pred)\n",
    "            model_dict['{}_score'.format(valid_type)].append(accuracy)\n",
    "        return({\"score\":model_dict, \"predictions\":y_pred, 'predict_prob':y_pred_prob})\n",
    "    \n",
    "\n",
    "    # define model from sklern by specified model_type\n",
    "    if isinstance(X_train, pd.DataFrame):\n",
    "        fit_model = model(**model_params).fit(X_train.to_numpy(), y_train.to_numpy())\n",
    "    else:\n",
    "        fit_model = model(**model_params).fit(X_train, y_train.to_numpy())\n",
    "\n",
    "    # save model, <model_type>, and train, validation and optional test accurace scores to dict\n",
    "    model_eval_dict['model'].append(fit_model)\n",
    "    model_eval_dict['model_type'] = model_type\n",
    "    train_pred = _calculate_accuracy(fit_model, model_dict = model_eval_dict, \n",
    "                                 valid_type = \"train\", X_eval = X_train, y_eval = y_train)\n",
    "    model_eval_dict = train_pred['score']\n",
    "    \n",
    "    # save validation accuracies\n",
    "    valid_pred = _calculate_accuracy(fit_model, model_dict = model_eval_dict, \n",
    "                                 valid_type = \"validation\", X_eval = X_valid, y_eval = y_valid)\n",
    "    model_eval_dict = valid_pred['score']\n",
    "    \n",
    "    if run_test: # only calculate test rmse if specified\n",
    "        test_pred = _calculate_accuracy(fit_model, model_dict = model_eval_dict, \n",
    "                                 valid_type = \"test\", X_eval = X_test, y_eval = y_test)\n",
    "        model_eval_dict = test_pred['score']\n",
    "        model_eval_dict['pred'] = test_pred['predictions']\n",
    "    \n",
    "    # returns all predictions if selected\n",
    "    if return_pred:\n",
    "        return_dict = {\"validate_pred\":valid_pred['predictions'],\n",
    "                       \"train_pred\":train_pred['predictions'],\n",
    "                       \"validate_pred_prob\":valid_pred['predict_prob'],\n",
    "                       \"train_pred_prob\":train_pred['predict_prob'],\n",
    "                       \"model_eval_dict\":model_eval_dict}\n",
    "        if run_test:\n",
    "            return_dict[\"test_pred\"] = test_pred['predictions']\n",
    "            return_dict[\"test_pred_prob\"] = test_pred['predict_prob']\n",
    "        return(return_dict)\n",
    "    else:\n",
    "        return(pd.DataFrame(model_eval_dict))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'model_type': 'decision',\n",
       " 'model': [DecisionTreeClassifier(max_depth=1)],\n",
       " 'train_score': [0.9972880388238653],\n",
       " 'validation_score': [0.9988584474885844]}"
      ]
     },
     "execution_count": 77,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# our selected decision tree model (no feature selection)\n",
    "tree = train_and_evaluate_model(model = DecisionTreeClassifier,\n",
    "                                 model_type = \"decision\",\n",
    "                                 model_params = best_params_decision, \n",
    "                                 X_train = train_stage2[best_features], y_train = train_stage2['certified'], \n",
    "                                 X_valid = valid_stage2[best_features], y_valid = valid_stage2['certified'],\n",
    "                                 run_test = False,\n",
    "                                 return_pred = True)\n",
    "tree['model_eval_dict']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'model_type': 'randomforest',\n",
       " 'model': [RandomForestClassifier(max_depth=15, n_estimators=50)],\n",
       " 'train_score': [1.0],\n",
       " 'validation_score': [0.9988584474885844]}"
      ]
     },
     "execution_count": 78,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# our random forest model  (no feature selection)\n",
    "rand = train_and_evaluate_model(model = RandomForestClassifier,\n",
    "                         model_type = \"randomforest\",\n",
    "                         model_params = best_params_rand, \n",
    "                         X_train = train_stage2[best_features], y_train = train_stage2['certified'], \n",
    "                         X_valid = valid_stage2[best_features], y_valid = valid_stage2['certified'],\n",
    "                         run_test = False,\n",
    "                         return_pred = True)\n",
    "rand['model_eval_dict']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'model_type': 'adaboost',\n",
       " 'model': [AdaBoostClassifier(base_estimator=DecisionTreeClassifier(), learning_rate=3,\n",
       "                     n_estimators=10)],\n",
       " 'train_score': [1.0],\n",
       " 'validation_score': [0.9971461187214612]}"
      ]
     },
     "execution_count": 79,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# our ada boost model (using decision tree as base classifier)  (no feature selection)\n",
    "ada = train_and_evaluate_model(model = AdaBoostClassifier,\n",
    "                         model_type = \"adaboost\",\n",
    "                         model_params = best_params_ada, \n",
    "                         X_train = train_stage2[best_features], y_train = train_stage2['certified'], \n",
    "                         X_valid = valid_stage2[best_features], y_valid = valid_stage2['certified'],\n",
    "                         run_test = False,\n",
    "                         return_pred = True)\n",
    "ada['model_eval_dict']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'model_type': 'knn',\n",
       " 'model': [KNeighborsClassifier(n_neighbors=3)],\n",
       " 'train_score': [0.96988295746503],\n",
       " 'validation_score': [0.944634703196347]}"
      ]
     },
     "execution_count": 80,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# our k neighbors classifier model  (no feature selection)\n",
    "knn = train_and_evaluate_model(model = KNeighborsClassifier,\n",
    "                         model_type = \"knn\",\n",
    "                         model_params = best_params_knn, \n",
    "                         X_train = train_stage2[best_features], y_train = train_stage2['certified'], \n",
    "                         X_valid = valid_stage2[best_features], y_valid = valid_stage2['certified'],\n",
    "                         run_test = False,\n",
    "                         return_pred = True)\n",
    "knn['model_eval_dict']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Models with feature selection implemented"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'model_type': 'decision',\n",
       " 'model': [DecisionTreeClassifier(max_depth=1)],\n",
       " 'train_score': [0.9972880388238653],\n",
       " 'validation_score': [0.9988584474885844]}"
      ]
     },
     "execution_count": 81,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# our selected decision tree model (with feature selection)\n",
    "tree2 = train_and_evaluate_model(model = DecisionTreeClassifier,\n",
    "                         model_type = \"decision\",\n",
    "                         model_params = best_params_decision_w_select, \n",
    "                         X_train = X_train_w_select, y_train = train_stage2['certified'], \n",
    "                         X_valid = X_valid_w_select, y_valid = valid_stage2['certified'],\n",
    "                         run_test = False,\n",
    "                         return_pred = True)\n",
    "tree2['model_eval_dict']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'model_type': 'randomforest',\n",
       " 'model': [RandomForestClassifier(max_depth=20, min_samples_leaf=50, n_estimators=10)],\n",
       " 'train_score': [0.9970025692263774],\n",
       " 'validation_score': [0.9988584474885844]}"
      ]
     },
     "execution_count": 82,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# our random forest model  (with feature selection)\n",
    "rand2 = train_and_evaluate_model(model = RandomForestClassifier,\n",
    "                         model_type = \"randomforest\",\n",
    "                         model_params = best_params_rand_w_select, \n",
    "                         X_train = X_train_w_select, y_train = train_stage2['certified'], \n",
    "                         X_valid = X_valid_w_select, y_valid = valid_stage2['certified'],\n",
    "                         run_test = False,\n",
    "                         return_pred = True)\n",
    "rand2['model_eval_dict']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model Results and Vetting"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Looking at our final results, here we plot some of the predictions on our test, validation and training data and how they compare"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWcAAAEICAYAAACDGjUCAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8vihELAAAACXBIWXMAAAsTAAALEwEAmpwYAAAiyElEQVR4nO3deZgdZZn+8e+dbrJBEoghGAIkQbZJkE1BEAdx2BcJ+JMhqAwgEJBNBmRnZJEow4WoCAgBhCAQDAjDooIIhG0wYQ8TFsGgISSEhCUJkK27n98fVR1POt2nT58+J6e6+v7kquvU/j7Vp/rJ22+9VaWIwMzMsqVHrQMwM7NVOTmbmWWQk7OZWQY5OZuZZZCTs5lZBjk5m5llkJOzmVkGOTl3gKQ+ku6TtEDSHZ3Yz7cl/amSsdWKpH+V9HqZ224u6QVJiySdXOnYKk3ScEkhqX41lrmepMfTn9FPV1e5Vnu5TM6SviXpWUkfS5oj6Y+SvlKBXX8TWA/4TEQcXO5OIuLWiNizAvFUVZqINim2TkQ8ERGbl1nEGcDkiOgXEVeUuY+8GwvMB/pHxGmrq1BJkyUdvbrKs1XlLjlLOhX4OfBjkkS6EXA1MLoCux8G/DUiGiqwry6vAjXIYcD0apW9Omu4VTQMeCXKuJU3J8fffUVEbgZgAPAxcHCRdXqRJO/Z6fBzoFe6bFdgFnAa8B4wBzgyXXYhsAxYnpZxFHABcEvBvocDAdSn00cAM4BFwFvAtwvmP1mw3ZeBZ4AF6eeXC5ZNBn4EPJXu50/AoDaOrTn+MwriPxDYF/gr8AFwTsH6OwBPAx+l614J9EyXPZ4eyyfp8R5SsP8zgXeB3zTPS7f5XFrGdun0+iS1vl1bifURoBFYku5/s/T7uxmYB/wDOA/oUfAzewr4WVrGxa3s8wLgTuAWYCFwdLFjTLcJ4DjgDeBD4CpA6bI64LL0GGYAJ7T4ftcH7k3jeRM4pkUsd6SxLAJeTo/x7PS7eRvYs53z+SaS821Z+jPandLO38LvpwdwFvA34H1gEjAwXb93Gt/76c/nGZIKzbgW382Vtf7d7o5DzQOo6MHA3kBD8y9PG+tcBPwFGAysC/wv8KN02a7p9hcBa5AktU+BddLlF7ByMm45Pbz5lxdYM00Qm6fLhgCj0vEjSJMzMDBNCoel2x2aTn8mXT45/cXaDOiTTl/SxrE1x//DNP5jSBLdbUA/YFT6C7dxuv4XgB3TcocDrwKnFOwvgE1a2f9/p0miDwXJOV3nmHQ/fYEHgcuKfBeTgaMLpm8G7kljHU7yH8pRBT+zBuCkNN4+rezvApJkdiBJUupT4jHeD6xN8lfWPGDvdNlxwGvAhun39CgrJ+fHSP4q6w1sk267W0EsS4C90rJvJvkP+tyC7+atEs7pmyj4j4jSzt/C7+eUdP0N0nnXAhPT9Y8F7ku/q7r0Z9W/te/GQw3yWa0DqOjBwLeBd9tZ52/AvgXTewF/T8d3BRZTkNxJajk7puMX0LHk/BHw/1omElZOzocBU1ssfxo4Ih2fDJxXsOx44IE2jq05/rp0ul8az5cK1nkOOLCN7U8B7i6Ybi05LwN6t5g3q8V+7iWpKU4jrdW1Ud6KBJAmh6XAyILlx5K0STf/zGa2891eADzezjqtHeNXCqYnAWel448AxxUs27Pg+92QpHbZr2D5T4CbCmJ5qGDZ10lqoS2/m7XbifcmVk7O7Z2/Lb+fV0n/w0inh5D8B1YPfJckuW9V7LvxUJshb23O7wOD2mlrW5/kT+Zm/0jnrdhHrNym/CmwVkcDiYhPSJoCjgPmSPq9pC1KiKc5pqEF0+92IJ73I6IxHV+cfs4tWL64eXtJm0m6X9K7khaStNMPKrJvgHkRsaSdda4DtgR+GRFL21m32SCgJ6t+N4U/h7dL2M9K65R4jG39fNdvsb/C2NYHPoiIRUXibflzn9/Kd9PRc6u987fl9zMMuFvSR5I+IknWjSTNF78h+evmdkmzJV0qaY0OxmNVkrfk/DTJn5IHFllnNskJ22yjdF45PiH5k7DZZwsXRsSDEbEHSW3lNZKk1V48zTG9U2ZMHfErkrg2jYj+wDmA2tmm6IUpSWuRtIPeAFwgaWCJscwnqdG1/G4Kfw6lXBRruU45x9hsDkkNuTCeZrOBgZL6FYm3Gto7f1se/9vAPhGxdsHQOyLeiYjlEXFhRIwkue6xP/AfbezHVrNcJeeIWEDS3nqVpAMl9ZW0hqR9JF2arjYROE/SupIGpevfUmaRLwK7SNpI0gCSiz3Aiv6pB0hak+TP9Y9Jaiwt/QHYLO3+Vy/pEGAkSTtotfUjaRf/OK3Vf6/F8rnAxh3c5y+A5yLiaOD3wDWlbJTWKCcB4yT1kzQMOJXyv5tm7R1jMZOAkyVtIGkdkgtrzfG+TdIk8BNJvSVtRXKR+NZOxtuejp6/15D8TIcBpNuNTse/JunzkupIfkbL+ec5Ws53bxWUq+QMEBGXk/xSn0dygeZt4ETgf9JVLgaeJWkPfRl4Pp1XTlkPAb9N9/UcKyfUHiS9PmaTXM3/Kkl7cct9vE9SYzmNpFnmDGD/iJhfTkwd9APgWyS9Ca4jOZZCFwAT0j+J/729naW/9HuTNOVA8j1sJ+nbJcZzEslfIzOAJ0kuZP66xG3b0t4xFnMdyZ/9L5GcJ3e1WH4oyXWG2cDdwPnpOVFNHT1/f0FyDeBPkhaRXBz8UrrssyS9WxaSNHc8xj8T/S+Ab0r6UJL7oNdAc5chMzPLkNzVnM3M8sDJ2azG0scMtDb8a61j6+4k/VrSe5L+r2DeQEkPSXoj/VynYNnZkt6U9LqkvQrmf0HSy+myKyS1e1HaydmsxiJirTaGJ2odm3ETyXWUQmcBD0fEpsDD6TSSRgJjSG722hu4Or3YCkmvobHApunQcp+rqPq9959c/B03atsqBlz0aK1DsAxqWPZOqd0c27R8/oySc84agzYuWl5EPC5peIvZo0lu+AGYQHLDzpnp/NvTvv1vSXoT2EHS30nuvHwaQNLNJN19/1isbNeczazbkjQ2fYJl8zC2hM3Wi4g5AOnn4HT+UFa+aWlWOm9oOt5yflF+apWZ5UtTa7cTtC4ixgPjK1Rya7XwKDK/KCdnM8uXxqo/0XeupCERMUfSEJLn70BSIy68o3QDkj7ws9LxlvOLcrOGmeVKRFPJQ5nuBQ5Pxw8neZJi8/wxknpJGkFy4W9q2vSxSNKOaS+N/yjYpk2uOZtZvjSVnXRXIWkiycW/QZJmAecDlwCTJB0FzAQOBoiI6ZImAa+QPLr1hIIHXX2PpOdHH5ILgUUvBoKTs5nlTfk14lV3FXFoG4t2a2P9cSQvK2g5/1mSJzWWzMnZzPKlAxcEs8zJ2czypYI151pycjazXInq99ZYLZyczSxfKnhBsJacnM0sX9ysYWaWQb4gaGaWQa45m5llkC8ImpllkC8Impllzz/vmO7anJzNLF/c5mxmlkFu1jAzyyDXnM3MMqhxea0jqAgnZzPLFzdrmJllkJs1zMwyyDVnM7MMcnI2M8ue8AVBM7MMcpuzmVkGuVnDzCyDXHM2M8sg15zNzDLINWczswxq8MP2zcyyxzVnM7MMcpuzmVkGueZsZpZBrjmbmWWQa85mZhnk3hpmZhkUUesIKsLJ2czyxW3OZmYZ5ORsZpZBObkg2KPWAZiZVVRjY+lDOyT9p6Tpkv5P0kRJvSUNlPSQpDfSz3UK1j9b0puSXpe0V2cOw8nZzPKlqan0oQhJQ4GTgS9GxJZAHTAGOAt4OCI2BR5Op5E0Ml0+CtgbuFpSXbmH4eRsZvlSoeScqgf6SKoH+gKzgdHAhHT5BODAdHw0cHtELI2It4A3gR3KPQwnZzPLl2gqeZA0VtKzBcPYFbuJeAe4DJgJzAEWRMSfgPUiYk66zhxgcLrJUODtgkhmpfPK4guCZpYr0VR6P+eIGA+Mb21Z2pY8GhgBfATcIek7RXan1oooOZgWnJzNLF8q15Vud+CtiJgHIOku4MvAXElDImKOpCHAe+n6s4ANC7bfgKQZpCxu1jCzfKlcb42ZwI6S+koSsBvwKnAvcHi6zuHAPen4vcAYSb0kjQA2BaaWexiuOZtZvlSo5hwRUyTdCTwPNAAvkDSBrAVMknQUSQI/OF1/uqRJwCvp+idERPv99drg5Gxm+VLBOwQj4nzg/Bazl5LUoltbfxwwrhJlOzm3QQOH0OsbJ66Y7rHOYJY9dicNUx9cZd0eQzam95EXsPSuX9L42jOdK7iunl4HHEePISOIxYtYeteVxIL59FhvI3rucyTq1Qeamlj21D00vjKlc2VZTV03/qfst+/uvDdvPtts2+rvupUjJw8+cptzG+KDOSy5/txkuOE8YvlSGl9/dtUVJXrudgiNM6Z1aP8aMIjeh527yvz6bXYllnzC4qtPY/mUB+j5b2OSeJYvY+k917D42rNYMvFSeu5xGPTqW9axWTbcfPMk9tv/27UOI38q28+5ZtpNzpIuldRf0hqSHpY0v53uJLlTN2IU8eF7xIL3V1lWv/2eNLz6DPHJwpW32XJneh95Ib2PHkfPfb8Laq2XTStlbbYdDdOeAKDx1anUjRgFQHzwLvHh3GT844+ITxegvv06c1hWY088OYUPPvyo1mHkT1OUPmRYKTXnPSNiIbA/SVeRzYDTqxpVxtSN3ImG6U+vMl/91qF+8y/S8PzDK8//zPrUj/wSSyZcxJLrz4WmJuq33Lmksnr0W4dY+EEyEU3E0k+hz1orr7P+xqiunvjwvVb2YNbNVfDZGrVUSpvzGunnvsDEiPhA7dQC07tsxgJcccAOfHf7TTsVZE31qKN+s+349NHfrrKo5x7fYdkjt6/SxlU3YhQ9hoyg93cvAkBr9CQ+TWrWvb55Clp7XVRXjwZ8ht5HJ9cOGp55kIaXHm+3hq211qbX6O+x9N5r6ET/drPciow3V5SqlOR8n6TXgMXA8ZLWBZYU26DwrptPLv5Ol84gdZtsTdO7f4cWzRYAPdYfQa+DkouG6tuP+k22Zml6YjRMe4Llj05aZZuld/48WX/AIHodcCxLfrPyhd2mhR+g/gOJRR+AeqBefWHxx8nCnn3odcgPWDb5Dpre+VvlDtIsTzLeXFGqdpNzRJwl6b+BhRHRKOkTklsau4X6Ua03aQAsvvLUFeM9vz6WxjdeoPGvz6FB69P7309l+ZQH4NOF0HtN1Kt3q23WLTX+9Xnqt/pXlr3zJnX/sgONf38lWdCjjt4Hn0LDy0/Q+GrZ/drN8i8nz3NuNzlLWgM4DNglbc54DLimynFlQ31P6kZsydI//Pqfs7b7NwAann+kzc1i/myWTb6D3t86E0lEUyPLHrippOTc8OJj9Bp9HH2O/ymx+GOW3n0lAHUjd6THRptT32ct6rfaBYBl911L09yZnTlCq6FbfnMVX91lJwYNGsjfZzzLhRddxo033V7rsLq+nNScFe30CZR0PUm7c/Mj8g4DGiPi6FIK6OrNGlYdAy56tNYhWAY1LHuntG5NRXzywzEl55w1L7q90+VVSyltzttHxNYF049IeqlaAZmZdUpOmjVK6UrXKOlzzROSNgay3QfFzLqvnPRzLqXmfDrwqKQZJM8rHQYcWdWozMzK1C260qXd5haQvGplMElyfi0ilq6G2MzMOi7jNeJStdmsIeloYDrwS+BFYHhEvOTEbGaZ1g2aNU4BRkXEvLSd+VaSh0mbmWVXxm/LLlWx5Lys+fUsETFDUq/VFJOZWdk68g7BLCuWnDeQdEVb0xFxcvXCMjMrUzdIzi2fPPdcNQMxM6uIvPfWiIgJbS0zM8usblBzNjPrepyczcyyJxpz3qxhZtYl5aTmXNYLXiXtX+lAzMwqIZqi5CHLyn379vYVjcLMrFK6wR2CbYqI8ysdiJlZReSjybnt5CzpG8U2jIi7Kh+OmVnnREM+snOxmvPX08/BwJeB5vcyfQ2YDDg5m1n25CM3F70J5UgASfcDIyNiTjo9BLhq9YRnZtYxWb/QV6pS2pyHNyfm1FxgsyrFY2bWOXmvOReYLOlBYCIQwBjAb+c0s0zqNjXniDhR0kHALums8RFxd3XDMjMrUzeqOQM8DyyKiD9L6iupX0QsqmZgZmbliIZaR1AZ7d6EIukY4E7g2nTWUOB/qhiTmVnZoqn0IctKuUPwBGBnYCFARLxB0r3OzCx7mjowZFgpyXlpRCxrnpBUT3Jh0MwscypZc5a0tqQ7Jb0m6VVJO0kaKOkhSW+kn+sUrH+2pDclvS5pr84cRynJ+TFJ5wB9JO0B3AHc15lCzcyqpcLNGr8AHoiILYCtgVeBs4CHI2JT4OF0GkkjSXqzjQL2Bq6WVFfucZSSnM8E5gEvA8cCfwDOK7dAM7NqikaVPBQjqT9JL7UbACJiWUR8BIwGmt8UNQE4MB0fDdweEUsj4i3gTWCHco+jaG8NST2AaRGxJXBduYWYma0uHbnQJ2ksMLZg1viIGJ+Ob0xSMb1R0tYk71H9PrBe8415ETFHUvM1uKHAXwr2NSudV5aiyTkimiS9JGmjiJhZbiFmZqtLNBWvEa+0bpKIx7exuB7YDjgpIqZI+gVpE0YbWiu47OtzpfRzHgJMlzQV+GRFiREHlFuomVm1VLCL3CxgVkRMSafvJEnOcyUNSWvNQ4D3CtbfsGD7DYDZ5RZeSnK+sNydm5mtbhGl15yL7yfelfS2pM0j4nVgN+CVdDgcuCT9vCfd5F7gNkmXA+sDmwJTyy2/2POcewPHAZuQXAy8ISIv996YWV5V+OaSk4BbJfUEZgBHknSkmCTpKGAmcDBAREyXNIkkeTcAJ0REY7kFF6s5TwCWA08A+wAjSRrDzcwyq6mdXhgdEREvAl9sZdFubaw/DhhXibKLJeeREfF5AEk30InquZnZ6tKRC4JZViw5L28eiYgGKR8HbGb51h2S89aSFqbjIrlDcGE6HhHRv+rRmZl1UOTk4RLFXlNV9m2HZma10h1qzmZmXU6lutLVmpOzmeVKYwV7a9SSk7OZ5YprzmZmGeQ2ZzOzDMp9bw0zs67INWczswxqbCrlHSLZ5+RsZrniZg0zswxqcm8NM7PscVc6M7MMcrNGiQZc9Gi1i7AuaPHsJ2odguWUmzXMzDLIvTXMzDIoJ60aTs5mli9u1jAzyyD31jAzy6DKvny7dpyczSxXAteczcwyp8HNGmZm2eOas5lZBrnN2cwsg1xzNjPLINeczcwyqNE1ZzOz7MnJW6qcnM0sX5pcczYzyx4/+MjMLIN8QdDMLIOa5GYNM7PMaax1ABWSj1cGmJmlmlT6UApJdZJekHR/Oj1Q0kOS3kg/1ylY92xJb0p6XdJenTkOJ2czy5UmVPJQou8DrxZMnwU8HBGbAg+n00gaCYwBRgF7A1dLqiv3OJyczSxXogNDeyRtAOwHXF8wezQwIR2fABxYMP/2iFgaEW8BbwI7lHscTs5mlisdadaQNFbSswXD2Ba7+zlwBit3AlkvIuYApJ+D0/lDgbcL1puVziuLLwiaWa50pCtdRIwHxre2TNL+wHsR8ZykXUvYXWvtJGV3u3ZyNrNcaaxcT7qdgQMk7Qv0BvpLugWYK2lIRMyRNAR4L11/FrBhwfYbALPLLdzNGmaWK00dGIqJiLMjYoOIGE5yoe+RiPgOcC9weLra4cA96fi9wBhJvSSNADYFppZ7HK45m1murIY7BC8BJkk6CpgJHAwQEdMlTQJeARqAEyKi7G7XTs5mlivVeIVgREwGJqfj7wO7tbHeOGBcJcp0cjazXPGzNczMMigvt287OZtZrvhh+2ZmGeRmDTOzDHJyNjPLIL8Jxcwsg9zmbGaWQe6tYWaWQU05adhwcjazXPEFQTOzDMpHvdnJ2cxyxjVnM7MMalA+6s5OzmaWK/lIzU7OZpYzbtYwM8sgd6UzM8ugfKRmJ2czyxk3a5iZZVBjTurOTs5mliuuOZuZZVC45mxmlj2uOVvJrhv/U/bbd3femzefbbZt9Y3q1oWd9+PLefypqQxcZ23+55ZrOr2/e/7wENdOuB2AYw8fw+h99wDgv37yM6a/9gYRwfANhzLu3NPo27dPp8vLm7x0petR6wC6g5tvnsR++3+71mFYlRy47x5cc/nFHd7uiBPP4J05c1eat2DhIn51421MvO7nTLzu5/zqxttYsHARAGeePJa7JlzN3Tf/iiHrDea2391XkfjzJjowZJmT82rwxJNT+ODDj2odhlXJF7f5PAP691tp3sxZszn21PP49++exH987wfM+MfbJe3rqSnPsdP22zKgfz8G9O/HTttvy1NTngNgrTXXBCAiWLJ0KcrJGz8qrYEoeciydps1JN3Hqv/JLACeBa6NiCXVCMysK7vw0iv44eknMWzDoUyb/hoXX3YVv/7lJe1uN3fefD47eN0V0+utO4i58+avmD5v3OU8/vQzfG74Rpx+0jFVib2r604XBGcA6wIT0+lDgLnAZsB1wGEtN5A0FhgLoLoB9OixZkWCNesKPv10MS++/CqnnvfjFfOWLV8OwN2//xO3TLoHgJnvzOZ7P/gv1qhfg6Hrr8cVP/kh0UpeUUEV+eJzT6WxsZEf/+xXPPDw4xy0357VPZguqDtdENw2InYpmL5P0uMRsYuk6a1tEBHjgfEA9T2H5uO/MbMSNUUT/fqtye8mXLXKsoP223NFQj3ixDMYd+5pDB2y3orlnx08iGdemLZieu68+Wy/7VYr7aOuro69d9uFG2/7nZNzK/JScy6lzXldSRs1T6Tjg9LJZVWJyqwLW2vNNRk65LM8+MgTQNJG/NobM0raducvfYH/nfo8CxYuYsHCRfzv1OfZ+UtfICKYOWv2iv1NfmoKI4ZtULVj6MqaOjBkWSk159OAJyX9DRAwAjhe0prAhGoGlxe3/OYqvrrLTgwaNJC/z3iWCy+6jBtvur3WYVmFnH7+JTzzwjQ++mghux34HY4/6jD++/wz+NFlV3LthIk0NDSwz25fZYtNN253XwP69+PYIw5lzNHfB+C4I7/FgP79aGpq4pyLf8onn3xKRLD5JiP4r9NPrPahdUmNrbUNdUGKEg5EUi9gC5Lk/FpHLgK6WcNas3j2E7UOwTJojUEbd7oPyreGHVRyzrntH3dnts9Lu80akvoCpwMnRsSLwIaS9q92YGZm5YgO/MuyUtqcbyRpW94pnZ4FdLzHvZnZapCXNudSkvPnIuJSYDlARCwmad4wM8ucJqLkIctKuSC4TFIf0htRJH0OWFrVqMzMypT15opSlVJzPh94gKSt+VbgYeCMqkZlZlamxoiSh2IkbSjpUUmvSpou6fvp/IGSHpL0Rvq5TsE2Z0t6U9LrkvbqzHG0W3OOiIckPQ/sSNKc8f2ImN/OZmZmNVHB5ooG4LSIeF5SP+A5SQ8BRwAPR8Qlks4CzgLOlDQSGAOMAtYH/ixps4hoLKfwNpNz4Y0nqZfTz76SNoqImeUUaGZWTZW60BcRc4A56fgiSa8CQ4HRwK7pahOAycCZ6fzbI2Ip8JakN4EdgKfLKb9Yzfn3JO3MhRf/guQ5G4OBunIKNDOrpo60ORc+Byg1Pn38RMv1hgPbAlOA9dLETUTMkTQ4XW0o8JeCzWal88rSZnKOiM+3EtyZwO7Aj1vbxsys1jrSrFH4HKC2SFoL+B1wSkQsVNvPam1tQdltLKXchLKppJuAPwLPASMj4pflFmhmVk0RUfLQHklrkCTmWyPirnT2XElD0uVDgPfS+bOADQs23wCYXe5xtJmcJW0paWIa2J+BLSPi+ohYXm5hZmbV1kiUPBSjpIp8A/BqRFxesOhe4PB0/HDgnoL5YyT1kjQC2BSYWu5xFGtzfgl4m6TteQdgh8LqfEScXG6hZmbVUsHeGjuTPK/+ZUkvpvPOAS4BJkk6CpgJHAwQEdMlTQJeIenpcUK5PTWgeHL+brk7NTOrlVKaK0rcz5O0fTd0q29qjohxwLhKlF/sgqAfB2pmXU7Wb8suVSm3b5uZdRl5uX3bydnMciUvD9t3cjazXMlLs0YpDz5ahR+2b2ZZlZdHhpaVnIHtKxqFmVmFVPImlFoqq1kjIs6vdCBmZpWQ9RpxqYo9le4bxTYsuJXRzCwzukNvja+nn4OBLwOPpNNfI3lEnpOzmWVOY2T97YClKXYTypEAku4nedjRnHR6CHDV6gnPzKxjst6WXKpS2pyHNyfm1FxgsyrFY2bWKblvcy4wWdKDwESSZ5OOAR6talRmZmXqDm3OAETEiZIOAnZJZ42PiLurG5aZWXmaulGzBsDzwKKI+LOkvpL6RcSiagZmZlaOblNzlnQMyTu2BgKfI3kn1jW08cg8M7Nayn1vjQInkDxsfwpARLxR8EJDM7NM6U7NGksjYlnzW1Ak1dOJlxaamVVTt2nWAB6TdA7QR9IewPHAfdUNy8ysPHmpOZfy4KMzgXnAy8CxwB+A86oZlJlZuaID/7KsaM1ZUg9gWkRsCVy3ekIyMytfY/nvVM2UojXniGgCXpK00WqKx8ysU7rTI0OHANMlTQU+aZ4ZEQdULSozszJ1p9u3L6x6FGZmFZL1GnGpij3PuTdwHLAJycXAGyKiYXUFZmZWjrz01ihWc54ALAeeAPYBRgLfXx1BmZmVK+u9MEpVLDmPjIjPA0i6AZi6ekIyMytfd7h9e3nzSEQ0NN8haGaWZblvcwa2lrQwHRfJHYIL0/GIiP5Vj87MrINy3+YcEXWrMxAzs0roDjVnM7Mupzv1czYz6zJcczYzy6Du0FvDzKzLyf0FQTOzrsjNGmZmGdQd7hA0M+tyXHM2M8ugvLQ5Ky//y3QFksZGxPhax2HZ4vPCWlPKOwStcsbWOgDLJJ8XtgonZzOzDHJyNjPLICfn1cvtitYanxe2Cl8QNDPLINeczcwyyMnZzCyDnJwLSDpIUkjaooR1T5HUtxNlHSHpyjbmz5P0oqRXJB1TbhnWeRk6J5okbVUw7/8kDS+3LMs+J+eVHQo8CYwpYd1TgLJ/Edvx24jYBtgV+LGk9apUjrUvK+fELODcKu3bMsjJOSVpLWBn4CgKfhEl1Um6TNLLkqZJOknSycD6wKOSHk3X+7hgm29Kuikd/7qkKZJekPTnjiTaiHgP+BswTNJu6T5elvRrSb3S/V+S1rCnSbqs8z8Ja5axc+J+YJSkzVuJc09JT0t6XtIdadxI2lfSa5KelHSFpPvL/2nY6ubk/E8HAg9ExF+BDyRtl84fC4wAto2IrYBbI+IKYDbwtYj4Wjv7fRLYMSK2BW4Hzig1IEkbAxuT1JpuAg6JiM+TPBPle5IGAgcBo9LYLi5131aSA8nOOdEEXAqcUzhT0iDgPGD3iNgOeBY4VVJv4Fpgn4j4CrBuCWVYhjg5/9OhJL8opJ+HpuO7A9dERANARHzQwf1uADwo6WXgdGBUCdscIulFYCJwLMkv1ltpkgCYAOwCLASWANdL+gbwaQdjs+KydE4A3AbsKGlEwbwdgZHAU+k5czgwDNgCmBERb6XrTexgjFZjfiodIOkzwL8BW0oKoA4ISWcAgpIeEFu4Tu+C8V8Cl0fEvZJ2BS4oYV+/jYgTC+LbptUCIxok7QDsRvJn94npcVgnZfCcaP6+fwqcWRgq8FBEHFq4rqRtS9mnZZdrzolvAjdHxLCIGB4RGwJvAV8B/gQcJ6keIG1KAFgE9CvYx1xJ/yKpB0lTQ7MBwDvp+OFlxvcaMFzSJun0YcBjadvigIj4A8nFqG3K3L+tKqvnxE0kNffmZoq/ADs3nxuS+krajOSc2bigR8chHSzHaszJOXEocHeLeb8DvgVcD8wEpkl6KZ0HyS23f2y++AOcRXLR5hFgTsF+LgDukPQEML+c4CJiCXBkup+XSdofryFJBPdLmgY8BvxnOfu3VmXynIiIZcAVwOB0eh5wBDAxPQ/+AmwREYuB44EHJD0JzAUWdKQsqy3fvm2WU5LWioiPJQm4CngjIn5W67isNK45m+XXMelFwukkTSnX1jYc6wjXnM3MMsg1ZzOzDHJyNjPLICdnM7MMcnI2M8sgJ2czswz6/xQWml/nbx9dAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "# Compute the four values tp, fp, fn, tn and plot them using plot_confusion_matrix\n",
    "\n",
    "def plot_confusion_matrix(pred_data, actual_data, model_type):\n",
    "    \"\"\"\n",
    "    Plots a confusion matrix using the values \n",
    "       tp - True Positive\n",
    "       fp - False Positive\n",
    "       fn - False Negative\n",
    "       tn - True Negative\n",
    "    \"\"\"\n",
    "    tn, fp, fn, tp = confusion_matrix(y_true = pred_data, \n",
    "                                  y_pred = actual_data.to_numpy()).ravel()\n",
    "    data = np.matrix([[tp, fp], [fn, tn]])\n",
    "\n",
    "    sns.heatmap(data,annot=True,xticklabels=['Actual Pos', 'Actual Neg']\n",
    "              ,yticklabels=['Pred. Pos', 'Pred. Neg']) \n",
    "    plt.title('Confusion matrix for {}'.format(model_type))\n",
    "    plt.show()\n",
    "    \n",
    "\n",
    "plot_confusion_matrix(tree['validate_pred'], valid_stage2['certified'], model_type = \"decision_tree\")\n",
    "plot_confusion_matrix(rand['validate_pred'], valid_stage2['certified'], model_type = \"random_forest\")\n",
    "plot_confusion_matrix(ada['validate_pred'], valid_stage2['certified'], model_type = \"ada_boost\")\n",
    "plot_confusion_matrix(knn['validate_pred'], valid_stage2['certified'], model_type = \"knn\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWcAAAEICAYAAACDGjUCAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8vihELAAAACXBIWXMAAAsTAAALEwEAmpwYAAAohUlEQVR4nO3deZwU1b3+8c8zLIoKxgWUTUHEKGjEqITExIuJuF0NaKJijOKKcYma5BcTlyQuIddrEpOYqAkuV4wrucYr7jtGjRsaFEGNKESBEdwBg8DMfH9/1BltYOjuaXpmeprnzateXXWqTp3T3TVfTp86VaWIwMzMKktNW1fAzMxW5eBsZlaBHJzNzCqQg7OZWQVycDYzq0AOzmZmFcjB2cysAjk4tyBJXSTdLulDSX9Zg/0cLum+ctatrUj6iqRXSsz7WUn/kLRI0qnlrlsR5U+WdNwa7mO2pD3LVSerXg7OgKRvSZoiabGkWkl3S/pyGXb9TWAzYJOIOLjUnUTE9RGxVxnq06IkhaSt820TEY9GxGdLLOIMYHJEdI2IS0rcx1pF0lGSHmvreljzrfXBWdL3gd8CvyALpFsAlwEjy7D7LYF/RkRdGfbV7knquIa72BKY3kZlm7WuiFhrJ2BDYDFwcJ5t1iEL3vPS9FtgnbRuODAH+AGwAKgFjk7rzgOWActTGccC5wLX5ey7HxBAx7R8FPA6sAiYBRyek/5YTr4vAc8AH6bXL+WsmwxcADye9nMfsOlq3ltj/c/Iqf8oYD/gn8B7wFk52w8FngA+SNv+Aeic1v0tvZeP0vs9NGf/PwLeAv7cmJbyDEhlfD4t9wLeAYY3UdeHgHrg47T/bdL3dy3wNvAv4BygJuczexz4TSrj503sc7XvJ60fAbycPuc/AI8Ax+XU/SHg3VTn64HPFHHMzQbOBGYA7wP/A6ybs/54YGaq8ySgV5Hf+1GsdOwA26XPqz59Zh+09d+cp+KnNq9Am7552AeoIwXH1WxzPvAk0APoDvwduCCtG57ynw90SkHt38BGaf25rBiMV17ulwJaR2B9YCHw2bSuJzA4zR9FCs7AxumP+oiU77C0vElaPxl4LQWvLmn5wtW8t8b6/zTV/3iyQHcD0BUYnP64t0rb7wwMS+X2A14CTs/ZXwBbN7H//yb7T64LOcE5bXN82s96wL3Ar/J8F5NJwTEtXwvcluraj+w/lGNzPrM64Lupvl2a2N9q3w+wafo+vpk+m++l/TUG563Jgvc6ZMfF34DfFnHMzQZeBPqm7/Jx0n8cwFfJAv3n035/D/yt0PdOkceOp/Y1tXkF2vTNZ62Ltwps8xqwX87y3sDsND8cWEJOcCdrgQ5L8+fSvOD8AfCNlQMJKwbnI4CnV1r/BHBUmp8MnJOz7iTgntW8t8b6d0jLXVN9vpCzzbPAqNXkPx24NWe5qeC8jBVbhsPJCc4pbRIwDXiB9KtkNeVN5tPg2AFYCgzKWX8CWZ9042f2RjOPh0/eD3Ak8GTOOpH9CjhuNXlHAf8ooozZwHdylvcDXkvzVwEX5azbgOyXV79833uxx46n9jWt7X3O7wKbFuiP7EX2k7nRv1LaJ/uIFfuU/032R9UsEfERWVfAd4BaSXdK2raI+jTWqXfO8lvNqM+7EVGf5pek1/k565c05pe0jaQ7JL0laSFZP/2mefYN8HZEfFxgmyuA7YHfR8TSAts22hTozKrfTe7n8Ga+HRR4P71y80cW6d7MydtD0k2S5qa811H4s2iqXrnH0wrfbUQsJjtGe6+8Lidv72YcO9aOrO3B+Qmyn+2j8mwzj+xEVKMtUlopPiL7+d5o89yVEXFvRIwg+1n6MlnQKlSfxjrNLbFOzXE5Wb0GRkQ34CyyFmU+ee9JK2kDsn78q4BzJW1cZF3eIWtVrvzd5H4Ohe6Hm+/91JJ1PTTWU7nLwH+l/X8u5f02hT+LRrn7yT2eVvhuJa1P1m0xd+V1OXnnQt5jx/cEbqfW6uAcER+S9bdeKmmUpPUkdZK0r6SL0mY3AudI6i5p07T9dSUWORXYXdIWkjYkOzEEgKTNJH09/UEuJTuBU9/EPu4CtknD/zpKOhQYBNxRYp2aoytZ3+bi1DI7caX184GtmrnP3wHPRsRxwJ3AH4vJlFr7E4FxkrpK2hL4Ps37bvK9nzuBwZIOSr+sTmXF/0y7kk6ySeoN/LAZ5Z4sqU/6j+gs4OaUfgNwtKQhktYha8k/FRGzyfO9Fzh25gN9JHVuRv2sAqzVwRkgIi4m+6M+h+xk2JvAKcD/pU1+Dkwh6w+dBjyX0kop636yP8QXyPpycwNqDdmoj3lkZ+r/g6y/eOV9vAvsn7Z9l2ykxf4R8U4pdWqm/wd8i2xEwBV8GlQanQtMkPSBpEMK7UzSSLKTst9JSd8HPi/p8CLr812yXyOvA4+RBberi8wLed5P+jwPBi4k+5wHkp28a3Qe2Ym7D8kC+V+bUe4NZKNoXk/Tz1OZDwI/AW4ha7kPAEandfm+93zHzkNkww/fktQax4iVibKuNDMzqyRrfcvZzKwS+aopszKStAXZBSZNGRQRb7Rmfaz9creGmVkFavGW85Lrznb0t1V0Peaatq6CVaC6ZXOLHY64Wsvfeb3omNNp063WuLyW4j5nM7MK5D5nM6suDU1dHtD+ODibWXWpr4479Do4m1lViWho6yqUhYOzmVWXBgdnM7PK45azmVkF8glBM7MK5JazmVnlCY/WMDOrQD4haGZWgdytYWZWgXxC0MysArnlbGZWgXxC0MysAvmEoJlZ5ckezN7+OTibWXWpkj5n32zfzKpLQ0PxUx6S1pX0tKTnJU2XdF5KP1fSXElT07RfTp4zJc2U9IqkvXPSd5Y0La27RFLBJ7C45Wxm1aV8LeelwFcjYrGkTsBjku5O634TEb/K3VjSIGA0MBjoBTwgaZvI+lkuB8YCTwJ3AfsAd5OHg7OZVZf65WXZTWRPv16cFjulKd/zCUcCN0XEUmCWpJnAUEmzgW4R8QSApGuBURQIzu7WMLPq0oxuDUljJU3Jmcbm7kpSB0lTgQXA/RHxVFp1iqQXJF0taaOU1ht4Myf7nJTWO82vnJ6Xg7OZVZdoKHqKiPERsUvONH6FXUXUR8QQoA9ZK3h7si6KAcAQoBb4ddq8qX7kyJOel4OzmVWXMp0QzBURHwCTgX0iYn4K2g3AFcDQtNkcoG9Otj7AvJTep4n0vByczay6lG+0RndJn0nzXYA9gZcl9czZ7EDgxTQ/CRgtaR1J/YGBwNMRUQsskjQsjdI4Erit0NvwCUEzqypRphOCQE9ggqQOZA3ZiRFxh6Q/SxpC1jUxGzgBICKmS5oIzADqgJPj0ytiTgSuAbqQnQjMezIQHJzNrNqUaShdRLwA7NRE+hF58owDxjWRPgXYvjnlOzibWXXxvTXMzCpQlVy+7eBsZtXFLWczswrklrOZWQWq8832zcwqj1vOZmYVyH3OZmYVyC1nM7MK5JazmVkFcsvZzKwCebSGmVkFioK3Sm4XHJzNrLq4z9nMrAI5OJuZVSCfEDQzq0D19YW3aQccnM2surhbw8ysAjk4m5lVoCrpc/bTt82sqkRDFD3lI2ldSU9Lel7SdEnnpfSNJd0v6dX0ulFOnjMlzZT0iqS9c9J3ljQtrbskPYU7LwdnM6suDQ3FT/ktBb4aETsCQ4B9JA0Dfgw8GBEDgQfTMpIGAaOBwcA+wGXpyd0AlwNjgYFp2qdQ4Q7OZlZd6uuLn/KIzOK02ClNAYwEJqT0CcCoND8SuCkilkbELGAmMFRST6BbRDwREQFcm5NntRyczay6NKPlLGmspCk509jcXUnqIGkqsAC4PyKeAjaLiFqA9Nojbd4beDMn+5yU1jvNr5yel08Imll1acZojYgYD4zPs74eGCLpM8CtkrbPs7um+pEjT3peDs6rsbSunmMmTGZ5XQN1DcGe2/XmpOGDV9hm1jsL+dmkKbz01gecssdgxnzxs2tc7rK6es657Rleqn2fDbt05r+/MYzen1mfeR98xA/+8gT1EdTVB4cNHcDBOw9Y4/Ksbe2913Auvvh8OtTUcPX/3MhFv7y0ravU/rXAjY8i4gNJk8n6iudL6hkRtanLYkHabA7QNydbH2BeSu/TRHpe7tZYjc4darjiiP9g4gkjuHnsnvz9tbd4Yc67K2yzYZfOnLHPEI4ctk2z9z/3g4849trJq6TfOnU23dbtzO2n7Mu3v7ANv3twGgDdu3ZhwtF7MHHsCK479qtc/fgrLFi0pKT3ZpWhpqaGS343jv0P+DY77LgHhx46iu22G9jW1Wr/ynRCUFL31GJGUhdgT+BlYBIwJm02BrgtzU8CRktaR1J/shN/T6euj0WShqVRGkfm5FmtgsFZ0kWSuknqJOlBSe9I+nahfO2dJNbrnP2wqGvIWs8rD37ZeP112b7XxnTssOqvljtf+BeHX/Ugh4y/nwvufJb6AsN2Gk1+ZR4H7LglAHsO6s3TsxYQEXTqUEPnjtmJ32V19USV3BZxbTZ015147bXZzJr1BsuXL2fixNv4+gF7F85o+TVE8VN+PYGHJb0APEPW53wHcCEwQtKrwIi0TERMByYCM4B7gJNTtwjAicCVZCcJXwPuLlR4Md0ae0XEGZIOJGueHww8DFxXRN52rb4hOOzKB3jzvcUcussAdui9SVH5Xn97IffOmMM1R+1Bpw41jLvrOe6a9sYnQTefBYuWsHm3LgB0rKlhg3U78cGSZWy03jq89eG/+e5Nj/Pme4s5fc8d6NG1yxq9P2tbvXpvzptzPv11O2duLUN33akNa1QlynRvjYh4AVjlC4mId4GvrSbPOGBcE+lTgHz91asoJjh3Sq/7ATdGxHuFxk+nM55jAX5/9L4c+9X2ecB1qBETx45g4cfL+P7EJ5i54EO27rFhwXxPz17AS7Xvc/hVDwKwdHk9G6+/DgDfm/h35n7wEXX1DdR++G8OGX8/AN8aOpBRQ/o12V3W+GlvvuF6/OWEESxYtITvTfw7I7brwyYbrFuW92qtr6m/I/8iWnOxFl2+fbukl4ElwEmSugMf58uQewZ0yXVnt/ujrdu6ndlly+48/tpbRQXnCDjgc1ty6td2WGXdbw75EpD1Of900jNcdeTwFdZv1q0Lby1cwmbd1qOuoYHFHy9nwy6dV9imR9cuDOjejefeeIcRg/pg7dPcObX07dPrk+U+vXtSWzu/DWtUJYrsQqx0BfucI+LHwBeBXSJiOfAR2WDrqvbeR0tZ+PEyAD5eXs9Ts+bTf5OuReUd2r8H9788l/c+yv4P+3DJMuZ98FFRef9jm57c/vy/AHhgxlx27dcDScxf+G8+Xp79XFu4ZBlT33yXfkXWxyrTM1OmsvXW/enXry+dOnXikENGcvsd97V1tdq/aCh+qmAFW86SOgFHALunn2GPAH9s4Xq1uXcWL+Ent02hIYKGCPYa1Ifdt+nFX559DYCDdx7AO4s/5ltXPshHS5cjieufmslfT9yLAd27ccrwwXzn+keJgI414sx9d6LXZ9YvWO6BO/Xn7P97mgP+cDfdunTmvw/6AgCvv7OIi+9/HJENkDzyi9swcLPCrXirXPX19Zx2+jncdecNdKip4ZoJNzNjxj/bulrtX5W0nFWoj0vSlWT9zo2XKx4B1EfEccUUUA3dGlZ+XY+5pq2rYBWobtncgjcEKuSjn44uOuasf/5Na1xeSymmz3nXdOOPRg9Jer6lKmRmtkYqvLuiWMVchFIv6ZNL0SRtBVTHc2DMrPqUb5xzmyqm5fxDsoHYr5ON6toSOLpFa2VmVqK1YihdGjb3ITCU7M5LAl6OiKWtUDczs+ar8BZxsVbbrSHpOGA68HtgKtAvIp53YDazirYWdGucDgyOiLdTP/P1ZDf2MDOrXGW6fLut5QvOyyLibYCIeF3SOq1UJzOzkhV6NmB7kS8495F0yeqWI+LUlquWmVmJ1oLg/MOVlp9tyYqYmZVFtY/WiIgJq1tnZlax1oKWs5lZ++PgbGZWeaK+yrs1zMzapSppOZf0gFdJ+5e7ImZm5RANUfRUyUp9+vauZa2FmVm5lOkKQUl9JT0s6SVJ0yWdltLPlTRX0tQ07ZeT50xJMyW9ImnvnPSdJU1L6y5RoWf9UWK3RkT8rJR8ZmYtrnxdznXADyLiOUldgWcl3Z/W/SYifpW7saRBwGhgMNALeEDSNukJ3JeTPVf1SeAuYB8KPIF7tcFZ0kH5MkbEX/O+LTOzNhB15YnOEVEL1Kb5RZJeAnrnyTISuCndf2iWpJnAUEmzgW4R8QSApGuBUZQanIED0msP4EvAQ2l5D2Ay4OBsZpWnGbFZ0liyFm2j8ekB1Stv1w/YCXgK2A04RdKRwBSy1vX7ZIH7yZxsc1La8jS/cnpe+S5COTpV6g5gUPpfBEk9gUsL7djMrC0050RfCsSrBONckjYAbgFOj4iFki4HLiB7nOcFwK+BY8huqbxKEXnS8yqmz7lfY2BO5gPbFJHPzKz1lXGYc3rA9S3A9Y1duRExP2f9FcAdaXEO0Dcnex9gXkrv00R6XsWM1pgs6V5JR0kaA9wJPFxEPjOzVleuoXRpRMVVwEsRcXFOes+czQ4EXkzzk4DRktaR1B8YCDydGreLJA1L+zwSuK3Q+yjYco6IUyQdCOyeksZHxK2F8pmZtYnytZx3A44ApkmamtLOAg6TNISsa2I2cAJAREyXNBGYQTbS4+Q0UgPgROAaoAvZicC8JwOh+KF0zwGLIuIBSetJ6hoRi4rMa2bWaqKuTPuJeIym+4vvypNnHDCuifQpwPbNKb9gt4ak44H/Bf6UknoD/9ecQszMWks0FD9VsmL6nE8ma94vBIiIV8mG15mZVZ6GZkwVrJhujaURsazxakNJHSliGIiZWVuo9BZxsYoJzo9IOgvoImkEcBJwe8tWy8ysNNUSnIvp1vgR8DYwjeys5F3AOS1ZKTOzUkW9ip4qWd6Ws6Qa4IWI2B64onWqZGZWumppOecNzhHRIOl5SVtExButVSkzs1JFQ2W3iItVTJ9zT2C6pKeBjxoTI+LrLVYrM7MSrRUt5+S8Fq+FmVmZRFR5y1nSusB3gK3JTgZeFVGua2/MzFrG2tBynkB2H9JHgX2BQcBprVEpM7NSNVT4KIxi5QvOgyJiBwBJVwFPt06VzMxKtzacEFzeOBMRdUU8j9DMrM2tDcF5R0kL07zIrhBcmOYjIrq1eO3MzJopquTmEvkeU9WhNStiZlYOa0PL2cys3an6oXRmZu1R/VowWsPMrN1xy9nMrAJVS59zMbcMNTNrNyKKn/KR1FfSw5JekjRd0mkpfWNJ90t6Nb1ulJPnTEkzJb0iae+c9J0lTUvrLlERY5MdnM2sqkSDip4KqAN+EBHbAcOAkyUNAn4MPBgRA4EH0zJp3WhgMLAPcJmkxlFvlwNjgYFp2qdQ4Q7OZlZV6htqip7yiYjaiHguzS8CXiJ7wPVIsttbkF5HpfmRwE0RsTQiZgEzgaGSegLdIuKJiAjg2pw8q+XgbGZVpTndGpLGSpqSM41tap+S+gE7AU8Bm0VEbVZW1PLpA697A2/mZJuT0nqn+ZXT8/IJQTOrKg3NGK0REeOB8fm2kbQBcAtwekQszNNd3NSKyJOel4OzmVWVcg6lk9SJLDBfHxF/TcnzJfWMiNrUZbEgpc8B+uZk7wPMS+l9mkjPy90aZlZVyjhaQ8BVwEsRcXHOqknAmDQ/BrgtJ320pHUk9Sc78fd06vpYJGlY2ueROXlWq8Vbzl2Puaali7B2aMm8R9u6ClalmtOtUcBuwBHANElTU9pZwIXAREnHAm8ABwNExHRJE4EZZCM9To6I+pTvROAaoAtwd5rycreGmVWVQqMwihURj9F0fzHA11aTZxwwron0KcD2zSnfwdnMqkqV3DHUwdnMqksZuzXalIOzmVUV3/jIzKwCVcnDtx2czay6xGrP4bUvDs5mVlXq3K1hZlZ53HI2M6tA7nM2M6tAbjmbmVUgt5zNzCpQvVvOZmaVp0qe7+rgbGbVpcEtZzOzyuMbH5mZVSCfEDQzq0ANq3/GX7vi4GxmVaW+8CbtgoOzmVUVj9YwM6tA1TJaw0/fNrOqEs2YCpF0taQFkl7MSTtX0lxJU9O0X866MyXNlPSKpL1z0neWNC2tuyQ9hTsvB2czqyoNKn4qwjXAPk2k/yYihqTpLgBJg4DRwOCU5zJJHdL2lwNjgYFpamqfK3BwNrOq0tCMqZCI+BvwXpFFjwRuioilETELmAkMldQT6BYRT0REANcCowrtzMHZzKpKvYqfJI2VNCVnGltkMadIeiF1e2yU0noDb+ZsMyel9U7zK6fn5eBsZlWlOS3niBgfEbvkTOOLKOJyYAAwBKgFfp3Sm+ooiTzpeXm0hplVlZa+QjAi5jfOS7oCuCMtzgH65mzaB5iX0vs0kZ6XW85mVlVCxU+lSH3IjQ4EGkdyTAJGS1pHUn+yE39PR0QtsEjSsDRK40jgtkLluOVsZlWlnC1nSTcCw4FNJc0BfgYMlzSErGtiNnACQERMlzQRmAHUASdHROMFiyeSjfzoAtydprwcnM2sqpTz8u2IOKyJ5KvybD8OGNdE+hRg++aU7eBsZlXFl2+bmVUg3zLUzKwCOTibmVUgPwnFzKwCuc/ZzKwC+Wb7ZmYVqKFKOjYcnM2sqviEoJlZBaqOdrODs5lVGbeczcwqUJ2qo+3s4GxmVaU6QrODs5lVGXdrmJlVIA+lMzOrQNURmh2czazKuFvDzKwC1VdJ29nB2cyqilvOZmYVKKqk5eynb5tZVWloxlSIpKslLZD0Yk7axpLul/Rqet0oZ92ZkmZKekXS3jnpO0ualtZdkp7CnZeDcyvZe6/hTH/xb7w84zHO+OHJbV0dK9HSpcsYfdxpHDTmJEYefgJ/uPLPq2zz4cJFnHrm+Rx45ImMPu40Xn199hqXu2zZMn7wk/9i30OO4bDjT2du7XwA5r01n0OO+S7fGHMyIw8/gZtvvXONy2rvGoiipyJcA+yzUtqPgQcjYiDwYFpG0iBgNDA45blMUoeU53JgLDAwTSvvcxUOzq2gpqaGS343jv0P+DY77LgHhx46iu22G9jW1bISdO7ciasvuZC/TriM/51wKY8/9SzPv/jSCttcce3NbDtwALdeezm/+Mn/48Lf/rHo/c+tnc9Rp5yxSvpf77iPbl034O6JV3PEoaO4+LKrAei+ycZc98dfc8uES7nxit9y1XUTWfD2u2v2Jtu5aMZUcF8RfwPeWyl5JDAhzU8ARuWk3xQRSyNiFjATGCqpJ9AtIp6IiACuzcmzWg7OrWDorjvx2muzmTXrDZYvX87Eibfx9QP2LpzRKo4k1luvCwB1dXXU1dWx8i/U12a/wbCddwRgqy37Mrd2Pu+89z4At9/7EKOPO41vjDmZ8y66hPr64m4N/9CjTzByvz0B2Gv4V3jq2alEBJ06daJz584ALFu+nIaojv7WNVFHFD2VaLOIqAVIrz1Sem/gzZzt5qS03ml+5fS8CgZnSbdLmrTS9GdJp0lat8g3s1br1Xtz3pwz75PlOXNr6dVr8zaska2J+vp6vjHmZHbf/zC+uOtOfG7wtius/+zWW/HAI38HYNqMV6idv4D5C97htdlvcM+Dj/Dn1NKtqanhjvseLqrMBW+/y+Y9NgWgY8cObLD+enzw4UIAaue/zYFHnsieBx7JsYcfTI/um5Tx3bY/0Yx/ksZKmpIzjV2DopvqR4486XkVM1rjdaA7cGNaPhSYD2wDXAEcsUoNszc4FkAdNqSmZv0iiqleTfX9h1s47VaHDh24ZcKlLFy0mNPOvIBXX5/NwK36fbL+uCMO5sLf/olvjDmZgQP6se3AAXTo0IGnpkxlxsszGX3saQAsXbqUjTf6DACnnnk+c+fNZ3ndcmrnv803xmTnJb59yEgO/M+9mjxeGo+rnpt159ZrL2fB2+9y6pnnM2KPL7Ppxhutsv3aojlD6SJiPDC+mUXMl9QzImpTl8WClD4H6JuzXR9gXkrv00R6XsUE550iYvec5dsl/S0idpc0vakMuW+4Y+fea30Umjunlr59en2y3Kd3T2rTCR1rv7p13YBdP/85HntyygrBeYP11+fnZ38fyP4T3vubR9Gn12Y8O3UaX993T7534tGr7OuS//opkPU5nz3u11zzh4tWWL9Zj015a8E7bN6jO3V19Sz+6N9s2K3rCtv06L4JW/ffkueef5G99vhKmd9t+9EKQ+kmAWOAC9PrbTnpN0i6GOhFduLv6Yiol7RI0jDgKeBI4PeFCimmz7m7pC0aF9L8pmlxWZFvZq32zJSpbL11f/r160unTp045JCR3H7HfW1dLSvBe+9/wMJFiwH4eOlSnnzmH/Tfsu8K2yxctJjly5cDcMvt97DzkB3YYP31GbbLEO6f/Bjvvv8BkI3qmPdWcf9J7/HlYdx21wMA3Df5Ub6w845I4q0Fb/Px0qWf7O8f02bQb4s++XZV9co8lO5G4Angs5LmSDqWLCiPkPQqMCItExHTgYnADOAe4OSIaDypcCJwJdlJwteAuwuVXUzL+QfAY5JeI+s76Q+cJGl9Pj1jaXnU19dz2unncNedN9ChpoZrJtzMjBn/bOtqWQnefvd9zv75r6hvaCAagr2/+hWG7/aFT4awHXrgf/L6v97krAt+RYeaGrbqtwXnn3k6AAP6b8l3jz+SsaefTUM00KljR87+/kn02nyzguUetP/enHnBL9n3kGPYsFtXfnnejwF4ffab/PIPVyCJiOCoww5imwH9W+z9twf1ZewyjIjDVrPqa6vZfhwwron0KcD2zSlbxfR9SloH2JYsOL8cER8XW4C7NawpS+Y92tZVsArUadOtCl6cUci3tjyw6Jhzw79uXePyWkoxozXWA34InBIRU4G+kvZv6YqZmZWiOaM1Klkxfc7/Q9a3/MW0PAf4eYvVyMxsDZSzz7ktFROcB0TERcBygIhYQtPj9szM2lyZL99uM8WcEFwmqQtp0LSkAcDSFq2VmVmJKr27oljFBOefkQ0L6SvpemA34KiWrJSZWanKOVqjLRUMzhFxv6TngGFk3RmnRcQ7LV4zM7MSVHp3RbFWG5xzLzxJpqXX9SRtERFvtFy1zMxKU+kn+oqVr+V8J6vetCPI7rPRA+jQVCYzs7ZU9X3OEbFD7rKkfsCPgD2BX7RstczMSlMt3RrFXIQyUNI1ZNeCPwsMioiCN+0wM2sLEVH0VMny9TlvD5xN9siVi4Bjc27iYWZWkeqrpOWcr8/5ebK7+t8JDCV73MonKyPi1JatmplZ81VLt0a+4HxMq9XCzKxMKr27olj5Tgj6dqBm1u6sDS1nM7N2p+qH0pmZtUdrzeXbZmbtSbV0axRzy9BV+Gb7ZlapquWWoSUFZ2DXstbCzKxMynkRiqTZkqZJmippSkrbWNL9kl5NrxvlbH+mpJmSXpG095q8j5KCc0T8bE0KNTNrKS3Qct4jIoZExC5p+cfAgxExEHgwLSNpEDCa7MK9fYDLJJV8D6J8VwgelC9jRPy11ELNzFpKK4zWGAkMT/MTgMlk9x0aCdwUEUuBWZJmkl3A90QpheQ7IXhAeu0BfAl4KC3vkSrj4GxmFac+ynrT0ADukxTAnyJiPLBZRNQCREStpB5p297Akzl556S0kuS7COVoAEl3kN3sqDYt9wQuLbVAM7OW1JwrBCWNBcbmJI1PAbjRbhExLwXg+yW9nG93TVWn6MqspJihdP0aA3MyH9im1ALNzFpSc0ZhpEA8Ps/6eel1gaRbybop5kvqmVrNPYEFafM5QN+c7H2Aec2s/ieKOSE4WdK9ko6SNIbsRkgPl1qgmVlLimb8y0fS+pK6Ns4DewEvApOAMWmzMcBtaX4SMFrSOpL6AwOBp0t9H8U8Q/AUSQcCu6ek8RFxa6kFmpm1pIbyXSG4GXBruhtnR+CGiLhH0jPAREnHAm8ABwNExHRJE4EZQB1w8prcZrnYKwSfAxZFxAOS1pPUNSIWlVqomVlLKddojYh4HdixifR3ga+tJs84YFw5yi8YnCUdT9ZhvjEwgOzs4x9XVzkzs7ZU5tEabaaYlvPJZJ3gTwFExKs5Q0fMzCpKGbs12lQxwXlpRCxrfAqKpI6swfAQM7OWtDbdMvQRSWcBXSSNAE4Cbm/ZapmZlaZaWs7FDKX7EfA2MA04AbgLOKclK2VmVqpyDaVra3lbzpJqgBciYnvgitapkplZ6epLH71WUfK2nCOiAXhe0hatVB8zszVSzluGtqVi+px7AtMlPQ181JgYEV9vsVqZmZWo0m+iX6xigvN5LV4LM7MyqfQWcbHy3c95XeA7wNZkJwOvioi61qqYmVkpqmW0Rr6W8wRgOfAosC8wCDitNSplZlaqSh+FUax8wXlQROwAIOkq1uDuSmZmrWVtuHx7eeNMRNQ1XiFoZlbJqr7PGdhR0sI0L7IrBBem+YiIbi1eOzOzZqr6PueIKPmpsWZmbWVtaDmbmbU7a9M4ZzOzdsMtZzOzCrQ2jNYwM2t3qv6EoJlZe1Qt3RrF3M/ZzKzdKOf9nCXtI+kVSTMl/bgVqv8Jt5zNrKqUq+UsqQNwKTACmAM8I2lSRMwoSwEFODibWVUpY5/zUGBmRLwOIOkmYCRQHcG5btlcX/edSBobEePbuh5WWXxclFdzYo6kscDYnKTxOd9Fb+DNnHVzgC+seQ2L4z7n1jW28Ca2FvJx0UYiYnxE7JIz5f4n2VSQb7WzjQ7OZmZNmwP0zVnuA8xrrcIdnM3MmvYMMFBSf0mdgdHApNYq3CcEW5f7Fa0pPi4qULpV8inAvUAH4OqImN5a5ataBmybmVUTd2uYmVUgB2czswrk4JxD0oGSQtK2RWx7uqT11qCsoyT9YTXpb0uaKmmGpONLLcPWXAUdEw2SPpeT9qKkfqWWZZXPwXlFhwGPkZ2VLeR0oOQ/xAJujoghwHDgF5I2a6FyrLBKOSbmAGe30L6tAjk4J5I2AHYDjiXnD1FSB0m/kjRN0guSvivpVKAX8LCkh9N2i3PyfFPSNWn+AElPSfqHpAeaE2gjYgHwGrClpK+lfUyTdLWkddL+L0wt7Bck/WrNPwlrVGHHxB3AYEmfbaKee0l6QtJzkv6S6o2k/SS9LOkxSZdIuqP0T8Nam4Pzp0YB90TEP4H3JH0+pY8F+gM7RcTngOsj4hKyweh7RMQeBfb7GDAsInYCbgLOKLZCkrYCtiJrNV0DHBoRO5ANgTxR0sbAgcDgVLefF7tvK8ooKueYaAAuAs7KTZS0KXAOsGdEfB6YAnxf0rrAn4B9I+LLQPciyrAK4uD8qcPI/lBIr4el+T2BP0ZEHUBEvNfM/fYB7pU0DfghMLiIPIdKmgrcCJxA9oc1KwUJgAnA7sBC4GPgSkkHAf9uZt0sv0o6JgBuAIZJ6p+TNgwYBDyejpkxwJbAtsDrETErbXdjM+tobcwXoQCSNgG+CmwvKcgGnIekM8iury9mMHjuNuvmzP8euDgiJkkaDpxbxL5ujohTcuo3pMkCs0HyQ4Gvkf3sPiW9D1tDFXhMNH7fvwZ+lFtV4P6IOCx3W0k7FbNPq1xuOWe+CVwbEVtGRL+I6AvMAr4M3Ad8R1JHgNSVALAI6Jqzj/mStpNUQ9bV0GhDYG6aH1Ni/V4G+knaOi0fATyS+hY3jIi7yE5GDSlx/7aqSj0mriFruTd2UzwJ7NZ4bEhaT9I2ZMfMVjkjOg5tZjnWxhycM4cBt66UdgvwLeBK4A3gBUnPpzTILrm9u/HkD/BjspM2DwG1Ofs5F/iLpEeBd0qpXER8DByd9jONrP/xj2SB4A5JLwCPAN8rZf/WpIo8JiJiGXAJ0CMtvw0cBdyYjoMngW0jYglwEnCPpMeA+cCHzSnL2pYv3zarUpI2iIjFkkT2RI9XI+I3bV0vK45bzmbV6/h0knA6WVfKn9q2OtYcbjmbmVUgt5zNzCqQg7OZWQVycDYzq0AOzmZmFcjB2cysAv1/BqGY27TVXasAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# confusion matrices for our training data\n",
    "plot_confusion_matrix(tree['train_pred'], train_stage2['certified'], model_type = \"decision_tree\")\n",
    "plot_confusion_matrix(rand['train_pred'], train_stage2['certified'], model_type = \"random_forest\")\n",
    "plot_confusion_matrix(ada['train_pred'], train_stage2['certified'], model_type = \"ada_boost\")\n",
    "plot_confusion_matrix(knn['train_pred'], train_stage2['certified'], model_type = \"knn\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import roc_curve\n",
    "\n",
    "def plot_roc_curve_model(actual_data, pred_prob,model_type = \"tree\"):\n",
    "    '''\n",
    "      Parameters:\n",
    "      ===================\n",
    "      first argument = true values, second argument = predicted probabilities\n",
    "      roc_curve returns 3 objects fpr, tpr, thresholds\n",
    "      fpr: false positive rate\n",
    "      tpr: true positive rate\n",
    "    '''\n",
    "\n",
    "    fpr, tpr, thresholds = roc_curve(actual_data.to_numpy(),\n",
    "                                     pred_prob[:, 1])\n",
    "\n",
    "    plt.plot(fpr, tpr)\n",
    "    plt.title('ROC curve for {} classifier'.format(model_type))\n",
    "    plt.xlabel('False Positive Rate')\n",
    "    plt.ylabel('True Positive Rate')\n",
    "    plt.grid(True)\n",
    "    plt.show()\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_roc_curve_model(actual_data=valid_stage2['certified'], pred_prob = tree['validate_pred_prob'], \n",
    "               model_type = \"decision tree\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_roc_curve_model(actual_data=valid_stage2['certified'], pred_prob = rand['validate_pred_prob'], \n",
    "               model_type = \"random forest\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_roc_curve_model(actual_data=valid_stage2['certified'], pred_prob = ada['validate_pred_prob'], \n",
    "               model_type = \"ada boost\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_roc_curve_model(actual_data=valid_stage2['certified'], pred_prob = knn['validate_pred_prob'], \n",
    "               model_type = \"knn\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Summary of Results\n",
    "Here we can see that the performance of the Decision Tree and Random Tree models performed the best on the validation set and thus we concluded with using the Decision Tree Classifier yielded the best results. Using feature selection, it did help improve the random forest classifier to match the decision tree model out of the best performing models prior. \n",
    "\n",
    "To summarize, I improved my initial predictions by imputing the grade variable. I also improved my initial predictions by using feature selection. I tried encoding some variables/feature engineering (see feature engineering selection) but they didn't add a lot of explanatory value to our dataset. Here the ROC curves for the decision tree and random forest models reflect a perfect test almost as it's able to discriminate between the classifed and non-classified with 100 % sensitivity and 100 % specificity as opposed to the knn model which struggles a bit, helping assert that the random forest and decision tree models are very efficient and have a very high true positive rate compared to false positive rate. \n",
    "\n",
    "Given the ethnical implications of the following: \"Their idea is to use the model to help predict which student groups (i.e. from specific countries or certain educational backgrounds) would make them the most money by getting the most participants to complete their paid certificate program. Are there any ethical considerations we should think about before acting on that plan?\"\n",
    "\n",
    "There are a lot of ethical considerations because what we are doing by predicting which student groups is assuming/generalizing an entire group of students. We have to consider if we are blocking access/resources for a given group of students simply because possibly the model we made happen to predict correctly majority of students there didn't get certified. We also have to consider any external factors that may not be taken into consideration (and we can't since every individual is unique/has a unique background) such as if the country the students are from are undergoing some war/conflict, students experiencing difficulties at home, etc. As well, every student's experience is unique based on their own educational background and cannot be used to determine if they will be certified or not. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [],
   "source": [
    "# the selected best model to run testing data on\n",
    "pred = train_and_evaluate_model(model = DecisionTreeClassifier,\n",
    "                                 model_type = \"decision\",\n",
    "                                 model_params = best_params_decision, \n",
    "                                 X_train = train_stage2[best_features], y_train = train_stage2['certified'], \n",
    "                                 X_valid = valid_stage2[best_features], y_valid = valid_stage2['certified'],\n",
    "                                 run_test = True,\n",
    "                                 X_test = test_data_tran[best_features],\n",
    "                                 y_test = pd.DataFrame({\"hi\":[12,33]}), return_pred = True)['test_pred']"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}