Phase2_Group79_Data Cleaning_EDA.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Group 79\n",
    "\n",
    "* Mohamed Thaha Jiffry Ahamed Hibishy (S3851674)\n",
    "* Rooban Kumar Gopi (S3806269)\n",
    "* Biswabal Gurung (S3820568)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Predicting Term Deposit Subscriptions in Bank"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Table of Contents\n",
    "- [Introduction](#intro)\n",
    "  + [Report Overview](#ro)\n",
    "  + [Overview of Methodology](#oom)\n",
    "- [Data Preperation](#dp)\n",
    "  + [Loading Dataset](#ld)\n",
    "  + [Splitting Data and target](#sd)\n",
    "  + [Encoding the Categorical Features](#en)\n",
    "  + [Scaling of Features](#sc)\n",
    "- [Predictive Modelling](#pm)\n",
    "  + [Feature Selection](#fs)\n",
    "  + [Model Fitting & Tuning](#mft)\n",
    "  + [Model Comparision](#mc)\n",
    "- [Critique & Limitations](#cl)\n",
    "- [Summary & Conclusions](#sc)\n",
    "  + [Project Summary](#ps)\n",
    "  + [Summary of Findings](#sof)\n",
    "  + [Conclusions](#c)\n",
    "- [References](#ref)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## INTRODUCTION <a id='intro'></a>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Report Overview <a id='ro'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this report, we are trying to predict whether a bank customer will subscribe for the term deposit or not during the bank marketing campaigns.\n",
    "\n",
    "This report includes overview of our methodology, summary of our data preparation, predictive modelling, model comparisons and a brief summary of our findings and limitations of our project."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Overview of Methodology <a id='oom'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The dataset was cleaned in the Phase 1. For Phase 2, we used the cleaned dataset and did further transformation such as encoding of categorical variables and scaling of the numeric variables. After transformation the number of descriptive features increased from 15 to 47. Our dataset had 11,162 observations so we considered all the observations for the modelling. The descriptive features and the target features were then split from the dataset. The datasets were then split into training dataset and test dataset with 70:30 ratio. The training dataset was used during building and hyper-parameter tuning of the model. The test dataset was used during the model comparison.\n",
    "\n",
    "Before building the models using F-score best 10 features were selected and visualized to comply with the assignment rubrics. However, full set of features were used during model building and hyper-parameter tuning because of the increased accuracy. We can also pipeline feature selection and grid search for hyper-parameter tuning but for this assignment we considered full set of features.\n",
    "\n",
    "We built 5 classifiers K-nearest neighbors (KNN), Decision Tree (DT), Naive Bayes (NB), Random Forest (RF) classifier and Support Vector Classification (SVC) models. The models were then performed hyper-parameter tuning on training dataset using 5-fold cross-validation. The tuned models were then compared based on AUC score in test dataset with 5-fold cross-validation and then performed paired t-tests to select the best model statistically.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## DATA PREPARATION <a id='dp'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Loading Dataset <a id='ld'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Importing some of the required modules, setting up the relevant random state and loading the cleaned dataset from Phase 1."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Suppressing python warnings\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "\n",
    "# importing the required modules\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "# making sure while displaying dataframes it display all the columns\n",
    "pd.set_option('display.max_columns', None) \n",
    "\n",
    "# Set a seed value\n",
    "seed_value = 999 \n",
    "# 1. Initialise `PYTHONHASHSEED` environment variable\n",
    "import os\n",
    "os.environ['PYTHONHASHSEED']=str(seed_value)\n",
    "# 2. Initialise Python's own pseudo-random generator\n",
    "import random\n",
    "random.seed(seed_value)\n",
    "# 3. Initialise Numpy's pseudo-random generator\n",
    "np.random.seed(seed_value)\n",
    "\n",
    "# loading dataset\n",
    "bank = pd.read_csv('Data_Group79_Cleaned.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "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>age</th>\n",
       "      <th>job</th>\n",
       "      <th>marital</th>\n",
       "      <th>education</th>\n",
       "      <th>default</th>\n",
       "      <th>balance</th>\n",
       "      <th>housing</th>\n",
       "      <th>loan</th>\n",
       "      <th>contact</th>\n",
       "      <th>day</th>\n",
       "      <th>month</th>\n",
       "      <th>campaign</th>\n",
       "      <th>pdays</th>\n",
       "      <th>previous</th>\n",
       "      <th>poutcome</th>\n",
       "      <th>deposit</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>59</td>\n",
       "      <td>admin.</td>\n",
       "      <td>married</td>\n",
       "      <td>secondary</td>\n",
       "      <td>no</td>\n",
       "      <td>2343</td>\n",
       "      <td>yes</td>\n",
       "      <td>no</td>\n",
       "      <td>unknown</td>\n",
       "      <td>5</td>\n",
       "      <td>may</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>unknown</td>\n",
       "      <td>yes</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>56</td>\n",
       "      <td>admin.</td>\n",
       "      <td>married</td>\n",
       "      <td>secondary</td>\n",
       "      <td>no</td>\n",
       "      <td>45</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "      <td>unknown</td>\n",
       "      <td>5</td>\n",
       "      <td>may</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>unknown</td>\n",
       "      <td>yes</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>41</td>\n",
       "      <td>technician</td>\n",
       "      <td>married</td>\n",
       "      <td>secondary</td>\n",
       "      <td>no</td>\n",
       "      <td>1270</td>\n",
       "      <td>yes</td>\n",
       "      <td>no</td>\n",
       "      <td>unknown</td>\n",
       "      <td>5</td>\n",
       "      <td>may</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>unknown</td>\n",
       "      <td>yes</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>55</td>\n",
       "      <td>services</td>\n",
       "      <td>married</td>\n",
       "      <td>secondary</td>\n",
       "      <td>no</td>\n",
       "      <td>2476</td>\n",
       "      <td>yes</td>\n",
       "      <td>no</td>\n",
       "      <td>unknown</td>\n",
       "      <td>5</td>\n",
       "      <td>may</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>unknown</td>\n",
       "      <td>yes</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>54</td>\n",
       "      <td>admin.</td>\n",
       "      <td>married</td>\n",
       "      <td>tertiary</td>\n",
       "      <td>no</td>\n",
       "      <td>184</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "      <td>unknown</td>\n",
       "      <td>5</td>\n",
       "      <td>may</td>\n",
       "      <td>2</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>unknown</td>\n",
       "      <td>yes</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   age         job  marital  education default  balance housing loan  contact  \\\n",
       "0   59      admin.  married  secondary      no     2343     yes   no  unknown   \n",
       "1   56      admin.  married  secondary      no       45      no   no  unknown   \n",
       "2   41  technician  married  secondary      no     1270     yes   no  unknown   \n",
       "3   55    services  married  secondary      no     2476     yes   no  unknown   \n",
       "4   54      admin.  married   tertiary      no      184      no   no  unknown   \n",
       "\n",
       "   day month  campaign  pdays  previous poutcome deposit  \n",
       "0    5   may         1     -1         0  unknown     yes  \n",
       "1    5   may         1     -1         0  unknown     yes  \n",
       "2    5   may         1     -1         0  unknown     yes  \n",
       "3    5   may         1     -1         0  unknown     yes  \n",
       "4    5   may         2     -1         0  unknown     yes  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# diplaying first 5 observations\n",
    "bank.head(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "age           int64\n",
       "job          object\n",
       "marital      object\n",
       "education    object\n",
       "default      object\n",
       "balance       int64\n",
       "housing      object\n",
       "loan         object\n",
       "contact      object\n",
       "day           int64\n",
       "month        object\n",
       "campaign      int64\n",
       "pdays         int64\n",
       "previous      int64\n",
       "poutcome     object\n",
       "deposit      object\n",
       "dtype: object"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# cheking data types\n",
    "bank.dtypes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Splitting Data and target <a id='sd'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Splitting data into Data and target."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# storing descriptive features to Data\n",
    "Data = bank.drop(columns='deposit')\n",
    "\n",
    "# storing target feature to target\n",
    "target = bank['deposit']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Encoding the Categorical Features <a id='en'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "All the categorical features irrespective of descriptive or target needs to be converted to numeric. For this we perform one-hot encoding."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Encoding the target Feature <a id='ld'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For the target Feature the positive class is yes, so encoding 'yes' as 1 and 'no' as 0."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "no     5873\n",
       "yes    5289\n",
       "Name: deposit, dtype: int64"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# looking at counts of target variables\n",
    "target.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    5873\n",
       "1    5289\n",
       "Name: deposit, dtype: int64"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# encoding 'yes' as 1 and 'no' as 0\n",
    "target = target.replace({'no': 0, 'yes': 1})\n",
    "\n",
    "target.value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Encoding Categorical Descriptive Features<a id='ecf'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Below, we make a list of columns with the data type 'objects'. Then we run a for loop to check if the feature has only two levels which we will make them binary. For the others we use regular one hot encoding."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# get the list of categorical descriptive features\n",
    "categorical_cols = Data.columns[Data.dtypes==object].tolist()\n",
    "\n",
    "# if a categorical descriptive feature has only 2 levels,\n",
    "# define only one binary variable\n",
    "for col in categorical_cols:\n",
    "    n = len(Data[col].unique())\n",
    "    if (n == 2):\n",
    "        Data[col] = pd.get_dummies(Data[col], drop_first=True)\n",
    "\n",
    "# for other categorical features (with > 2 levels), \n",
    "# use regular one-hot-encoding \n",
    "Data = pd.get_dummies(Data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "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>age</th>\n",
       "      <th>default</th>\n",
       "      <th>balance</th>\n",
       "      <th>housing</th>\n",
       "      <th>loan</th>\n",
       "      <th>day</th>\n",
       "      <th>campaign</th>\n",
       "      <th>pdays</th>\n",
       "      <th>previous</th>\n",
       "      <th>job_admin.</th>\n",
       "      <th>job_blue-collar</th>\n",
       "      <th>job_entrepreneur</th>\n",
       "      <th>job_housemaid</th>\n",
       "      <th>job_management</th>\n",
       "      <th>job_retired</th>\n",
       "      <th>job_self-employed</th>\n",
       "      <th>job_services</th>\n",
       "      <th>job_student</th>\n",
       "      <th>job_technician</th>\n",
       "      <th>job_unemployed</th>\n",
       "      <th>job_unknown</th>\n",
       "      <th>marital_divorced</th>\n",
       "      <th>marital_married</th>\n",
       "      <th>marital_single</th>\n",
       "      <th>education_primary</th>\n",
       "      <th>education_secondary</th>\n",
       "      <th>education_tertiary</th>\n",
       "      <th>education_unknown</th>\n",
       "      <th>contact_cellular</th>\n",
       "      <th>contact_telephone</th>\n",
       "      <th>contact_unknown</th>\n",
       "      <th>month_apr</th>\n",
       "      <th>month_aug</th>\n",
       "      <th>month_dec</th>\n",
       "      <th>month_feb</th>\n",
       "      <th>month_jan</th>\n",
       "      <th>month_jul</th>\n",
       "      <th>month_jun</th>\n",
       "      <th>month_mar</th>\n",
       "      <th>month_may</th>\n",
       "      <th>month_nov</th>\n",
       "      <th>month_oct</th>\n",
       "      <th>month_sep</th>\n",
       "      <th>poutcome_failure</th>\n",
       "      <th>poutcome_other</th>\n",
       "      <th>poutcome_success</th>\n",
       "      <th>poutcome_unknown</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>59</td>\n",
       "      <td>0</td>\n",
       "      <td>2343</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>56</td>\n",
       "      <td>0</td>\n",
       "      <td>45</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>41</td>\n",
       "      <td>0</td>\n",
       "      <td>1270</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>55</td>\n",
       "      <td>0</td>\n",
       "      <td>2476</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>54</td>\n",
       "      <td>0</td>\n",
       "      <td>184</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   age  default  balance  housing  loan  day  campaign  pdays  previous  \\\n",
       "0   59        0     2343        1     0    5         1     -1         0   \n",
       "1   56        0       45        0     0    5         1     -1         0   \n",
       "2   41        0     1270        1     0    5         1     -1         0   \n",
       "3   55        0     2476        1     0    5         1     -1         0   \n",
       "4   54        0      184        0     0    5         2     -1         0   \n",
       "\n",
       "   job_admin.  job_blue-collar  job_entrepreneur  job_housemaid  \\\n",
       "0           1                0                 0              0   \n",
       "1           1                0                 0              0   \n",
       "2           0                0                 0              0   \n",
       "3           0                0                 0              0   \n",
       "4           1                0                 0              0   \n",
       "\n",
       "   job_management  job_retired  job_self-employed  job_services  job_student  \\\n",
       "0               0            0                  0             0            0   \n",
       "1               0            0                  0             0            0   \n",
       "2               0            0                  0             0            0   \n",
       "3               0            0                  0             1            0   \n",
       "4               0            0                  0             0            0   \n",
       "\n",
       "   job_technician  job_unemployed  job_unknown  marital_divorced  \\\n",
       "0               0               0            0                 0   \n",
       "1               0               0            0                 0   \n",
       "2               1               0            0                 0   \n",
       "3               0               0            0                 0   \n",
       "4               0               0            0                 0   \n",
       "\n",
       "   marital_married  marital_single  education_primary  education_secondary  \\\n",
       "0                1               0                  0                    1   \n",
       "1                1               0                  0                    1   \n",
       "2                1               0                  0                    1   \n",
       "3                1               0                  0                    1   \n",
       "4                1               0                  0                    0   \n",
       "\n",
       "   education_tertiary  education_unknown  contact_cellular  contact_telephone  \\\n",
       "0                   0                  0                 0                  0   \n",
       "1                   0                  0                 0                  0   \n",
       "2                   0                  0                 0                  0   \n",
       "3                   0                  0                 0                  0   \n",
       "4                   1                  0                 0                  0   \n",
       "\n",
       "   contact_unknown  month_apr  month_aug  month_dec  month_feb  month_jan  \\\n",
       "0                1          0          0          0          0          0   \n",
       "1                1          0          0          0          0          0   \n",
       "2                1          0          0          0          0          0   \n",
       "3                1          0          0          0          0          0   \n",
       "4                1          0          0          0          0          0   \n",
       "\n",
       "   month_jul  month_jun  month_mar  month_may  month_nov  month_oct  \\\n",
       "0          0          0          0          1          0          0   \n",
       "1          0          0          0          1          0          0   \n",
       "2          0          0          0          1          0          0   \n",
       "3          0          0          0          1          0          0   \n",
       "4          0          0          0          1          0          0   \n",
       "\n",
       "   month_sep  poutcome_failure  poutcome_other  poutcome_success  \\\n",
       "0          0                 0               0                 0   \n",
       "1          0                 0               0                 0   \n",
       "2          0                 0               0                 0   \n",
       "3          0                 0               0                 0   \n",
       "4          0                 0               0                 0   \n",
       "\n",
       "   poutcome_unknown  \n",
       "0                 1  \n",
       "1                 1  \n",
       "2                 1  \n",
       "3                 1  \n",
       "4                 1  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# displaying the first 5 observations after one-hot-encoding \n",
    "Data.head(5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Scaling of Features<a id='sc'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Scaling 'age' and 'balance' numeric variables before modeling."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "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>age</th>\n",
       "      <th>default</th>\n",
       "      <th>balance</th>\n",
       "      <th>housing</th>\n",
       "      <th>loan</th>\n",
       "      <th>day</th>\n",
       "      <th>campaign</th>\n",
       "      <th>pdays</th>\n",
       "      <th>previous</th>\n",
       "      <th>job_admin.</th>\n",
       "      <th>job_blue-collar</th>\n",
       "      <th>job_entrepreneur</th>\n",
       "      <th>job_housemaid</th>\n",
       "      <th>job_management</th>\n",
       "      <th>job_retired</th>\n",
       "      <th>job_self-employed</th>\n",
       "      <th>job_services</th>\n",
       "      <th>job_student</th>\n",
       "      <th>job_technician</th>\n",
       "      <th>job_unemployed</th>\n",
       "      <th>job_unknown</th>\n",
       "      <th>marital_divorced</th>\n",
       "      <th>marital_married</th>\n",
       "      <th>marital_single</th>\n",
       "      <th>education_primary</th>\n",
       "      <th>education_secondary</th>\n",
       "      <th>education_tertiary</th>\n",
       "      <th>education_unknown</th>\n",
       "      <th>contact_cellular</th>\n",
       "      <th>contact_telephone</th>\n",
       "      <th>contact_unknown</th>\n",
       "      <th>month_apr</th>\n",
       "      <th>month_aug</th>\n",
       "      <th>month_dec</th>\n",
       "      <th>month_feb</th>\n",
       "      <th>month_jan</th>\n",
       "      <th>month_jul</th>\n",
       "      <th>month_jun</th>\n",
       "      <th>month_mar</th>\n",
       "      <th>month_may</th>\n",
       "      <th>month_nov</th>\n",
       "      <th>month_oct</th>\n",
       "      <th>month_sep</th>\n",
       "      <th>poutcome_failure</th>\n",
       "      <th>poutcome_other</th>\n",
       "      <th>poutcome_success</th>\n",
       "      <th>poutcome_unknown</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.532</td>\n",
       "      <td>0</td>\n",
       "      <td>0.104</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.494</td>\n",
       "      <td>0</td>\n",
       "      <td>0.078</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.299</td>\n",
       "      <td>0</td>\n",
       "      <td>0.092</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.481</td>\n",
       "      <td>0</td>\n",
       "      <td>0.106</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.468</td>\n",
       "      <td>0</td>\n",
       "      <td>0.080</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     age  default  balance  housing  loan  day  campaign  pdays  previous  \\\n",
       "0  0.532        0    0.104        1     0    5         1     -1         0   \n",
       "1  0.494        0    0.078        0     0    5         1     -1         0   \n",
       "2  0.299        0    0.092        1     0    5         1     -1         0   \n",
       "3  0.481        0    0.106        1     0    5         1     -1         0   \n",
       "4  0.468        0    0.080        0     0    5         2     -1         0   \n",
       "\n",
       "   job_admin.  job_blue-collar  job_entrepreneur  job_housemaid  \\\n",
       "0           1                0                 0              0   \n",
       "1           1                0                 0              0   \n",
       "2           0                0                 0              0   \n",
       "3           0                0                 0              0   \n",
       "4           1                0                 0              0   \n",
       "\n",
       "   job_management  job_retired  job_self-employed  job_services  job_student  \\\n",
       "0               0            0                  0             0            0   \n",
       "1               0            0                  0             0            0   \n",
       "2               0            0                  0             0            0   \n",
       "3               0            0                  0             1            0   \n",
       "4               0            0                  0             0            0   \n",
       "\n",
       "   job_technician  job_unemployed  job_unknown  marital_divorced  \\\n",
       "0               0               0            0                 0   \n",
       "1               0               0            0                 0   \n",
       "2               1               0            0                 0   \n",
       "3               0               0            0                 0   \n",
       "4               0               0            0                 0   \n",
       "\n",
       "   marital_married  marital_single  education_primary  education_secondary  \\\n",
       "0                1               0                  0                    1   \n",
       "1                1               0                  0                    1   \n",
       "2                1               0                  0                    1   \n",
       "3                1               0                  0                    1   \n",
       "4                1               0                  0                    0   \n",
       "\n",
       "   education_tertiary  education_unknown  contact_cellular  contact_telephone  \\\n",
       "0                   0                  0                 0                  0   \n",
       "1                   0                  0                 0                  0   \n",
       "2                   0                  0                 0                  0   \n",
       "3                   0                  0                 0                  0   \n",
       "4                   1                  0                 0                  0   \n",
       "\n",
       "   contact_unknown  month_apr  month_aug  month_dec  month_feb  month_jan  \\\n",
       "0                1          0          0          0          0          0   \n",
       "1                1          0          0          0          0          0   \n",
       "2                1          0          0          0          0          0   \n",
       "3                1          0          0          0          0          0   \n",
       "4                1          0          0          0          0          0   \n",
       "\n",
       "   month_jul  month_jun  month_mar  month_may  month_nov  month_oct  \\\n",
       "0          0          0          0          1          0          0   \n",
       "1          0          0          0          1          0          0   \n",
       "2          0          0          0          1          0          0   \n",
       "3          0          0          0          1          0          0   \n",
       "4          0          0          0          1          0          0   \n",
       "\n",
       "   month_sep  poutcome_failure  poutcome_other  poutcome_success  \\\n",
       "0          0                 0               0                 0   \n",
       "1          0                 0               0                 0   \n",
       "2          0                 0               0                 0   \n",
       "3          0                 0               0                 0   \n",
       "4          0                 0               0                 0   \n",
       "\n",
       "   poutcome_unknown  \n",
       "0                 1  \n",
       "1                 1  \n",
       "2                 1  \n",
       "3                 1  \n",
       "4                 1  "
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Scaling the numeric descriptive feature\n",
    "from sklearn import preprocessing\n",
    "\n",
    "Data['age'] = preprocessing.MinMaxScaler().fit_transform(Data['age'].values.reshape(-1,1))\n",
    "Data['balance'] = preprocessing.MinMaxScaler().fit_transform(Data['balance'].values.reshape(-1,1))\n",
    "\n",
    "# display first 5 observations\n",
    "Data.head(5).round(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## PREDICTIVE MODELLING <a id='pm'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature Selection <a id='fs'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this section, we are trying to look at the best 10 features based on F-Score and visualize them. However, for the model building and hyper-parameter tuning we will use the full set of features because of the increased accuracy."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['age', 'default', 'balance', 'housing', 'loan', 'day', 'campaign',\n",
       "       'pdays', 'previous', 'job_admin.', 'job_blue-collar',\n",
       "       'job_entrepreneur', 'job_housemaid', 'job_management', 'job_retired',\n",
       "       'job_self-employed', 'job_services', 'job_student', 'job_technician',\n",
       "       'job_unemployed', 'job_unknown', 'marital_divorced', 'marital_married',\n",
       "       'marital_single', 'education_primary', 'education_secondary',\n",
       "       'education_tertiary', 'education_unknown', 'contact_cellular',\n",
       "       'contact_telephone', 'contact_unknown', 'month_apr', 'month_aug',\n",
       "       'month_dec', 'month_feb', 'month_jan', 'month_jul', 'month_jun',\n",
       "       'month_mar', 'month_may', 'month_nov', 'month_oct', 'month_sep',\n",
       "       'poutcome_failure', 'poutcome_other', 'poutcome_success',\n",
       "       'poutcome_unknown'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Display all features\n",
    "Data.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([45, 30, 46, 28,  3, 39,  7,  8, 38, 41], dtype=int64)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# selects best 10 features (code from Feature Ranking website)\n",
    "from sklearn import feature_selection as fs\n",
    "\n",
    "num_features = 10\n",
    "\n",
    "fs_fit_fscore = fs.SelectKBest(fs.f_classif, k=num_features)\n",
    "fs_fit_fscore.fit_transform(Data, target)\n",
    "fs_indices_fscore = np.argsort(np.nan_to_num(fs_fit_fscore.scores_))[::-1][0:num_features]\n",
    "\n",
    "# display selected features\n",
    "fs_indices_fscore"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['poutcome_success', 'contact_unknown', 'poutcome_unknown',\n",
       "       'contact_cellular', 'housing', 'month_may', 'pdays', 'previous',\n",
       "       'month_mar', 'month_oct'], dtype=object)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# map the above selected feature number with the actual feature names\n",
    "best_features_fscore = Data.columns[fs_indices_fscore].values\n",
    "\n",
    "# display selected features name\n",
    "best_features_fscore"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 281,
       "width": 473
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# store selected features score\n",
    "feature_importances_fscore = fs_fit_fscore.scores_[fs_indices_fscore]\n",
    "\n",
    "# plot the feature score in a graph to see the selected feature importance (code from Feature Ranking website)\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline \n",
    "%config InlineBackend.figure_format = 'retina'\n",
    "plt.style.use(\"ggplot\")\n",
    "\n",
    "def plot_imp(best_features, scores, method_name):   \n",
    "    plt.barh(best_features, scores)\n",
    "    plt.title(method_name + ' Feature Importances')\n",
    "    plt.xlabel(\"Importance\")\n",
    "    plt.ylabel(\"Features\")\n",
    "    plt.show()\n",
    "    \n",
    "\n",
    "plot_imp(best_features_fscore, feature_importances_fscore, 'F-Score')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Based on F-Score feature selection technique 'poutcome_success' is the most important feature and 'month_oct' is the least important feature out of the selected 10 features."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Model Fitting & Tuning <a id='mft'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Train-Test Splitting"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Splitting the dataset into train and test with 70:30 ratio before fitting and tuning the model. The 70% of the dataset is used to train and tune the model, the remaining 30% is used during the comparison of the 5 models which we are going to implement below."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Saving Data to X and target to y (following naming convention)\n",
    "# Storing it to numpy array\n",
    "X = Data.values\n",
    "y = target.values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "# splitting train and test\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.3, random_state=seed_value)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(7813, 47)\n",
      "(3349, 47)\n",
      "(7813,)\n",
      "(3349,)\n"
     ]
    }
   ],
   "source": [
    "# checking train and test size\n",
    "print(X_train.shape)\n",
    "print(X_test.shape)\n",
    "print(y_train.shape)\n",
    "print(y_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### K-Nearest Neighbors"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For tuning each of the model below including KNN we are going to use 5-fold stratified cross-validation. As we are using all the features for fitting & tuning the model we haven't implemented pipeline in our approach. For measuring prediction accuracy of all the below models we considered AUC score.\n",
    "\n",
    "We implemented grid search for KNN hyper-parameter tuning along with 5-fold stratified cross-validation. We considered following hyper-parameters for KNN:\n",
    "* n_neighbours\n",
    "* p"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import StratifiedKFold, GridSearchCV\n",
    "\n",
    "# defining our 5-fold cross-validation technique\n",
    "cv_method = StratifiedKFold(n_splits=5, random_state=seed_value)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "\n",
    "\n",
    "params_KNN = {'n_neighbors': [1, 5, 10, 15, 20, 25, 30],\n",
    "                   'p': [1, 2]}\n",
    "\n",
    "gs_KNN = GridSearchCV(estimator=KNeighborsClassifier(), \n",
    "                           param_grid=params_KNN, \n",
    "                           cv=cv_method,\n",
    "                           refit=True,\n",
    "                           n_jobs=-2,\n",
    "                           scoring='roc_auc',\n",
    "                           verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 14 candidates, totalling 70 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-2)]: Using backend LokyBackend with 3 concurrent workers.\n",
      "[Parallel(n_jobs=-2)]: Done  44 tasks      | elapsed:   11.0s\n",
      "[Parallel(n_jobs=-2)]: Done  70 out of  70 | elapsed:   16.2s finished\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=StratifiedKFold(n_splits=5, random_state=999, shuffle=False),\n",
       "             error_score='raise-deprecating',\n",
       "             estimator=KNeighborsClassifier(algorithm='auto', leaf_size=30,\n",
       "                                            metric='minkowski',\n",
       "                                            metric_params=None, n_jobs=None,\n",
       "                                            n_neighbors=5, p=2,\n",
       "                                            weights='uniform'),\n",
       "             iid='warn', n_jobs=-2,\n",
       "             param_grid={'n_neighbors': [1, 5, 10, 15, 20, 25, 30],\n",
       "                         'p': [1, 2]},\n",
       "             pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n",
       "             scoring='roc_auc', verbose=1)"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# fitting the model with train dataset\n",
    "gs_KNN.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'n_neighbors': 25, 'p': 1}"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# display the best parameters\n",
    "gs_KNN.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7526711361429341"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# display the AUC score of the KNN model\n",
    "gs_KNN.best_score_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The KNN model with n_neighbors = 25 and p = 1 (Manhattan Distance) is the model which gives the AUC score of 0.753.\n",
    "\n",
    "Now using the code from the Feature Ranking website we will try to visualize the performance of KNN model with various different combination of KNN hyper-parameters."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# custom function to format the search results as a Pandas data frame (code from Feature Ranking website)\n",
    "def get_search_results(gs):\n",
    "\n",
    "    def model_result(scores, params):\n",
    "        scores = {'mean_score': np.mean(scores),\n",
    "             'std_score': np.std(scores),\n",
    "             'min_score': np.min(scores),\n",
    "             'max_score': np.max(scores)}\n",
    "        return pd.Series({**params,**scores})\n",
    "\n",
    "    models = []\n",
    "    scores = []\n",
    "\n",
    "    for i in range(gs.n_splits_):\n",
    "        key = f\"split{i}_test_score\"\n",
    "        r = gs.cv_results_[key]        \n",
    "        scores.append(r.reshape(-1,1))\n",
    "\n",
    "    all_scores = np.hstack(scores)\n",
    "    for p, s in zip(gs.cv_results_['params'], all_scores):\n",
    "        models.append((model_result(s, p)))\n",
    "\n",
    "    pipe_results = pd.concat(models, axis=1).T.sort_values(['mean_score'], ascending=False)\n",
    "\n",
    "    columns_first = ['mean_score', 'std_score', 'max_score', 'min_score']\n",
    "    columns = columns_first + [c for c in pipe_results.columns if c not in columns_first]\n",
    "\n",
    "    return pipe_results[columns]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "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>mean_score</th>\n",
       "      <th>std_score</th>\n",
       "      <th>max_score</th>\n",
       "      <th>min_score</th>\n",
       "      <th>n_neighbors</th>\n",
       "      <th>p</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.752670</td>\n",
       "      <td>0.008228</td>\n",
       "      <td>0.763644</td>\n",
       "      <td>0.742223</td>\n",
       "      <td>25.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.752429</td>\n",
       "      <td>0.007755</td>\n",
       "      <td>0.761862</td>\n",
       "      <td>0.743117</td>\n",
       "      <td>30.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.751235</td>\n",
       "      <td>0.007175</td>\n",
       "      <td>0.759547</td>\n",
       "      <td>0.740551</td>\n",
       "      <td>20.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.747135</td>\n",
       "      <td>0.006664</td>\n",
       "      <td>0.756757</td>\n",
       "      <td>0.737468</td>\n",
       "      <td>15.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.740962</td>\n",
       "      <td>0.005551</td>\n",
       "      <td>0.751352</td>\n",
       "      <td>0.735329</td>\n",
       "      <td>30.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    mean_score  std_score  max_score  min_score  n_neighbors    p\n",
       "10    0.752670   0.008228   0.763644   0.742223         25.0  1.0\n",
       "12    0.752429   0.007755   0.761862   0.743117         30.0  1.0\n",
       "8     0.751235   0.007175   0.759547   0.740551         20.0  1.0\n",
       "6     0.747135   0.006664   0.756757   0.737468         15.0  1.0\n",
       "13    0.740962   0.005551   0.751352   0.735329         30.0  2.0"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Convert grid search results to dataframe using above function\n",
    "results_KNN = get_search_results(gs_KNN)\n",
    "\n",
    "# display the dataframe\n",
    "results_KNN.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 281,
       "width": 394
      },
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot a graph to display KNN model AUC score based on various combination of hyper-parameters \n",
    "# (code modified from feature ranking website)\n",
    "\n",
    "for i in results_KNN['p'].unique():\n",
    "    temp = results_KNN[results_KNN['p'] == i]\n",
    "    plt.plot(temp['n_neighbors'], temp['mean_score'], marker = '.', label = i)\n",
    "    \n",
    "plt.legend(title = \"p\")\n",
    "plt.xlabel('Number of Neighbors')\n",
    "plt.ylabel(\"AUC Score\")\n",
    "plt.title(\"KNN Performance Comparison\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There seems to be an increase in AUC score of KNN model on increasing the k-neighbors using Manhattan distance than using Euclidean distance."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Decision Tree"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In order to build decision tree we considered gini, we implemented grid search for decision tree hyper-parameter tuning along with 5-fold stratified cross-validation. We considered following hyper-parameters:\n",
    "* max_depth\n",
    "* min_samples_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 6 candidates, totalling 30 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-2)]: Using backend LokyBackend with 3 concurrent workers.\n",
      "[Parallel(n_jobs=-2)]: Done  30 out of  30 | elapsed:    0.5s finished\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=StratifiedKFold(n_splits=5, random_state=999, shuffle=False),\n",
       "             error_score='raise-deprecating',\n",
       "             estimator=DecisionTreeClassifier(class_weight=None,\n",
       "                                              criterion='gini', max_depth=None,\n",
       "                                              max_features=None,\n",
       "                                              max_leaf_nodes=None,\n",
       "                                              min_impurity_decrease=0.0,\n",
       "                                              min_impurity_split=None,\n",
       "                                              min_samples_leaf=1,\n",
       "                                              min_samples_split=2,\n",
       "                                              min_weight_fraction_leaf=0.0,\n",
       "                                              presort=False, random_state=999,\n",
       "                                              splitter='best'),\n",
       "             iid='warn', n_jobs=-2,\n",
       "             param_grid={'max_depth': [3, 4, 5], 'min_samples_split': [2, 5]},\n",
       "             pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n",
       "             scoring='roc_auc', verbose=1)"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.tree import DecisionTreeClassifier\n",
    "\n",
    "\n",
    "params_DT = {'max_depth': [3, 4, 5],\n",
    "                  'min_samples_split': [2, 5]}\n",
    "\n",
    "gs_DT = GridSearchCV(estimator=DecisionTreeClassifier(criterion='gini', random_state=seed_value), \n",
    "                          param_grid=params_DT, \n",
    "                          cv=cv_method,\n",
    "                          refit=True,\n",
    "                          n_jobs=-2,\n",
    "                          scoring='roc_auc',\n",
    "                          verbose=1) \n",
    "\n",
    "gs_DT.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'max_depth': 5, 'min_samples_split': 5}"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# display the best parameters\n",
    "gs_DT.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7202096974886911"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# display the AUC score of the Decision tree model\n",
    "gs_DT.best_score_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The Decision tree model with max_depth = 5 and min_samples_split = 5 is the best DT model with AUC score of 0.720.\n",
    "\n",
    "Visualizing the grid search results of Decision tree model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "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>mean_score</th>\n",
       "      <th>std_score</th>\n",
       "      <th>max_score</th>\n",
       "      <th>min_score</th>\n",
       "      <th>max_depth</th>\n",
       "      <th>min_samples_split</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.720205</td>\n",
       "      <td>0.015545</td>\n",
       "      <td>0.745205</td>\n",
       "      <td>0.698887</td>\n",
       "      <td>5.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.719737</td>\n",
       "      <td>0.015222</td>\n",
       "      <td>0.743990</td>\n",
       "      <td>0.698887</td>\n",
       "      <td>5.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.710452</td>\n",
       "      <td>0.011274</td>\n",
       "      <td>0.728010</td>\n",
       "      <td>0.695276</td>\n",
       "      <td>4.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.710363</td>\n",
       "      <td>0.011136</td>\n",
       "      <td>0.727565</td>\n",
       "      <td>0.695276</td>\n",
       "      <td>4.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.698449</td>\n",
       "      <td>0.018512</td>\n",
       "      <td>0.734054</td>\n",
       "      <td>0.681929</td>\n",
       "      <td>3.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   mean_score  std_score  max_score  min_score  max_depth  min_samples_split\n",
       "5    0.720205   0.015545   0.745205   0.698887        5.0                5.0\n",
       "4    0.719737   0.015222   0.743990   0.698887        5.0                2.0\n",
       "2    0.710452   0.011274   0.728010   0.695276        4.0                2.0\n",
       "3    0.710363   0.011136   0.727565   0.695276        4.0                5.0\n",
       "0    0.698449   0.018512   0.734054   0.681929        3.0                2.0"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Convert grid search results to dataframe using above function\n",
    "results_DT = get_search_results(gs_DT)\n",
    "\n",
    "# display the dataframe\n",
    "results_DT.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAyIAAAIyCAYAAAA37H5xAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAAWJQAAFiUBSVIk8AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nOzdd5hkVZ3w8e/pGRgGJM4gkhRFEXGNIOCCMoKL6BpWZY++qyhmBDPqGjBg1jWCEQOCAT0mMAIrCAbEgMIKgiICAgPCADODME7ovu8f59b07epK3V1zu6rn+3meeircc885Vbeq+/7uSaEoCiRJkiSpTiOzXQFJkiRJGx8DEUmSJEm1MxCRJEmSVDsDEUmSJEm1MxCRJEmSVDsDEUmSJEm1MxCRJEmSVDsDEUmSJEm1MxCRJEmSVDsDEUmSJEm1MxCRJEmSVDsDEUmSJEm1MxCRJEmSVDsDEUnagEIIu4QQPh5C+HMIYVUIoShvx8123STVI4Tw2Mpv/9mzXR9pUMyf7QpI2jBCCLsBV7fZfCewArgNuBT4HfD9oigu75DfNcC9Zlita4ui2K3XxCGEJcBP2mxeQ34PVwDnAJ8riuKGGdavr0II9wR+A9x9tuuiuSOEsAPwDOAQ4IHAYmBz8u/havJ37nTgnKIoxmarnpLUjYGItHHaorztBPwL8EzgAyGEXwBvKorip7NZuR5tCmxf3h4FvC6E8NKiKL40u9Wa4C2MByEfAL4P3F4+//us1EhDK4SwAHgn8DJgYYski8vbI4CjgatDCMcVRfHV+mopSb0LRVHMdh0kbQBNLSK/BZ5X2bwJsC2wC7Af8HRgh3LbGPDeoigmdB0KIexBPvlv5WjgpeXj44Az2qRbUxTFn6fwHpYw3iJyRpl3w6bAfYEXAY+t1P1xRVH8uNcyNqQQwt+AXYFfFUWx/2zXR8OrbAU5g/x7BVgNfBP4X+Aa4A5yELIn8O/k38QIcGtRFIvrrq8k9cIWEWnjcGdRFJe22XZqCOHV5GDi/eQT/DeHEFYWRfGBRqJOAUQI4ebK0xs6lDUTy1vk+zsghRA+Qa7/CPBuYCACEXKLE8CfZrUWGmohhE2ZGIScDzynKIq/tUh+NnBCCGEv8u/5kfXUUpKmzsHqkiiKYk1RFB8FnkJuVQB4Twjh/rNYral4A/kKMcAjQgjbz2ZlAEII84F55dO1s1kXDb23MzEIObRNELJeURR/LIriScDrNnDdJGnaDEQkrVcUxZnAJ8qn84A3z2J1elYUxR3AZeXTANyzVboQwpYhhGNDCOeGEG4KIawJIdwaQvhFCOGNIYSt25URQnhhZdabA0P27BDCWSGEpSGEdSGE34YQ3hVCKJgYfLygsm8RQvhLmzKeEEL4Wgjh2hDCP0MIK0IIl4YQPhZCuF+nzyCE8PNq3iGEu4cQ3hlCuCSEcHu57WWV9NeXr/24fL5bCOGEEMJfQgh3hRBuCCF8K4Tw8KZyFoUQ3hpC+L8Qwh0hhOXl5/mELvWbH0I4NITwobKuN4cQ1pZ5/CmE8IUQwn5d8rhv86xjIYSHhhBODiFcE0JYHUJYFkL4Ubf6NOX7+BDCKSGEK8v63FnOcvbDEMJLQwj36LDvSAjh/5Wf1d9CnhltZQjhDyGEj4YQ7tNrPdrkvx3w8vLpanJLyJpe9y+K4uQOeW8fQjg+hPDr8newJoRwY/m+jwwhzOuw76RZoEIITwwh/KDMY1UI4YryO7h10777hhC+GkK4uvyeXx9C+GwIYZcplvcf5bG+scznmjKfbr+Vzcp9P1F572vL7/IfQp7lbq8uebR7/98JIVxX5resU/oWeW4RQnhNCOG88vexJuS/AX8JIZxffpYdW7hCCAeUv4eryu/xP8rv8mdD02+5xb5fLuu3rny+aQjhFSGEX4X8N+SuEMLlIYT3hxAWdcpL6llRFN68eZuDN2A3oChv501xv7Fyv5XAvB72eXulrCP7+B6WVPL9Ype0F1bS7tdi+6HALZU0rW43AQe0yf+FlXSPBX7UYv/fAu/qUkYB/KUp77sB3+uyz1rg2A7v/+eNvMlXz29qkcfLKumvL1/7MXn2peVtyl0NPLHc50HAtR3q+PIO9ft4D59LAXyEcvxiizzuW0l3HPAS8uxp7fJ6W5fvzA7AeT3U6Ztt9r83cHGXfdcAL53Bb+DoSl5f6uNv60nk33enul8C7Npm/8dW0j0bOLFLPovK/V4DjLZJdxNw3x7L+2yH8lYBz+jw3r/fwzEfBV7TIY9qfY4AvtAij2Xt6t8ivz3IY3261euKNvWZD5zUZd8x4EPASJs8vlymW0eeZOO3HfK6Brhnv76P3jbem2NEJE1QFMU1IYQ/kQe9bgk8jPwPaWCFEDYBHlB56Yam7YeRT/Tnk0+4P0We4vQ68ns8BHgF+cT0RyGEfYuiuKJDkR8EHgKcST4B+SuwDflk4lvA18gtSheX6b8NvK2yf6MbGSGEUG7/t/KlK8gnC5cAm5FPYF5DDlY+GEJYUxTFiR3qtlWZ31bA/5R1XEE+ib+pRfpdgG+Qp3R+OzmgC8ATyF3eNgVOKa+m/rCsx9uAc4F/koPFt1Xqd2ZRFFe2KGc+Ofj5blnGX8knjDsCDyXPBHUP4FXA38gBSSePB/Ynj7/5GPmzDsDBwBsb9Qwh/Lgoil8071y2NFwANFos/kA+kbu4/Cx2APYFntaq8BDCzsAvy3TrgK+SP+ury3o8Anhlmf8nQx5z9ZUu76mVgyuPvzeN/ScJeRKI75C/o6PA58jfmVvJ35OXAgcBDwbOCyE8vCiKFR2yPIZ8LM4mf4ZXk4/lq8nf3wcD7wshnEX+bv+WfMyuALYmB/nPJH+WJzHxPbfycvKxuRj4MPDHMp+nAkeRfzdfCSEsLYriZy32nw/8mfx5/pZ8Ur2W8ck7ji7z+1AI4ZqiKL7dpT7Hkv8eXAh8EricPJ3yvl32q/oK49Ojf538d+QG8m9ke/JFgEOBndvs/xng+eXjG8i//V+Rv4sHAK8v83lNpc7tBPL34yFlvqcDN5M/n1cBjynr+lngcb2/RamF2Y6EvHnztmFuTLNFpNz3y5V9n9VD+rdX0h/Zx/ewpJLvFzukO7aS7k9N27YBlpXbfgps0yaP+zPeYnJmi+3VFpECeF+Xus+vpP1ch3TVfM8DFrZI8y/kaX8L8sn/pKvUjLeIFMA/gL271O/6SvorgR1apHl1Jc3N5ee4R4t0T6uk+0Cb8nanQ+sa+cTtp2UetwFbtEhTbREpyCf+m7VId2glzdfalPe1SppP0uYqcZl20pVfcktSAdwIPLDNfncjBytF+dndbRq/gasq9bxPH35T8xm/8j5K2drVlCYwscXhEy3SVK/wF8AJbcq6iPGWodvJJ7jzW6Q9o5LXXj2UdxawSYt0T2K8xeVyWrSutfoON23fvty3IAdLrfJors9XunyH2raIAPerbPtgl7ot6pL3ZW3S7Fw57mPAvi3SVP/urwX+rc0x/Wkl3Z4z/U5627hvjhGR1MqyyuOB7AscQtgkhLBXCOHD5DU6Gt7blPSl5PewFvivoiiWt8qvKIo/kWfcAnhcyIsRtvNn+jd+5lXl/RrgiKIoVrWo26XAm8qnCxifKrmd9xdFcdEU6vDyoiharWvy2bJekE/O3ly0nj3tO8DS8vFBrQooiuKqoihG21WgKIq7GL9auy35qmsnq8jjJf7ZIq+zgd+XT5c0bw95Kur/LJ/+hvz+2y78VzQNDC/76R9SPn1lURSXTd4LiqL4B+PHahFtWle6qE6924+1Z/6D8SvvJxVF8f3mBEVRFORWh2vLl54XQtimQ57X0eIKe1EU64BPl083KW8vKF9v9onK45bfoYrVwPOLopg0CURRFN8DTi2f7kkOSpvTdJxCvCiKW8hrAEG+QPHALvVZDhzV6TvUxY6Vx+d2qdutLV5+VeXx81ulKfJir43vYmjap5UTiqL43xb5rCO3CDcs6ZKP1JGBiKRW/lF5vOWs1WKi51YGezausF5Gvmrf+Fv2waIovti039PL+58VRXF9lzLOqzw+oEO60zqdVPeq7N7TOMn5UVEU13VIfgpwV/m4W3eIqSzqeCv56vIk5Yn0VeXTMXIrQqt0BbkrGYx3deoo5IkDdgshPDCE8C8hhH8hX2FteFiXLM4qiuLmDtt/Xd7vEELYomnbkxj/znxkGsey8Z1aTQ7C2iqK4mJy1zjo/J1qp/H7K4qiuHMa+zerfnc+0y5RGeCdXD5dSF40tJ1vtAoKSpdUHp9VFMVtbdJdXHnc7Tt0Znli3c5nK4+7dh0KIWwTQrhP03exOiFAt+/iGUWeMGO6qu/luSHPuNeTsltqoyvbJUVR/Kpd2qIofsT42lKHlt1C2zm1w7ZfVx7v3lNFpTYcIyKplWrwsXLWatGb5eRFD08siuIn1Q3lP+mHlk8PLgOYXu3YYdvFHbZNxYMrj3/ZKWFRFHeFEC4G/hV4cAghlAFAsxVFUVwzhTr8uU0+DY0WpL8XnccJNNJt1S5BOZvRq8kL7nVqcYKJLQGtXN5le/WEdyvyuI+GvSuPz++STyuNvv8LgDWdz+cm6PSdaucf5N9jCCFs0YdgpPGduxP4vy5pL6g8fgjtx6h0Wien2gLZa7q236FS25Pt0kXk7lnzmPgbWy+EsDd5XNih5PEsnXT7Ls7o70FRFFeFEM4ntwRF4JEhhG+Rv5sXdAm470sOFKHL35BKmnuTW+h2JnfRbKXTGLnm35Y0bQYiklqprsPR7gpm3ZpXVl9LvtL89w4n0osYX8tjqjbvsO32aebZrNrtrdVA8mY3lvfzyScArQKDqdbtri7bG91Nek3X8vMupyz9PHnwey86ff5TqU+rOjW+3wW9fe7N7j6NfaD7e2plGeMXBnYgD/KficZ37pYeuhLdWHncqYtmp2NRLaPXdN1+s51OzCmKYnUI4XZyADGp3iGENwDvIXdR6kW349aPvwfPIE94cDCwK7nr1KsAQghXkIPAzxRFcVXTftP9G9LYt1UgMloUxeoWrzdM5VhJHRmISGqlesW425XnurRaWb2b6t+47wBvncK+nfrjz7hbVgtTaa3plH5D1G1GQl4Y8wvkMQJ3kmfEOpM81fDyxklPyCuIN06Aem5mmIGpfuYNje/V9eTZu3r1j+5JJvk9+Qo2wD7MPBBp6Nf3bTb0UpeW358QwsGMjyO7hTyL17nkLksri3KNlnIcUaMFp9t3cca/uXKM1iHl+KOnAY8mdwnbhDzWZU/g1SGE/y6K4sPtsplqsdOtr9QvBiKSJgh5AbbGgmAr6F83pNlQHXS/3TQCmQ2tOqi0l247jS4k64CZ9Emv2wvIJ1QATymK4pw26barqT63lPcj5M90aYe07fbfndwycmWXq8czdS7jg9yfDKQZ5ncrZd1DCCNdWkWqXZYGpWUUcstQW2VA21hEsXng9lHl/TrgUeUkFa3U9V2coCiKX1J2sQohLCR3xfxP4Hnk1sQPhRAuqfyGpvs3BAbrmGoj5WB1Sc1ew/gVwG/PYCaYWVcOuG3MaPTILjP/zIZqH/39OyUsT0oa413+r8u4jkHzoPL+lg5BCOS1N+pQXRen2wxNrTRmJNuU7mtezNRpjI9vObzLbG69aHzntmD8uLTzr5XHl7RNVb/9umzfm/ELrc3jYBrv+fcdghCo77vYVlEUq4qiOKcoiqOAIyubnll5/BfyDHLQ5W9IqbEy+600rbckzQYDEUnrlQv/NaZ4HCX3ox52jcXINgVeN5sVaVbO/NMIlB4fQtilQ/LnkE8eoc0sVwOscVK4sMtMPcfUURlyf/tGgP3qEMJU/xdWF7h70zT271k5y1RjatsFwKnlFf+ehBCe1/RS9bvz4g77LSBfhYd8ottqYcDZclgIodPV/xdWHp/dtK3xXWyeSW29ctaql0yzbhtK9bitH8NXzlbWmPL3oSGEtgFUCOFxjM9IdvaQXczQHGUgIokQwqYhhFeRB4Q3/i68viiKv8xitfrlo4wPJn1DCOE5nRKHEBaFEOo6IYZcP8gnmaeUJ4DNddqL8X7tq8krww+TxroNdyMPyp0khPDf1LRKc5FXfm90cXoEcGKnYCKEsGvT/ueSF3UDOBD4RKcpV8vf1wtCCNu3S9PFW8nrnUBuwTmruU4tytwzhPBd8grbVaczvj7IS0II7ca4fIy8KCrAye3W35klC4AvlLPiTRBCeALjrQd/ZnIg0vgu7hlCmDSdchkof4zua4f0TQjh4SGEh3dJdljlcfM4oY9WHn8hhLBtizJ2ZPzvRtG0jzRrHCMibRy2KOfGb9iEvOL4ruRuDk9jvO/wGPCuDgMih0pRFLeFECLwQ/L7PiWEcAS5y8sfySf225JPPA4mDz6+iYkLrG1InyefnD+2LP935SKNlwCbla+/hvGZk17XZb2RQfRF4Ojy8ckhhH3IV3hvI49XeB75ROtndF6vop+OIU/De5+ybgeGEE4ij4m6i3zVeR/gcPKJ3+FN+/8XeRrZncnjDpaEED5H7va1ghx03Y/cvemp5BmK7s34+JSelbNAPQX4blmnJcCVIYRvAP9LXjH7jrKMPcnf4ceRZzS6tSmvdSGEI8krw88Dvlu+7+8wfjyOZnyhuqsZX0xzUPya/H25MITwEfKEGluRF2s8mnwxZRR4cYuupScDTyjT/CCE8CHy9+5O4AHkFuH9qfe7+HDgsyGES4Dvk7v+LSW/h3uQj+WLyrSrmbhOCkVR/DiE8AXg+cC/AJeEEP6H/DkF8vo1r2d8trePFEVRXQtEmjUGItLGYR/gDz2k+znwhqIofrGB61Or8h/1wcBXyOtXPLa8tdNpvYy+KoqiCCE8lRwYPRHYC/hci6TryMfmxLrq1i9FUfwmhPBGcqvOZuRVuJtX4v41eQ2FG6lBGaAeCHyDfKL2YODjbZJPmqmqKIobQgj7kY/bo8gBwAeb01X8k4mL5E21vjeGEB4NvIt8srwQeHZ5a+cq4I0t8jqv/M59hRzgHs14oFj1f8ATu6wfMxtOBB5DPvFutXhnY+X1SWvEFEXxzRDCp8nB49bAO1rs/0Pgv+ntb2Y/PaS8tbMSOKIoilYzGb6EfBHpheQLTCe0SNNoCRmoLqrauBmISBunu8j/1G4l/7O9CPh+URSdFrEaakVR/LxcUO9Z5JW19yZf9Z5PXkztr+TuL2dR8xiMcgXzJ4UQ/p08FuSR5KuXa4DrgHPICzZeWWe9+qkoiveFEH4NvJL8/rYlX4G/HPg6TVd5a6rTjeSWkKcA/498Jfzu5CvRS8kDgU9n4piQ6v43AI8OIRxKHkD8r+SZi7YgT9X7N3LL1o+B02d6Ql8UxSrg2PJq9zOBQ8gteYvJAd4KcgvGb8gtHOe2m2yiKIrvhRB2B15GbiHYndyKcyu5VSgBp05j1flaFEXxghDCj8gtBQ8hf59uIrcQ/U+ngehFUbw0hPBjcjCyN/l9LyMfqy8VRfHVEMJ9N/R7qDgVuJLcInogOZDYgbx+yQryb+Qs4KR2ixsWRbEOeFEI4Yvkz+RR5O9iQf4unwd8uiiKi1rtL82W4FglSZI0yEIIjyUHGZBbBb48m/WR1B8OVpckSZJUOwMRSZIkSbUzEJEkSZJUOwMRSZIkSbUzEJEkSZJUO2fNkiRJklQ7W0QkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLt5s92BbRBOAOBJEmS6hSmuoMtIpIkSZJqZ4vIHLZ06dJay1u8eDEAy5Ytq7VcdeZxGTwek8HkcRk8HpPB5HEZPLN5THbaaadp72uLiCRJkqTaGYhIkiRJqp2BiCRJkqTaGYhIkiRJqp2BiCRJkqTaGYhIkiRJqp2BiCRJkqTaGYhIkiRJqp2BiCRJkqTaGYhIkiRJqp2BiCRJkqTaGYhIkiRJqp2BiCRJkqTazZ/tCkiSJEmanmL1aladfxajNy9lbPOtCA/bn7DpgtmuVk8MRCRJkqQhVFx9JWMffycrVy4ff22rbRh52VsI977fLNasN3MmEIkx7gK8AzgMWATcCJwOHJ9Sur2H/ZcAP+mhqHumlK4r99kZeBrwBOABwI7AP4DfAZ9KKX27Q3lPBF4LPAyYB1wGfDKldEoPdZAkSdJGrFizmrGPvxMqQQgAK5cz9vF3MvLezw58y8icCERijLsDFwB3B84ArgD2BV4JHBZjPCCldGuXbK4Bjm+z7UHkgOOyRhBSejnw38DV5CDmJuBeZdrHxhg/klJ6TYv6vgw4EbgV+DKwBjgc+GKM8UEppdd2fdOSJEkbiaIoYGwMxkbz/ejo+OPqa8UYjI613rb+eeO1MShGYXSMorFtfR5N+4+NNuXbvK1F+rHxuhRd04+1f1+T3kP52rq1+bVWVi6n+P2FhP0OqvdATdGcCESAT5KDkFeklE5svBhj/DDwauDdwFGdMkgpXQO8vdW2GONp5cOTmjb9GliSUjq/Kf0DgAuBV8cYv5JSuqiybTfgg8BtwD5lucQY3wH8Bjg2xvitlNIvO9VXkiTNHUVRVE6i252Edj7ZbX1SPr5f0e1EvfkEvHKiXk27ctNNKEZHGbvrrvG8qyfULU/i25x4dzqJH6uUXYzN9iEaPrfcNNs16GroA5EY432AQ8ktGp9o2vw24MXAETHGY1NKd04j/0XAU4FVwJeq29p1vUopXR5j/DrwImAJcFFl8/OBBcD7G0FIuc/tMcb3AJ8nB00GIpKkOaVoeSW59ckuY6OsvXNFvpp8662drwyPVa46t7xSXi2jw4n3hDo1n5SPUrRM3+4kvvm1sQ7byvshsWq2K6DebH+P2a5BV0MfiAAHl/dnp5Qm/IpTSnfEGH9BDlT2B86ZRv5HkgOHU3sZa1Kxtrxf1/R6o75nttjnR01pJElDpGX3kbYnqFPo3jHW1H1kpl1Umq6UN068i5bdV1qcULc8kW5VdtO2opjS53nbBjpOGlIjI+Vt3vj9vHlNr1UeT9rW7vURwoRtLdKHEZjXLa/GayNl+vFtYX0eLerUIj3d0o+MUKxbR/HWY+COFZM/q622ITxs//qP0RTNhUDk/uX9n9tsv5IciOzB9AKRF5b3n+l1hxjjVsDTgQI4u2lz2/qmlG6MMd4J7BJj3DyldFeXci5q9XpKCYDFixf3WuW+mD9//qyUq848LoNnmI/J+Il24wp0efI8OvFqdPWEtmic8E5IU3blGF033mVk0v6V7iST9q+UPWH/deXrk6+QF6OjE8tu2v/28vVQ1mlCeaPrJgQB43mNNp28D89VbQ2gENafCIfKSXaYN37SGxonvPPmESonzqFyAj2efn7leT6Bbbl/Nd8J6eeN5zXSSD+feZvMh5F5jIUwuV6N/SfUa2RiHbvWfXIaRkYIIcz2ERo4a9/yIZa/5/WMLR8P20e22Y5t3vQBNtlp51msWW/mQiCydXnfIhyc8Po2U804xngQsCd5kPoFPe4TgM8BO5Bnwbq8KUkv9d2iTNcxEJG0YRSVq77jJ6wTT14nnThXrga3P/FeB6OjjIQAo6OMrl3bVEabk/mmNC1PvJtPnHs9mW9Osz7/dRNOrqsn3tKMdDlBrZ5Ih3nzCfPmUYSRaZ28hkkn45NP5jufeDelaTwemV/Jq3EyP79SdtMJf6Xs1kFGY5/hWGe6cTFl3brmTh+q2yb324vFn/4Wa3/zM9b9/QbC9juy2X4HERYM9mxZDXMhEOmmET5PrT04e3F533NrCPAh4D+BnwGTZszqQc/1TSnt3WZTAbBs2bJpFD89xerV3O0vlzJ681LuGrLFdOa6xlX36veh+ap2r7N+dO3esf6ktVv6HruotOmPPekKdMv30KIPd6eBkNXXpth9RJqgUxeOll0tmre1S9/UfaRtV45euq80lV294t2pi8qEriGtuqg0vzYCYTz9VE+0FzX9/Rr6X+YYZavZGOM9uIdPq/8rml2LD3wsm5KPyZ133AF33FFb2TvttNO0950LgUijZWHrNtu3akrXkxjjduTuVZMGqXfY53/Is3T9FPj3lNLqFslWAIvL+raaUrhR35VTqe9sGtTFdIrqiWabgZDtT7xbncw2XwGfQn/sSX2ox7oObJw0A0nbQZVtTrzLbTdTNF2lH3P2Ec1MCK1POCedtLbqQ93hxLt6dbv5ZL7lSXz1ZLf3/tWh1Yn6yAhbb7cdjMxjxR139Pb+2pyo231EknozFwKRP5X3e7TZ3jgTbjeGpJ3nkgepn5JSWt4tcYzxI8CryOuJPLHD+I4/kQORPWiaGSvGuCO5W9b13caHDIqOi+l88E2w/5L8T7nF1edi0kn85CvfbU+8J530tziZFzAHriDOlileSe564l3ZtmDh5oSREVavXdd0kt3tKnqrq8zNJ+VNV7XbDYTs9v46BBbD0n1kqjYtr/IGr/JKUi3mQiDSWA390BjjSHXmrBjjlsAB5FaNC6eY74vK++a1QyYox4R8HDga+F/gKSmlTjPbnVvW6TAmT9H7+EqaoVD8/sLJQUjDmtXw07M8ER5Unbpw9NR9pN0JbVMXkqme7La8Uj7xhDpMOCnvVKepvL9K+rBhr2pvY7cGSZKGPxBJKV0VYzybPDPWMeQVyxuOJ7cwfKa6hkiMcc9y3yta5RljfBTwAODSToPUyyDkJPLMWj8CnpZS+meXKp8MvB54WYzx5MqChtsCbyrTfLpLHoNj0BfLWX/C2qELR4f+2J27d8y8P3anbh6hRf/q9lfd27+/RdvfHUZGuHX58olXte0+IkmSZtHQByKlo4ELgBNijIcAlwP7AY8hd8l6c1P6xkxW7c7EGoPUO7aGAG8lByGrgIuBN8QYm9NcnFI6vfEkpXR1jPF1wAnAb8uFD9cAhwO7AB8aqlXVuy2Ws//BhN33aN29YzoDIafQRWWudh+ZqpEt87CjsKpbjCxJklSfORGIlK0i+wDvIHd5egJwI/lk//iUUs9rIpUtE4fT2yD1e5f3C4E3tklzCnB69YWU0okxxmuA1wLPAUaAPwLHpZRO6bWugyA8bH+KrbZp3T1rq20YOeKlzp4lSZKkSULhNJVzUQGwdOnSegorZ82aEIwMwKxZypxmcfB4TAaTx2XweEwGk8dl8MzmMalM3zvlPt9zohTMG4AAACAASURBVEVEsyvc+36MvPez3O0vlzH69xtcR0SSJEldGYioL8KmC1j46EMB+KdXSCRJktSFo3klSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLtDEQkSZIk1c5ARJIkSVLt5s92BfolxrgL8A7gMGARcCNwOnB8Sun2HvZfAvykh6LumVK6rrLfC4B9gYcCDwIWAu9OKR03zXLen1J6Qw/1kCRJkobWnAhEYoy7AxcAdwfOAK4gBwevBA6LMR6QUrq1SzbXAMe32fYg4GnAZdUgpPQhYGvgdmApsHuP1T4fOK/F6z/vcX9JkiRpaM2JQAT4JDkIeUVK6cTGizHGDwOvBt4NHNUpg5TSNcDbW22LMZ5WPjypxeZnApenlK6NMR4JnNxjnc9LKbUsT5IkSZrrhn6MSIzxPsCh5BaNTzRtfhtwJ3BEjHGLaea/CHgqsAr4UvP2lNKZKaVrp5O3JEmStLGaCy0iB5f3Z6eUxqobUkp3xBh/QQ5U9gfOmUb+RwILgFN7GWsyBfeNMb4M2Aq4CfhZSunKPuYvSZIkDay5EIjcv7z/c5vtV5IDkT2YXiDywvL+M9PYt5Nnlbf1YozfAl7Ua8ATY7yo1espJQAWL148wypOzfz582elXHXmcRk8HpPB5HEZPB6TweRxGTzDekyGvmsWeaA4wIo22xuvbzPVjGOMBwF7kgepXzCNurVyC/AG8gD4LYHtgccDvweeDnwvxjgXjoskSZLU1lxoEekmlPfFNPZ9cXnft9aQlNJlwGWVl/4BnBljvAC4GDgAeBJ59q9uee3dZlMBsGzZsplVdooaUXjd5aozj8vg8ZgMJo/L4PGYDCaPy+CZzWOy0047TXvfuXDlvdHisXWb7Vs1petJjHE7cgtFy0Hq/ZZSWgl8tXz66A1dniRJkjSb5kIg8qfyfo822+9X3rcbQ9LOc8mD1FNKafl0KjYNt5T305rhS5IkSRoWcyEQaaxSfmjz2IoY45bkrk6rgAunmO+LyvtWa4dsKPuX93+tsUxJkiSpdkMfiKSUrgLOBnYDjmnafDy5deHUlNKdjRdjjHvGGPdsl2eM8VHAA4BL+zhIvZH3Aa0Go8cYnw08A1gDpH6WKUmSJA2auTJY/WjgAuCEGOMhwOXAfsBjyF2y3tyU/vLyPtBaY5B619aQGOMLgQPLp/ct758UY9ylfHxFSul9lV2+AoyUg9OvBzYDHgHsC6wDXlKu8i5JkiTNWXMiEEkpXRVj3Ad4B3AY8ATgRuAE4PiU0m295hVj3BY4nN4HqR9IHk9S9eDyBnA+UA1EPgU8ltxlbDE5GLoB+CLw0ZTSJb3WVZIkSRpWoSimM6utBlwBsHTp0loLdTq/weRxGTwek8HkcRk8HpPB5HEZPAMyfW+7nkZtDf0YEUmSJEnDx0BEkiRJUu0MRCRJkiTVzkBEkiRJUu0MRCRJkiTVzkBEkiRJUu0MRCRJkiTVzkBEkiRJUu0MRCRJkiTVzkBEkiRJUu0MRCRJkiTVzkBEkiRJUu0MRCRJkiTVzkBEkiRJUu0MRCRJkiTVzkBEkiRJUu0MRCRJkiTVzkBEkiRJUu0MRCRJkiTVzkBEkiRJUu0MRCRJkiTVzkBEkiRJUu0MRCRJkiTVzkBEkiRJUu0MRCRJkiTVzkBEkiRJUu0MRCRJkiTVzkBEkiRJUu0MRCRJkiTVzkBEkiRJUu0MRCRJkiTVzkBEkiRJUu0MRCRJkiTVbv6GyjjGGIBFwOYppb9tqHIkSZIkDZ++ByIxxkcCbwQeA2wOFNVyYozbAB8qXz8mpbS633WQJEmSNNj62jUrxngM8FPgicAWQChv66WUlpNbSp4HPL6f5UuSJEkaDn0LRGKM+wIfA0aB1wO7An9vk/xkcoDy9H6VL0mSJGl49LNr1mvIwcXbUkofBIgxtkt7fnm/bx/LlyRJkjQk+tk161Hl/ae6JSy7Z60Edulj+ZIkSZKGRD8DkcXAypTSyh7TF30uX5IkSdKQ6GcgsALYMsa4oFvCGOM9gK2BW/pYviRJkqQh0c9A5BLyGJElPaQ9qrz/VR/LlyRJkjQk+hmInEoORN4bY9y6XaIY47OBN5O7Zn2hj+VLkiRJGhL9nDXry8BzgEOAi2KMpwCbAcQYnwjsRZ6udx9ywPKdlNKP+li+JEmSpCHRtxaRlFIBPBU4A7gP8HZgq3LzGcB7gUeQg5BvA0f0q2xJkiRJw6WfLSKklP4BPDXGeAhwJPBIYEdywPN34JfAF1NKZ/WzXEmSJEnDpa+BSENK6RzgnA2RtyRJkqTh17dAJMb44fLhR1NKf+tXvpIkSZLmnn62iLwCWAe8to95SpIkSZqD+hmI3AxsllIa62OekiRJkuagfq4jcgGwdYxx1z7mKUmSJGkO6mcg8kFgtLyXJEmSpLb61jUrpXRhjPFZwOdjjOcDHyZP13tLucaIJEmSJAH9nTVrtPL0wPLW2NZutyKltEGmEJYkSZI0uPoZBISa9pEkSZI05PoZiNy7j3lJkiRJmsP6OUbk2n7lJUmSJGlu6+esWZIkSZLUkw02UDzGeHfg4cD25Uu3AL9LKd28ocqUJEmSNBz6HojEGA8E3gU8qs32nwLHpZR+0e+yJUmSJA2HvnbNijEeBfyEHIQE8gKHN5e30fK1g4DzYowv6WfZkiRJkoZH3wKRGOPDgI8D84BfAI8Dtkwp7ZhS2hHYEjis3DYP+Hi5jyRJkqSNTD+7Zh1LDmwS8F8ppbHqxpTSauDsGOOPga8BhwOvAY7oYx0kSZIkDYF+ds06CCiAVzcHIVXltleVaZf0sXxJkiRJQ6Kfgcj2wPKU0o3dEqaUlgLLGZ9RS5IkSdJGpJ+ByEpgyxjjFt0Slmm2KveRJEmStJHpZyDyO/Ig9Ff0kPaVZdqL+li+JEmSpCHRz8HqJwGHAu8sWzz+J6W0opogxrgj8DpysFKU+0iSJEnayPQtEEkpfTvG+CXyLFhvBI6NMV4C3AAsAO4F3A/YhLyeyCkppe/0q3xJkiRJw6PfK6sfCVwOvIE8BmTfFmlWAu8BPtjnsiVJkiQNib4GIimlAnhfjPEEcjethzM+M9Yt5HEkZ6eU7upnuZp9q9eNcdYVN3Pjyn+yZVjL/rtuyYL5/RyCJEmSpLmk3y0iAJSBxunlTXPclbeu4l3nXc/yf46uf22bzW7muCW7cL9FC2exZpIkSXPbMF8M3iCBiDYeq9eNTQpCAJb/c5T/Puta7r3tAuaNhPLVwPpHgfWPG89zivEnYf1zJj4v04znOr5/JXn5WmhbXqjsMaH8FmkCoU09Jpc38fnkOoZqyTP8HEKovovWn8Pmm/+DEOCuu1ZNTjPV99jj59Dbe+wtTbvPsfoee/kc2qWZ/Nn29jm0/86Nl9fu+7T1qvmEACtW3JnzallGtV6dy2vk0PH71Kpencprfj+t3mNzvbqU1yhj4vOm73lzmjbltf5su38O1e+BJA27Yb8Y3LdAJMa4C/B8YGlK6XNd0r4E2AH4bC8LIGpwXXjdHZOCkIbRAv5y2+qaa6TJbpvtCmiS62a7AqJVQJefBIr1KToHZJUgqm1gO77DjAKyCfXoQ0BWqWTroK+pvBbbx/dp8zlMqhdT/hw23fTvAKxdu6a+z6GS4VQ+h+aLKzP5HJovnFTz7DlNmMJ7nMLnEAhssUX+337XXXe2eI+9fA5hwvbmMnp+jx3StP4dhZ4+h471qnxHJ76nNu9xUhnVevXyObT/nq8dHeMdP7meO9aMUbX8n6O867zrOekpuw98y0g/W0SeC7yNPD1vN/cA3gqsBt7fxzqoZn//x9rZroIkTUsj3CiK6otFixTdctCGc+dsV0AaSsv/OcqF193BQffeerar0lE/A5Enlvdn9JD2K+Sg5ckYiAy1He62Scftz3zQYh624xYUFOv/ZxdU/n0XlZMBignnAAVMeg5QVF4siomnAsX6Moo2ZTSeFxOeN5c3Mc+iTRmd07StV8v3VH2t2+dQTC6jxXtspN188y0AuPPOOzt+DhPKmOHn0L5ebcpryrNR9+YyJj6f2ufQuozG817q1aK8pjoVlRc7fQ6bzM9/etesXdu2vMq3vM1n3eFzaKpX759DI9XUPodJv6NKJSZ/56rPixb7bMDPQZI2IsNwsbifgchuwDrg6h7S/rVMu1u/Ci+7hr0DOAxYBNxIHix/fErp9h72XwL8pIei7plSWt+vIsb4AvI0xQ8FHgQsBN6dUjquS3lPBF4LPIy8yvxlwCdTSqf0UIeBsf+uW7LNZje37J61zWbzeNpe2w18s+Bct3jxYgCWLVs2yzVRg8dkdlWD6mrAtWjRIgpg2bJb6RaQVcPy1kFf7wHZhIsrLerVsbwJaSrpOgWK3cqbUEYxYfvkMnsorymP5jJa1qt8uOVWWwGwcuXKCeW1/6wnBtztA9vx8loHtq3TdPscer2I1ClN9+/T1D+Hyd+n9u+xW5oC2GyzPPZg1apVrb/nPX4O7cuc+ufQ9eJJx3p1u9DRqMfEz2G8Hkzpc6h+4p0vnnSu151rRll21zra6XaxeBD0MxDZFrgzpTTWLWFKaSzG+A9ywDBjMcbdgQuAu5NbZK4gBwevBA6LMR6QUrq1SzbXAMe32fYg4GnAZdUgpPQhYGvgdmApsHsP9X0ZcCJwK/BlYA1wOPDFGOODUkqv7ZbHoFgwf4TjluzSYqDUPI5bsotBiKSB0zxgv2H+vPz3apN5YeIGzZrFi7cDYNmyrqcWqpEXUwbD6nVjvPiMq9peDN5/1y1noVZT089A5BZgpxjj9imlWzoljDFuD2wD3NSnsj9JDkJekVI6sVLOh4FXA+8GjuqUQUrpGuDtrbbFGE8rH57UYvMzgctTStfGGI8ETu5UToxxN/JijrcB+5TlEmN8B/Ab8or030op/bJTPoPkfosWctJTduey5bB0CKeOkyRJGjZz4WJwPwORXwFPJZ/wv7NL2qPL+1/PtNAY433IiydeA3yiafPbgBcDR8QYj00pTXnUW4xxEfl9rQK+1Lw9pXTmFLN8PrAAeH8jCCnzuT3G+B7g8+TPcGgCEcg/hkP39AqJJElSXYb9YnA/a3kSuS37LeW4iZZijC8CjiN3b+s4zW+PDi7vz27uFpZSugP4BbA5sP808z+SHDh8o5exJj1o1LdVAPOjpjSSJElSW/li8N05ct97ctC9tx6aIAT62CKSUjo7xvgV4FnASTHG1wA/BP5WJrkX8HhgT3LA8vWU0vf7UPT9y/s/t9l+JbnFZA/gnGnk/8Ly/jPT2LeVtvVNKd0YY7wT2CXGuHm5Qn1bMcaLWr2eUgLG+3DWZX45E1Dd5aozj8vg8ZgMJo/L4PGYDCaPy+AZ1mPS75XVnw+sJHctegA56KgK5JaQTwOv6lOZjQmSV7TZ3nh9m6lmHGM8iPweLkspXTCNurXSS323KNN1DEQkSZKkYdXXQCSltBY4Jsb4CeAIcneoHcgByE3AhcCXUkp/7Ge5XTSmPik6pmrtxeV9v1pDetFzfVNKe7fZVED9YzWcRWMweVwGj8dkMHlcBo/HZDB5XAbPbB6TnXbaadr79rtFBIAy0Hjjhsi7hUbLQrulI7dqSteTGON2wNNpM0h9BlYAi8n1bTWlcKO+K/tYpiRJkjRQhmc0S3t/Ku/3aLP9fuV9uzEk7TyXPEg9pZSWT6dibbStb4xxR3K3rOu7jQ+RJEmShtkGaRFpFmN8CPnEew1wcUrp2j5m/5Py/tAY40h15qwY45bAAeRWjQunmO+LyvtWa4fMxLnkOh3G5Cl6H19JI0mSJM1Z024RiTFuHmN8cnnbtE2aPWKMvwV+B3wN+Dbw1xjjd2KMUx483kpK6SrgbGA34JimzceTWxhOra4hEmPcM8bYPJC+Wu9HkQfbX9rHQeoNJwOrgZeVixs2ytwWeFP59NN9LlOSJEkaKDNpEXkscDrwh5TSd5s3lmMszgF2YnwAdsOTgTOAg2ZQftXRwAXACTHGQ4DLgf2Ax5C7ZL25Kf3l5X1zvRoag9S7tobEGF8IHFg+vW95/6QY4y7l4ytSSu9rpE8pXR1jfB1wAvDbGOPXyS1FhwO7AB8aplXVJUmSpOmYyRiRJeX9qW22vwHYuXz8eeBfgYcAHyhfOzDG+LQZlL9e2SqyD/BFcgByLLA7+WT/kSmlVoPCWypbJg6n90HqB5LHkzyX3OUK4MGV1w5rUd8TycHYZcBzyIHPTcCRKaXX9lpXSZIkaVjNpEVkP/I0sT9qs/2Icvv3U0ovqrz+hrJb1ouBZ5C7a81YSuk64Hk9pm3XEkK5evrCKZR7JHn19SlJKX0P+N5U95MkSZLmgpm0iOwErGW8m9N6McYHktcPAfhYi30brz18BuVLkiRJGlIzCUR2AO5IKbVaeG+/8n4V8NMW2/8EjAL3mEH5kiRJkobUTAKRMWCbGGOrPB5R3l+SUlrXvLGcYncFeZ0OSZIkSRuZmQQi15f7P6TFtoPI40N+1WrHGGMgryzu6uGSJEnSRmgmgcgvyvvXVV+MMR4INNboOLvNvg8E5gFXz6B8SZIkSUNqJrNmfZI8W9QzYoxbAD8gT9f7cnJryLW0D0QaU9r+dgblS5IkSRpS024RSSldBLyLvCjgE4FPAccB25ADkZeXY0FaeW6Z5ifTLV+SJEnS8JpJ1yxSSm8DngX8mjxDVmOWrMellH7Qap9y5fMdyQv4/XAm5UuSJEkaTjPpmgVASuk04LQppD8HWDzTciVJkiQNrxm1iEiSJEnSdBiISJIkSaqdgYgkSZKk2hmISJIkSaqdgYgkSZKk2hmISJIkSaqdgYgkSZKk2hmISJIkSaqdgYgkSZKk2s1oZfUY4xLg0cAdKaWP9LjPa4C7AeemlH4+k/IlSZIkDadpt4jEGDcDvgy8DbhuCrteB7wdOCXGuMl0y5ckSZI0vGbSNevpwE7AeSmlb/a6U0rpG8D5wG7AU2dQviRJkqQhNZNA5D+AAvj4NPY9EQjkYEaSJEnSRmYmgcje5f2Pp7FvY599ZlC+JEmSpCE1k0Dk7uRB6ndMdceU0krgDmCHGZQvSZIkaUjNJBAJfdg/zGB/SZIkSUNqJoHEMmCLGOO2U92x3OduZR6SJEmSNjIzCUQuKe8Pm8a+Tyjv/28G5UuSJEkaUjMJRH5E7lp1XIxxQa87lWnfTJ5x6wczKF+SJEnSkJpJIPJF4CZgT+AbMcYtu+0QY7wb8I1yn5uBU2ZQviRJkqQhNe1AJKW0CngJuWXj34HLYozHxhj3aE4bY9wjxvha4LIy7RjwkjIPSZIkSRuZ+TPZOaX0vRjjC4FPAbsAHwA+EGNcDdxeJtsWaHTdCsBq4JiU0ndnUrYkSZKk4TWTrlkApJS+COzL+HiPAGwG7FjeNmN8mt4fAPunlL4w03IlSZIkDa8ZtYg0pJT+ADwpxrgTsAR4ALCo3HwrcDlwfkrphn6UJ0mSJGm49SUQaUgpLQW+2s88JUmSJM09M+6aJUmSJElTZSAiSZIkqXbT7poVY/zrFJKvIo8V+T3w7ZTS+dMtV5IkSdLwm8kYkd2msc8BwMtijGcBz04p3TaD8iVJkiQNqZkEIsdPIe3m5Kl8HwnsDjwOOD3GuCSlNDaDOkiSJEkaQtMORFJKUwlE1osx/gfwJXLryDOA06ZbB0mSJEnDqfbB6iml04HXkRc5/H91ly9JkiRp9s3WrFlfBQpg71kqX5IkSdIsmpVAJKW0EljO+OrrkiRJkjYis7mOyKbAmlksX5IkSdIsmZVAJMa4B7AFcP1slC9JkiRpds1Wi8hx5DEiP52l8iVJkiTNopmsIzIlMcb5wL7AscBTgTHgk3WVL0mSJGlwTDsQiTH+dQrJFwLbNZX3lpTS/023fEmSJEnDayYtIrtNc7+rgeNSSi5kKEmSJG2kZhKITGVl9X8CtwK/B36XUipmUK4kSZKkITftQCSlNJVARJIkSZLWm811RIgxbjWb5UuSJEmaHbXNmtUQYwzA44DnAk8mryciSZIkaSNS5/S9DyQHH88C7gEE8loikiRJkjYyGzQQiTEuAv6LHIA8rHw5AGuBc4FvbcjyJUmSJA2mvgci5cKFTyIHH48vy2i0fnwf+AbwvZTSin6XLUmSJGk49C0QiTHuQw4+nklevLARfPwMeHSZ7IiU0sp+lSlJkiRpOM0oEIkx7gQ8mxyA7EkOPgD+AHwFOC2ldF2McWxGtZQkSZI0p0w7EIkxngkcQp4COAB/A04DvpJSurQ/1ZMkSZI0F82kReRQcterrwKfSSn9vD9VkiRJkjTX9WOMyH8AxBjvBvxvSmm0D3lKkiRJmsNmsrL6U4EzgE3Ja4P8ALgxxnhijPFf+1E5SZIkSXPTtFtEUkpnAGfEGLcjByLPBR4OHAMcHWO8ltxt67R+VFSSJEnS3DHjrlkppduAE4ETY4x7AUeSA5PdgDeWt4Z7Ag5klyRJkjZyfV3QMKX0R+D1McY3kAezHwk8GdiMPLPWJTHGS4BvA99KKV3ez/IlSZIkDYeZjBFpK6U0llI6M6X0TOAewFHABeRg5KHA8cClMcY/bojyJUmSJA22vraItFKupH4ScFKMcXdyK8mzgXsB99/Q5UuSJEkaPBs8EKlKKV0FvAV4S4zxMcBz6ixfkiRJ0mCoNRCpSin9BPjJbJUvSZIkafZskDEikiRJktSJgYgkSZKk2hmISJIkSaqdgYgkSZKk2hmISJIkSaqdgYgkSZKk2hmISJIkSardrK0j0m8xxl2AdwCHAYuAG4HTgeNTSrf3sP8SelvX5J4ppeua9t0LeDuwBNgKuBb4GvC+lNKqprS7AVd3yP/rKaVn9lAPSZIkaWjNiUAkxrg7cAFwd+AM4ApgX+CVwGExxgNSSrd2yeYa4Pg22x4EPA24rEUQsh9wLrAJ8E3gOuBg4K3AITHGQ1JKq1vkeQk5UGp2aZd6SpIkSUNvTgQiwCfJQcgrUkonNl6MMX4YeDXwbuCoThmklK4ht2pMEmM8rXx4UtPr84CTgc2Bp6SUvlu+PgIk4Oll+e9rke3FKaWW5UmSJElz3dCPEYkx3gc4lNyi8YmmzW8D7gSOiDFuMc38FwFPBVYBX2rafBDwAOCnjSAEIKU0Bry+fHpUjDFMp2xJkiRprhr6QITcDQrg7DIAWC+ldAfwC3KLxf7TzP9IYAHwjRZjTRpln9m8U0rpr8CfgXsB92mR704xxpfEGN9U3j94mvWTJEmShs5c6Jp1//L+z222X0luMdkDOGca+b+wvP/MNMveo7xd1bTt38rbejHG84DnppT+1kvFYowXtXo9pQTA4sWLe8mmb+bPnz8r5aozj8vg8ZgMJo/L4PGYDCaPy+AZ1mMyF1pEti7vV7TZ3nh9m6lmHGM8CNiTPEj9gj6VfRfwTmBvYNvydhB5xq4lwDnT7UYmSZIkDYu50CLSTWN8RjGNfV9c3rdqDZlW2Smlm8kzalX9NMZ4KPBzYD9yK8zHumWeUtq7zaYCYNmyZVOt74w0ovC6y1VnHpfB4zEZTB6XweMxGUwel8Ezm8dkp512mva+c6FFpNHqsHWb7Vs1petJjHE78qxXrQap973slNI64HPl00f3WE1JkiRpKM2FQORP5f0ebbbfr7xvN46jneeSB6mnlNLymsq+pby3a5YkSZLmtLkQiDRWQz+0XL9jvRjjlsAB5FaNC6eY74vK+5M6pDm3vD+seUM5rfAe5FXW/9pjmY2ZvXpNL0mSJA2loQ9EUkpXAWcDuwHHNG0+nty6cGpK6c7GizHGPWOMe7bLM8b4KPL6IJe2GaTecD5wOfDoGOOTK/uPAO8vn346pVRUtu0XY9y0RZkHkxc/BPhyhzIlSZKkoTdXBqsfDVwAnBBjPIQcHOwHPIbcLerNTekvL+/bLTTYGKTeqTWElNJojPF55JaRb8YYvwn8DTgE2Ie8hslHmnZ7P/DAcqre68vXHsz4miRv6RL8SJIkSUNvTgQiKaWrYoz7AO8gd5N6AnAjcAJwfErptl7zijFuCxxO50Hq1bJ/FWN8BLn15VBgS3J3rHcA70sprW7a5UvkldofATwe2AT4O5CAj6eUftZrXSVpQ1m3bh2rVq1izZo1FMV0Jh0cPrfdlv9VjI2NdUk5N4QQ2HTTTVm4cOH6NQgkqU5hY/kHs5EpAJYuXVproU7nN5g8LoNn0I/JunXrWLFiBQsXLmTBggWMjIwQQrsG5LmjcTK+bt26Wa7JhlcUBWNjY6xevZpVq1ax9dZbD2QwMui/lY2Vx2XwErYaigAAIABJREFUDMj0vVP+RzH0Y0QkSf21atUqFi5cyOabb868efM2iiBkYxNCYN68eWy++eYsXLiQVatWzXaVJG2EDEQkSROsWbOGBQsWzHY1VJMFCxawZs2a2a6GpI2QgYgkaYKiKBgZ8d/DxmJkZGSjGQckabD4n0aSNIndsTYeHmtJs8VARJIkSVLtDEQkSZIk1c5ARJKkIXT44Yez8847z3Y1JGnaDEQkSdOy8847s/POO7PLLrtwzTXXtE3XOGHeeeed+frXv15fBbto1Klx23XXXdlrr7143OMex2tf+1rOPfdcRkdHZ61+r3rVq9h555257rrrZq0OkrQhDd7qRZKkoTF//nzWrVvHaaed9v/bu/P4qKr7/+OvQGSJEDZBJSAhKVbUr6jsIIIbKn4FFTyuVBSxIlVLRFEU64ZKRVBAtCJCS0U5wldjLVgXIPxEUIooouJCILKIEFnCDmHy++PcidkmmWEmk0nyfj4eeUxyl3PPnTN3cj/3bDzwwAPF1mdmZrJs2bL87WJRWloa4EYL27VrF2vWrGHu3Lm8/vrrtGvXjkmTJpGamlrBuRQRqXoUiIiIyFE77rjjOP7447HWcu+99xabnfv1118nLy+PCy+8kPfee6+Cclm6e+65Byg8s/q2bdt46KGHePfdd7n22muZP39+/szFIiISGWqaJSIiYbn++uvZunUrH374YaHlhw8f5s0336RDhw6cfPLJJe67atUqHn74YS688EJOO+00UlJS6N69O48++ig7d+4stO3OnTvp3LkzrVu3ZtWqVYXW+Xy+/CZgc+fODfucmjZtyosvvkjXrl3ZvHkzEydOLLbNjh07eOqpp+jZsyepqamccsopGGPIyMgotu3s2bPzm6Z9+OGH9O3bl9/97neceuqpDBkyhMzMzELbJyUl8eabbwLQpUuX/OZjnTt3LpZ2bm4uEydOpHv37rRu3ZoOHTowZswYTVIoIjFPgYiIiITliiuuICEhgVmzZhVa/v7777Nt2zauv/76gPu+9tprpKenk5qaijGGG2+8keOPP56XX36ZK664gj179uRv27BhQ6ZMmYLP52Po0KHs3r07f9348eNZunQpxhj69+8fkfOqUaMGd999NwDp6emFJv3buHEjl156KZMnT6ZJkybceOON9O3blx9//JEbbriB1157rcQ058+fz+DBgznxxBMZPHgw7du3Z968eVx++eX8+OOP+dulpaVx6qmnAjB48GDS0tJIS0vj1ltvLZbmsGHDmD59Op07d2bgwIHUqVOHKVOmMHLkyIi8DyIi5UVNs0REJCz16tWjX79+WGvZvHkzzZs3B2DWrFnUr1+fvn37llijAHDnnXfy5JNPUrNmzULLX3/9dUaMGMHf//53hg0blr+8ffv2jBw5kjFjxjBy5EimTJnCkiVLeP7552nTpg1jxoyJ6Ll16tSJ+Ph4srOz2bBhAyeddBLgOpJv3LiRKVOm0K9fv/ztd+3axYABA3j44Yfp3bs3TZs2LZTeBx98wIwZM7jooovyl73yyiv85S9/YdSoUVhrAddcbMOGDXzzzTcMGTKEli1bBsxjVlYWCxYsoFGjRgDcf//9XHTRRcyZM4cHHniAZs2aRez9EBGJJNWIiIhI2K6//nqOHDmSPyrWxo0bWbx4MVdeeSV169YNuF+LFi2KBSEA1157LfXr12fRokXF1g0dOpTzzjuP9PR0Jk+ezJ133kmtWrV48cUXSUhIiNg5AdSuXTv/Bv/XX38F4Ouvv2bp0qX06dOnUBAC0KBBA0aMGMGBAwf497//XSy97t27FwpCAG6++WaSk5NZsmQJGzduDDmPo0aNys8jQEJCAldeeSU+n48vv/wy5PRERKJFNSIiIhK2s88+m7Zt2/LGG29w9913M2vWLHw+HzfccEOp+x0+fJh//vOfpKen88MPP5CTk4PP58tfv2XLlmL7xMXF8fzzz9O7d2+eeuopAMaOHUvbtm0je1Ief5OsuLg4AFasWAHA7t27efbZZ4tt7w9YCja18uvatWuxZTVr1qRjx46sX7+e1atX06JFi5Dy165du2LL/LVSu3btCiktEZFoUiAiIiIRcf311zN69GgWLlzI7NmzOeOMMzj99NNL3Wfo0KHMnz+fVq1a0bt3b5o1a0atWrUA12QpUIfrJk2a0LlzZ9LT02nUqFHE+oUUdeDAgfxO802aNAFcJ3WAxYsXs3jx4oD77t27t9iyQCNv+ZtP5eTkhJzHBg0aFFvmr2WqyHlQRETKokBEREQion///vl9N7Zs2cLw4cNL3f7LL79k/vz59OjRg5kzZ3LMMcfkr/P5fEyZMiXgvunp6aSnp9O4cWO2b9/Oww8/zDPPPBOxc/H77LPPyM3NpWnTpvn9NBITEwF47LHHGDx4cEjpZWdnl7h869athdIWEakO1EdEREQiokGDBlx22WX8/PPPJCQkcMUVV5S6vX829t69excKQgBWrlzJgQMHAu5333330aRJE9577z26dOnCrFmzSE9Pj8h5+Pl8vvxO9gXP5eyzzwbg008/DTnNpUuXFlt25MgRli9fDlCoBkm1GiJS1SkQERGRiBk5ciTTpk1j1qxZ1KtXr9Rt/X0hit6cZ2dn8+CDD5a4z6FDhxg6dCh79+7lueeeIykpicmTJ9OoUSPuu+8+1q1bF5HzyM7OZujQoSxdupSkpCTuuuuu/HXt2rWjc+fOzJ8/nzfeeKPE/b/99tsSaz+WLFnCBx98UGjZ9OnTWb9+Pd26dSvUP8TfAX3Tpk2ROCURkZijplkiIhIx/on3gnHmmWfSsWNH5s2bR9++fenUqRPbtm1j4cKFpKamcsIJJxTb54knnmDVqlXcdtttnH/++QCceOKJTJgwgUGDBnHHHXeQnp6e388kGAU7nO/atYs1a9awfPlyDh06xFlnncWkSZNo3LhxoX0mT56MMYZ77rmHadOmcdZZZ9GgQQN+/vlnvv32W9asWcM777xTrE/IRRddxK233soll1xCcnIy33zzDQsWLKBhw4b5He/9zjnnHF588UXuu+8+LrvsMhISEmjQoAE333xz0OcmIhLLFIiIiEiFqFmzJtOnT+evf/0rCxYs4NVXX+WEE07guuuu489//jO9evUqtP3777/PtGnTaNeuHaNGjSq07qKLLmLIkCFMnTqVJ554gsceeyzofIwfPx6AWrVqUa9ePZKSkhgwYAB9+vShZ8+e1KhRvPFA8+bNee+993j11VeZN28eb731Fj6fj6ZNm9KmTRtuvvnmEkfxuvTSS7nhhhuYOHEiH330Eccccwx9+vTh/vvvJzU1tdC2vXr14uGHH2bWrFlMnTqVQ4cO0aJFCwUiIlJlxBWcKVaqjDyAzZs3R/Wg/id/gTpjSsVQucSeWC+T7OzsgKM7VWXx8e7ZXG5ubsTTnj17NmlpaYwfP55rrrkm4umHK1bLPNavlepK5RJ7KrJM/MOFA3Gh7qs+IiIiIiIiEnUKREREREREJOoUiIiIiIiISNSps7qIiEg5u+aaa2Kyb4iISEVSjYiIiIiIiESdAhEREREREYk6BSIiIiIiIhJ1CkRERERERCTqFIiIiIiIiEjUKRAREREREZGoUyAiIiIiIiJRp0BERERERESiToGIiIiIiIhEnQIRERERERGJuviKzoCIiEhl17lzZzZu3FjiuqZNm/LFF18EndbmzZsZN24cixYtYseOHTRr1oyLL76YtLQ0GjZsGKksi4hUOAUiIiIiEZCYmMitt95abPmxxx4bdBrr16+nX79+ZGdnc/HFF/O73/2OlStXMm3aNBYtWsTbb79N48aNI5ltEZEKo0BEREQkAhITE7nnnnvCSmPUqFFkZ2fz+OOPc8stt+Qvf+SRR5g6dSpjx45l7Nix4WZVRCQmqI+IiIjEvLyDB/EtW4Tv3dn4Ps0g79DBis5SxGVlZZGRkUHLli0ZNGhQoXUjRowgISGBuXPnsm/fvorJoIhIhKlGREREYlreuh/wTX4ccnb+tiyxITX+NJq41m0qMGeFHTp0iLlz57Jp0yYSEhJo27YtXbp0oWbNmkHtv2TJEgDOPfdcatQo/JywXr16dOzYkYyMDFasWEGPHj0inn8RkWhTICIiIjEr79DBYkEIADk78U1+nBpPTSWuVu2KyVwRW7du5a677iq07KSTTmL8+PF07dq1zP3Xrl0LQEpKSonrW7duTUZGBpmZmQpERKRKUCAiIiLl4siQvuV7gJyd+IZdHZGkak59J6z9r7nmGjp16sTvf/976tWrR1ZWFtOnT+e1117jxhtv5J133uG0004rNY3du3cDrq9JSerXrw9ATk5OWHkVEYkV6iMiIiISprS0NM455xyaNm1K3bp1OeWUUxg7diy33XYbBw4cYPz48WEfIy8vD4C4uLiw0xIRiQUKRERERMrJwIEDAVi2bFmZ25ZV47Fnz55C24mIVHZqmiUiIuUi3OZO4PUReWBI8T4iAIkNY6qPSEmOO+44APbv31/mtqmpqQBkZmaWuH7dunVA4D4kIiKVjWpEREQkZsXVqk2NP42GxCIzivtHzYrhIARgxYoVgOu0XpZu3boBsHjxYnw+X6F1e/bsYfny5dSpU4f27dtHPqMiIhVANSIiIhLT4lq3ocZTU8lbuQy2bYGmJxB3VpeYCUK+++47mjVrRqNGjQot37hxIw8++CAAV111Vf7yw4cPk5WVRXx8PMnJyfnLk5OT6dmzJxkZGcyYMaPQhIbjxo1j37593HjjjSQkJJTvCYmIRIkCERERiXlxtWoT17lnRWejRO+++y4vvPAC3bp1o2XLltSrV4/169ezYMECDhw4wPnnn8/tt9+ev/2WLVvo2bMnLVq04NNPPy2U1pNPPkm/fv0YPXo0H3/8MW3atOHzzz/nk08+ISUlhZEjR0b79EREyo0CERERkTB069aNtWvXsnr1alasWMG+fftITEykY8eO9O/fnwEDBgQ90lVycjLz5s1j3LhxLFq0iAULFtCsWTMGDx7M8OHDi9W6iIhUZgpEREREwtC1a9egJiz0a9myJZs2bQq4PikpiQkTJkQiayIiMU2d1UVEREREJOoUiIiIiIiISNQpEBERERERkahTICIiIiIiIlGnQERERERERKJOgYiIiIiIiESdAhEREREREYk6BSIiIiIiIhJ1CkRERERERCTqFIiIiIiIiEjUKRAREREREZGoUyAiIiIiIiJRp0BERERERESiToGIiIiIiIhEnQIRERGRcjBnzhySkpJISkpi1qxZIe37/fff88c//pEzzjiDlJQUevTowbhx49i/f3855VZEJPoUiIiIiETYpk2bGD16NMcee2zI+37++ef06dOH//znP/To0YPBgwdTv359JkyYwHXXXcfBgwfLIcciItGnQERERCSC8vLySEtLo1GjRgwcODCkfY8cOUJaWhr79+/n5Zdf5oUXXuDBBx/k3XffpU+fPixfvpypU6eWU85FRKJLgYiIiMS8g7k+Fq3bhf0qm4x1uziY66voLAU0bdo0lixZwvjx40lISAhp36VLl/LDDz/QpUsXevfunb+8Ro0aPPTQQwDMnDmTvLy8iOZZRKQixFd0BkRERErzw6/7eWLRRnYeOJK/rGGdrTzUqwVtmtStwJwV98MPP/DUU08xePBgunTpwpIlS0La3799r169iq1r1aoVKSkpZGZmkpWVRXJycgRyLCJScVQjIiIiMetgrq9YEAKw88ARnli0MaZqRnJzc7nrrrto3rw5999//1GlsXbtWgBSUlJKXN+6dWsAMjMzjy6TIiIxRDUiIiJSLvq9tqZc09954Ahm9vcRSSv9hlPCTmPChAmsXr2at956i7p1j66mZvfu3QDUr1+/xPWJiYkA5OTkHF0mRURiiGpEREREwrRy5UomTZrEH//4Rzp06FBux/H3DYmLiyu3Y4iIRIsCERERkTD4m2SlpKRw7733hpWWvybEXzNSVFk1JiIilYmaZomISLmIRHOng7k+bktfW6yPCEDDOjV5uV8qteMr9pna3r178/tsBOrbce+993LvvfcyePBgHnvssYBppaamAoH7gKxbt67U44iIVCYKREREJGbVjq/BQ71alDBqVk0e6tWiwoMQgFq1anHdddeVuO6rr75i9erVdOrUidTUVNq3b19qWt27d2fixIksWrSIO++8s9C6rKwsMjMzadGiBa1atYpY/kVEKkqVCUSMMS2Ax4BLgCbAz8DbwKPW2h1B7N8LWBjEoU6y1m4osu+pwCNALyARyALeAJ621u4PcLxuwENAF6AO8CPwKjDJWlv80Z+ISDXVpkldXu6XyrINu/llz2GOr3cMXVrWj4kgBKBu3bqMGzeuxHXPPvssq1ev5uqrr+b666/PX75//342bdpE3bp1SUpKyl/etWtX2rRpw7Jly3j//ffz5xLx+XyMGTMGgIEDB6qPiIhUCVUiEDHGpAKfAM2AdGAN0Am4G7jEGNPdWvtrGcmsBx4NsO5/gKuAr0sIQjoDC4BjgDnABuB84GHgAmPMBdbag0X26QfMBQ4As4HtwOXABKA7cHXZZy0iUn3Ujq9Bz9YNKjobEbNy5Uquvvpqunbtypw5c/KX16xZk/Hjx2OM4bbbbuOyyy4jKSmJjz/+mC+//JKOHTsyZMiQCsy5iEjkVIlABJiCC0LustZO8i80xowHhgNjgNtLS8Baux5Xq1GMMeZ179eXiyyvCUwHEoB+1tp3vOU1AAv0947/dIF9EoGpwBGgl7X2v97y0biAZoAx5lpr7RtBnLeIiFQxZ599NvPmzWPcuHFkZGSwd+9ekpKSGD58OMOGDaN27doVnUURkYiI8w8FWFkZY1KAtbgajVRrra/Auvq4JlpxQDNr7d6jSL8JsAnwAUkFm3kZY84HPgIWW2t7BshXFtDaWpvnLb8FmAb8w1p7U5F9AqYXojyAzZs3h5FE6I477jgAsrOzo3pcKZ3KJfbEeplkZ2fn57E6iY93z+Zyc3MrOCfRF6tlHuvXSnWlcok9FVkmzZs39/8acpvR2GhgG57zvdf3CwYhANba3cASXI1Fl6NMfxBQG3izhL4m/mO/V3Qna20m8D3QCkgJZh9gMbAP6GaM0SMvEREREamyqkLTrN97r4Gm1/0B6A2cjKttCNWt3uvfjvLYJ3s/a8vax1qba4xZB5yGC16+LS1jxpgVJS231gJE/emW/2liLD5Vq85ULrEn1stk+/bt+XmsjqrjudeoUSMmP4+xfq1UVyqX2FNZy6Qq1Ij4ey/uCrDev7xhqAkbY3oCp+A6qX8SoWOXW35FRERERCqL6vDYx99e7Wg6w9zmvZZUG1Jexw56H2ttoAHp8yD67QTVZjQ2qVxiT6yXic/nq5b9JKpzHxGfzxeTn8dYv1aqK5VL7ImRPiIhqwo1Iv4ahEDjOiYW2S4oxpjGuFGv9gMzI3jscsmviIiIiEhlUhUCke+815MDrG/jvQbqxxHITbhO6tZauzOCxw64jzEmHmgN5AKZIeVWRERERKQSqQqBiH829N7e/B35vOF7u+NqNZaFmK5/xqiXS9lmgfd6SdEV3vC9J+OG780MZh/gXNwIX58UnQRRRERERKQqqfSBiLV2LfA+kAwMK7L6UeBY3Jwd+XOIGGNOMcacEihNY0wPoC2wOkAndb8M3MhW5xpj+hbYvwYw1vvzJf8cIp45QDZwrTGmQ4F96gBPeH++WMoxRUREREQqvarSWf0O4BNgojHmAlxw0Bk4D9cs6sEi2/uHxQ008Yq/k3pptSFYa48YY27G1XLMMcbMAX4CLgA64OYwmVBknxxjzBBcQLLIGPMGsB3oixvadw4wu9SzFRERERGp5Cp9jQjk14p0AGbgApB7gFRgItDVWvtrsGkZYxoBAyi9k3rBY38KdATScfOVDMd1RH8MuKikJlbW2reBnrgJDPsDdwKHgTTg2iI1KCIiIiIiVU5cXp7ueaugPIDNmzdH9aAazi82qVxiT6yXSXZ2dqWbFCsSqvPwvbFa5rF+rVRXKpfYEyPD9wZqaRRQlagRERERERGRykWBiIiIiIiIRJ0CERERERERibqqMmqWiIhIhRkzZgxffvklmZmZ7Nixgzp16pCUlMQll1zCoEGDaNy4cdBpbd68mXHjxrFo0SJ27NhBs2bNuPjii0lLS6Nhw4bleBYiItGlGhEREZEwTZ06lX379nHuuecyePBgrrzySuLj43n22We56KKL2LRpU1DprF+/nksvvZTZs2dz5plnMmTIEE466SSmTZtG37592b59ezmfiYhI9KhGREREJExr1qyhTp06xZY//fTTTJo0icmTJ/PUU0+Vmc6oUaPIzs7m8ccf55Zbbslf/sgjjzB16lTGjh3L2LFjS0lBRKTyUI2IiIjEvNzcPDauP8T3Xx9gY9YhjuTG1tDzJQUhAJdffjkA69atKzONrKwsMjIyaNmyJYMGDSq0bsSIESQkJDB37lz27dsXdn5FRGKBAhEREYlpO3/NZcG/c1j56T6+W32Alcv28dG/c9j5a+zP9/HBBx8A0LZt2zK3XbJkCQDnnnsuNWoU/vdcr149OnbsyP79+1mxYkXkMyoiUgHUNEtERGLWkdw8Pvt4LwcPFK4BOXjALb/gskRqxoc8h1a5eemll9i7dy85OTmsWrWKzz77jLZt2/KnP/2pzH3Xrl0LQEpKSonrW7duTUZGBpmZmfTo0SOi+RYRqQgKREREpFz8a/bOck3/4IE85s3dFZG0Lr8mMqNRvfTSS2zbti3/7/POO48JEybQpEmTMvfdvXs3AImJiSWur1+/PgA5OTkRyKmISMVT0ywREZEI+eKLL9i0aRNffPEFr7zyCllZWVx88cV89dVXYaedl+dqheLiYqcGSEQkHApEREREIqxp06ZceumlvP766+zYsYO77767zH3KqvHYs2dPoe1ERCo7Nc0SEZFyEYnmTkdy8/jo3znF+ogA1K4TF3N9RIpq0aIFbdq04euvv2b79u2lTmyYmpoKQGZmZonr/SNvBepDIiJS2ahGREREYlbN+Dg6nXMstesUDjZq13HLYzkI8fvll18Aio2EVVS3bt0AWLx4MT6fr9C6PXv2sHz5curUqUP79u3LJ6MiIlGmQERERGJawybxXHBZImd1SeD3p9fhrC4JXHBZIg2bxEal/o8//sjWrVuLLff5fDz99NNkZ2fToUMHGjZ0NUSHDx/mxx9/ZP369YW2T05OpmfPnmzYsIEZM2YUWjdu3Dj27dvHgAEDSEhIKK9TERGJqtj4FpdKLzc3j7Xf7Wb37sMQd4gTk46pFE8qRaRyqBkfR4tWtSo6GyVauHAhTzzxBJ07dyY5OZlGjRqxbds2li1bRlZWFs2aNeOZZ57J337Lli307NmTFi1a8OmnnxZK68knn6Rfv36MHj2ajz/+mDZt2vD555/zySefkJKSwsiRI6N9eiIS4yrzPZgCEQnbzl9zvXH+fxtG8xuv2USsPLEUESkvPXr04IYbbmD58uV888035OTkkJCQQEpKCv379+eWW26hUaNGQaWVnJzMvHnzGDduHIsWLWLBggU0a9aMwYMHM3z48KDTEZHqobLfg8X5hwOUKiUPYPPmzeV+oMrekbQ6OO644wDIzs6u4JyIX6yXSXZ2dn4eq5P4ePdPOzc39mdsj7RYLfNYv1aqK5VLbIiVe7DmzZv7fw35YLEfKklM+3nT4RIvAIjsZGMSjvKdVE6ORmyXSbsuPnbWqH4341Adz9nZt9fHvz6Kxc9lLOZJVC6x7+CBPH7edDhmm7T6qbO6hGXfHl/ZG4mIiIhIVFWGezQFIhKWhHr6CImIiIjEmspwj6amWRKWE5OO4Zs6cRXePlECU1ve2BPrZZKdnU3DxtXv30N17iOS66sRkQkoIy3Wr5XqSuUSG8rqI3Ji0jEVkKvQxH6oJDGtKkw2JiIiIlLZVIV7sOr3yEsizj/Z2J6cOuzOOQxxByrVGNYiIiIilVFlvwdTICIRUTM+jtST6wOqqhURERGJlsp8D6amWSIiUozmmKo+VNYiUlEUiIiISCFxcXH4fLE/7KNEhs/nIy6ucjTjEJGqRYGIiIgUUqtWLQ4ePFjR2ZAoOXjwILVqxfakZyJSNSkQERGRQurWrcv+/fvZt28fR44cUdOdKigvL48jR46wb98+9u/fT926dSs6SyJSDamzuoiIFBIfH0+DBg3Yv38/O3furDaBSI0a7tlcdWmWFhcXR61atWjQoEH+HCoiItGkbx4RESkmPj6e+vXrV3Q2okqTtImIRJeaZomIiIiISNQpEBERERERkahTICIiIiIiIlGnQERERERERKJOgYiIiIiIiESdAhEREREREYk6BSIiIiIiIhJ1cdVloqpqRoUqIiIiItEUF+oOqhEREREREZGo08zqVVPIEWkkGGNWAFhr21fE8aVkKpfYozKJTSqX2KMyiU0ql9hTWctENSIiIiIiIhJ1CkRERERERCTqFIiIiIiIiEjUKRAREREREZGoUyAiIiIiIiJRp3lEREREREQk6lQjIiIiIiIiUadAREREREREok6BiIiIiIiIRJ0CERERERERiToFIiIiIiIiEnUKREREREREJOoUiIiIiIiISNTFV3QGJDYZY5oAVwKXAf8DJAGHgK+A6cB0a60vhPRaAI8BlwBNgJ+Bt4FHrbU7Ipv7qiuS5WKMWQ+0CrD6F2vtCWFnuJowxowFOgAnA8cB+4Es3Gd8srX21xDS0rUSIZEqF10r5ccYMxD4h/fnEGvtKyHseyrwCNALSMSV7RvA09ba/ZHNafVytOVijCltcrpPrbVdws5cNRHJ751Y/r+iQEQCuRp4EfdhXQj8BBwPXAW8AlxqjLnaWlvmjJjGmFTgE6AZkA6sAToBdwOXGGO6h3KjVs1FrFw8u4DnSli+JwJ5rU6GA58yhN0tAAAVfklEQVQDHwBbgWOBLribpNuMMV2stRvKSkTXSsRFpFw8ulYizBjTEpiEew/rhbhvZ2ABcAwwB9gAnA88DFxgjLnAWnswsjmuHsIpF08WMKOE5RvDyFZ1Ffb3Tqz/X1EgIoF8D/QF/l3wCbsxZhTwGdAfd/M7N4i0puAugLustZMKpDUed6MwBrg9clmv0iJZLgA7rbWPRDqT1VCitfZA0YXGmDHAKOAB4I4g0tG1ElmRKhfQtRJRxpg4XC3ur8D/ASNC2Lemt28C0M9a+463vAZgcd+Dw4GnI5ztKi+ccilgva6ViInE905M/19RHxEpkbV2gbX2X0Wb+VhrtwAveX/2KisdY0wK0BtYD7xQZPVfgL3AQGPMseHmuTqIVLlIZJV0s+tf5b22KSsNXSuRF4lykXJzF64G42bcZzsUPYG2wGJ/EALgfS/e5/15u3dTLaEJp1wkxlSG/yuqEZGjcdh7zQ1i2/O91/dLuHnebYxZgrtIugAfRS6L1VIo5eJX2xhzI3AS7gtpFe6f+5FIZ66autx7XRXEtrpWoieUcvHTtRIhxpi2uNqK5621i40x55e1TxH+7d8rusJam2mM+R7XLygFWBtWZquRCJSLX0NjzC3ACbimRSustcsilc9qJtzvnZj/v6JAREJijIkH/uD9WeyfQAl+771+H2D9D7iL4GR0c3XUjqJc/E4AZhZZts4Yc7O1NiMimatGjDEjcG2qG+A6SZ+D+8cRTBMRXSvlJMxy8dO1EgHed9VMXP+2UUeZTDDXysnejwKRIESoXPzaAdOKpP8lMNBa+1WYaVc34X7vxPz/FTXNklA9DZwOzLPW/ieI7Rt4r7sCrPcvbxhuxqq5UMsFXDvgC3BfdMfiRuH6G5AMzDfGtCuHfFZ1I3DV3X/G3ey+B/S21m4LYl9dK+UnnHIBXSuR9DBwFjAojJGtdK1EXiTKBWA80B1oCtQHOuIGE2gHLDDGJIWb0WokEt87MX+tqEZEgmaMuQu4BzfiwsAIJetvwxvsKE9SxNGWi7X20SKLVuPaVe/x0nsEN1SwBMk/nKIx5nigGy5AXGmM+V9r7edhJq9r5SiFWy66ViLDGNMJ97T9WWvt0nI8lK6VEESyXKy19xRZ9F/gamPMHNwgAiNwHaSlDFH63qnwa0U1IhIUY8ww4HngG+A8a+32IHf1R9sNAqxPLLKdhCCMcimNv9P7uRFIq1qy1v5irX0LV+XdhN/G4y+NrpVydpTlUhpdK0Eq0PTne2B0mMnpWomQCJdLaXStRE4o72XMXysKRKRMxpg/A5Nx0fh53ghNwfrOez05wHr/qDWB2i9KAGGWS2m2eq8anSlM1tosXJB4mjHmuDI217USJSGWS2l0rQSvHu6z3RY4YIzJ8//gms0BTPWWlTRvQkG6ViInkuVSGn8zSF0r4QvleyfmrxUFIlIqY8xIYALwBe5md2sZuxS10Hvt7Y3xXjDt+ri2pPsBjagRggiUS2m6eq+ZEUyzOmvuvZY1yomulegKtlxKo2sleAdxHZhL+lnpbfOx93dZzYMWeK+XFF3hDVd6Mm5SPZVL2SJZLqXxz6iuMglfKN87Mf9/RX1EJCBjzGjgMWAFrmNnwGY/xphjgFTgsLU2f5QSa+1aY8z7uKYQw3Cztfo9iovo/2at1XjlQYpEuRhjTgN+LrqvMaYVrpYF4J+RzntVZIw5BTfp1JYiy2sAj+MmkvrEWrvDW65rJQoiVS66ViLD6wB9a0nrjDGP4DpK/91a+0qB5Qm4YUv3WWt/KrBLBvAtcK4xpm+RCQ3Hetu8ZK1VH5EyRLJcjDFnA98V/Y4yxpyBmzQPdK0EJdTvncr8f0WBiJTIGHMT7mb3CPD/gLuMMUU3W2+tneH9noT7x5CFG9GhoDuAT4CJxpgLvO06A+fhqgMfjPwZVE0RLJergfuNMQuBdcBu3JfYZUAdYB4wrlxOouq5BHjGGLMYN1Tor8DxuEnXUoAtwJAC2+taiY5IlYuulYrTCfdEN4MCE7Vaa48YY27G1YzM8TpC/4QbYagDsARXYyzlo8RywU2GeJUxZgGwAVfbcgruWqwJTAVej2pOK69Qv3cq7f8VBSISSGvvtSZuyMuSZAAzykrIi8g74G6gLwH6AD8DE4FHI9TBurqIVLksxI0vfhaumvdYYCeuCn4mMFNPE4P2IfAyroq7HW4YxL24L/iZwMRgP+O6ViIqUuWiayUGWWs/NcZ0xD3V7Y0bKjYLd+08ba09WJH5q6bexnV+PgM3kV4d3AOA+cBUf82VBCVi3zux/n8lLi9P358iIiIiIhJd6qwuIiIiIiJRp0BERERERESiToGIiIiIiIhEnQIRERERERGJOgUiIiIiIiISdQpEREREREQk6hSIiIiIiIhI1CkQERERERGRqFMgIiIiIiIiUadAREREREREok6BiIiIiIiIRF18RWdARKQ6Mcbkeb+2ttaur8i8VGeVpRyMMXHAMOBW4GSgrrcqpvNd3owxvYCFQJa1NrnIuhnATcCj1tpHop03EQmeAhERkQAK3NAA5AJJ1tqtpWx/BfBWgUU3W2tnlFsGQ2SMuQr4A9AeaAYcBLYAWcDHwCJgsbU2L1AaEnWjgCe83w8Av3i/H6mY7BRnjEnABUpXAqcDDYAc3GfreyAD+NBa+3WFZdJjjHnE+/U5a+3OisyLiCgQEREJVjxwPfBcKdv8IYh0vvNeD4edoyB5N4pzgEsLLD6EC65SgTbAhd7yRoBu0GLH3d5rGu7mOaaCRGPM74D3cJ8jv71ALeA07+dK4EvgzChl62fcdZZdwrq/eK8z0OdcpMIpEBERKdtPwEm4QKPEQMQY0xi4DNiDu8lvXNJ21tpTyimPpZmAC0IOA88ALwM/WWvzjDF1cTUkfQkukJIoMcY0A5p6f06NwSAkHngbF4RsAR4FrLV2u7e+MdAdGIBrVhYV1toHgAeidTwROXoKREREyrYUdxN/ljHmtABNTK7FPQV+nd9qFyqcMSYRGOT9OcpaO67gemvtflyzrI+NMQ/hakkkNvj7g2Ct3VORGQngQlyNB8Dl1tr/FlzpBST/Av7lBbwiIoUoEBERCc5M4BFcrcHIEtb7axP+QSmBSKBO0l7b9b8Af7fWDjLG3ATcAZwK5AErgCettR+EmO/f4wIkgHdL29BaeyhAnjsDVwDnAS1xT+l3Ap8Dr1hr5wTYbwZep2HgSeA+4AagFa7ZzNvAX6y1O7zt2wMPAt1w/QxWA2OstW+XkPYgYDqQYa3t5b1fQ/nt/fov8Iy19r3SzjkQY0w94C5cs6KTgdrABmC+l+6GEvapgfsc3AScASTi3qetwKe42oIy81OgI3bBZQVrQwp1wvaCzTQvr/4mUpm4/krjrbW7SjjGI3ifN+AW3GftD965NgDOstZ+UUZW/8d7/aVoEFKUF/AWzcMMfvt8PI2rxbgG9/nYDXyE+3x8X0Y+Aqbrf5+K9PcCWGeMKbjb3621g0I5joiET8P3iogEZ6b3eoN3w5nPGHMy0Bl3o7oo3AMZY17BtWFvD/iA+kAv4D1jTP8wkk46irzUA5YB9+POsSmu03RT4GLgTWPM38pIphbwIfA4kAzE4QKaO4H3jTF1jDH9gCW4gKeO99MB+D9T5I6xhDxOwL1fHXGduOsD5wPzjTEjQjtjMMa0xQuCvDwk4GqKfufl+UtjTPcSdp2JC4564Zrm7cUFI6cCN+MC2WAcwnVKL9jH4ZcCP/m1I14fjVW4oOIM3HsbhwsSHgZWGWPalHKsOOD/gEnA2bggLlSNjTF1jmI/v9q4wOthoDXu/JviahlXGmPODSNtv1381tEf3Htb8D0tFqyJSPlTICIiEgRrbSbuRjkJd5NbkL825DVrrS/MQ/XD1RoMBRKttQ2AFGAx7jt7ktc2P1hfA/6n0c8YY5JDzI8PmAdchzv3OtbaRFyn9jtxN8W3GWOuLiWNO3Ad4v8XOBaohws4duNu9P1P5l8DmltrG+JG9UrH3Sg/V8o5nwn8GRgLNLbWNvLy+Zq3/q/GmHOCPVljTAPvfFvhamzOBupaa+vhbpJneuc+1xjTsMB+5+IGM/ABw3Fl1xAXUDXHNY/7OJg8WGs/sdaegAus/MtOKPAzzjtmLWCul9cNQG/ce1sPVyvn79v0ljGmdoDDXQVcgiujRO/9Ox5Xo1IWfy3IMcBLxpj6wZxfCYbigqibgHreZ/4sXI1bAmCNMY2OMm0ArLV3e++pX8ci7+ndAXcWkXKjplkiIsH7B67z7R9wT/j98zzcWGB9uBoCN1pr/TfSWGvXGWOuBdYDJ+KaLi0OJjFr7T5jzDO4p81nAT8aYz4FPgE+A5aV1Myo4P64TvhFl+8EJhtjduHO+w7gzQDJNAD6WWszCixL9/L1GK62ZaG1dnCB9LcZY27AjYBU2jk3wDUPu7/Avj8bYwbiAoDzcDURwfbbuRdXa5MOXFWwg7jXlO4PxpgmQB/ckLX+PjddvNf3rbXPFdgnzzuHvwd5/FBcg7uBzwX6WGtXF1j3kTGmD7AS14/jBuDVEtKoB/zRWvtygTwHHKK6IGvtQmNMBtATF0QMMMYswjVDW477bAUzMlUDin/mvzDGXAyswQVGw/htGGMRqSJUIyIiEjyLa5Z0lTHmWG9ZT9wT6f9aa7+NwDF+AmYVO7C1P+MCB3BzNYTiEdx8FHuAmrib+hG48/nJGPO1MeZPxphjjiK///JeuxhjagbYZmmRIMTvwwK/P1V0pbV2L65ZGJR+zk+WsG9egTTP90ZwCoa/H8GEUkapet17vajAshzvtVnRpnvlaID3+naRIAQAb1AFf/+dQM3bfqXkACVYfXGBqA9X23UZLricD/xqjFnoBUSlyaLkz3w24G/2N6DoehGp/FQjIiISJGvtTmPMv4Crgf64G7CCndQj4b+l3ABv8l5Daqbivyk3xryAaxLVC+gEnIILTE7F9RG4xhhzsVcLks9rFnUT7rzb4fo/1KKwOl6+Spq74asAWSv45L3YjbTH364/0Dn/ZK1dF2Ddx7g+IzVxTbgWBNgOAGNMS6CF9+ebxphAzez8596ywLIPcX0bzgYWGWNeBhZYazeXdswwne29LixlmwW4ZnVnB1j/X2vtUY+UZq3NAW4yxozGNfPqgevb1Ar3sLMX0MsYM95ae0+AZDJK+cxn4ILo040xtQINqCAilZNqREREQuMPOAZ6Q5L2xw3t+3rgXUKyu5R1B7zXo6m5wFqbY639h7X2Fmvt6cBxuJtU/3DE5+A6aOfzOqtnAK/gOqefgLu538ZvHX39jqVkPwdYnj87uFfjU9o2gc55U4Dl/pGadnh/Ng20XQEnFvi9Ka5JUEk//qAoocCxfsT1ddiPuxmfCWwyxqwzxrxojDkriOOHyn9OAd8DYKP32sRrRljUtkhkxFr7k7X2OWttf2ttMi4QSQO2e5ukeQMSlKS0/PvX1STEAFxEYp9qREREQvMe7kn++cCfcKMiveM1I6lUvPb7bxhj3sZ1PD4N93T7ngKd7kfjmnJlA/cA7xXsQ+A1x/I/US/pRreihZKngg/nGnhP+4NmrX3VGPNv3GhP5+ECu2TgduCPxpiHrLXFmpFFQKCO6ME4UvYmobPW/gRM8N6Plbig7RZc35tQxOJnSkQiRDUiIiIh8JqxvIH7/vTXHswMvEfss9Ye4LdRphpRuPbAPxrWnV5tStGOzMeXd/7K0DzQCm9IWf/IVsE8+S9Yu3Pq0WTGWvuLtfZ5a+0VuPexE24+jzjgcWPMGUeTbgD+c2pVyjb+pma/VsTM7N4cIP7RwgLNrh6wDPmtluoIv9VuiUgVoUBERCR0/uZZx+Bujv5VyraVxd4Cvxdsh++/kV0ZYL+KnkW+VSlDEp+Da9KTB5Q1OR9eXxN/MHJVuBmz1uZZa5fjgrmNuP+5QQ8lHITPvdfzStnGP9T056VsU978n61A/Tt6lrKvf93qCPUP8QdjqmkRiQEKREREQmStXYEbiepZ4M/W2oMVm6PAjDHHGWPOLGObGrihYAGy/DOde/wTvf0PRXj9Rx6MSEbD80DRBV5/CP+Qvh9Za7cX3SaAGd7rHd7EhiUyxsR5c474/y7aeT+ftfYIrh8RhNeMqij/iFiXltQHxRhzGr+NNmUjeFx/+qcbY04oY5vj+S0YChQMJhtjrith38bAbd6fgYaGDpW/uV3DUrcSkahQHxERkaNgrX20ovMQpBNws1N/iGt+tdBamwX5TZe6AA/h+oEATCyy/we4Du3jjTHbgMXW2jxjTEdgMq7De0XKwU2o+Csw1lq7y7s5/itwAe4JeChl9TSuBiMFyDDG3AvMtdbugfyRtS7BdUyfyG+By5Nezcw/ce/Rdm/743HBWmsvLx8c/akWMxs3DPMZwNvGmMG4oCvPGHMBbljeY3CDEbwWOJmj1gs3SeYcXKCzpMB5JwKX4wL2BrjhfScHSGcXMNXrb/SGtTbXa8I2Dde8bSswJUJ5/hr3Wf+DMWaVFySKSAVRICIiUrXl4m6AL/R+MMYcBPZRfBSiKcBzRZY9hJsvoyWwCDhgjDmCGyFrP2444P+UU96DsdL7eQC4zxiTg3va7W96c5+1NqgZzSF/iOaLgXeAtrhA41VjzE6grvfjV7DPRTxuBLX+AF4+4oCCs40/VNJ8H0fLWnvIGNMfN3RwK1yQs88YA7+N6PUTbmLG8qi1O4wbtvlG7wdjzB7c+1LwvA8At3vN1EryIi6omQm84n0+E711+4Cri9TSheMVXCDyZ+B2Y8xWL79zrLUjInQMEQmSmmaJiFRh1to1uJvUYbgn6N/ibiATcUMFfwW8DHS11g4rMFqWf/9MXIfrf+KeTNcEduKesHe01r4fpVMJyFo7HBgErMAFBHtwc2tcaq0dV8qugdL7ETcL/R1eOttx71cusAo350pPCg9SMAG4Czcq1Pe4IKQ2sAH3vp9bHiNmeXlth5tEsGCQsxp4HDjD6zAecdbav+HmJ3kQN4FhFq4Gpi5uosRluMkm21prS5tZ/iCun8tjXhq1cB3x3wDOttYujmCepwNDcJOD5uIC7FZUfM2eSLUUl5cX9UE0REREwmKMGQRMx02G16ticyNHwxgzAzdR5qPW2kcqNjciUhFUIyIiIiIiIlGnQERERERERKJOgYiIiIiIiESdAhEREREREYk6dVYXEREREZGoU42IiIiIiIhEnQIRERERERGJOgUiIiIiIiISdQpEREREREQk6hSIiIiIiIhI1CkQERERERGRqFMgIiIiIiIiUadAREREREREok6BiIiIiIiIRJ0CERERERERiToFIiIiIiIiEnUKREREREREJOoUiIiIiIiISNT9f0TLBigMBv8/AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 281,
       "width": 401
      },
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# visualizing the grid search of DT\n",
    "\n",
    "for i in results_DT['max_depth'].unique():\n",
    "    temp = results_DT[results_DT['max_depth'] == i]\n",
    "    plt.plot(temp['min_samples_split'], temp['mean_score'], marker = '.', label = i)\n",
    "    \n",
    "plt.legend(title = \"Max Depth\")\n",
    "plt.xlabel('Min Samples for Split')\n",
    "plt.ylabel(\"AUC Score\")\n",
    "plt.title(\"DT Performance Comparison\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We see that the AUC score increases on increasing the max_depth, however the AUC score doesn't change much on increasing the min_samples_split."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Naive Bayes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For tuning Naive Bayes model we are using only var_smoothing as hyper-parameter, in order to do that we are using Randomized Search instead of grid search because this helps us by choosing random combinations hyper-parameter values unlike grid search and it saves the running time for the purpose of this assignment. Before fitting the NB model the descriptive features are converted to Gaussian distribution."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import PowerTransformer\n",
    "\n",
    "# Converting descriptive features to GAussian distribution\n",
    "X_train_transformed = PowerTransformer().fit_transform(X_train)\n",
    "\n",
    "from sklearn.naive_bayes import GaussianNB\n",
    "from sklearn.model_selection import RandomizedSearchCV\n",
    "\n",
    "params_NB = {'var_smoothing': np.logspace(1,-9, num=100)}\n",
    "\n",
    "n_iter_search = 20\n",
    "gs_NB = RandomizedSearchCV(estimator=GaussianNB(), \n",
    "                          param_distributions=params_NB, \n",
    "                          cv=cv_method,\n",
    "                          refit=True,\n",
    "                          n_jobs=-2,\n",
    "                          scoring='roc_auc',\n",
    "                          n_iter=n_iter_search,\n",
    "                          verbose=1) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 20 candidates, totalling 100 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-2)]: Using backend LokyBackend with 3 concurrent workers.\n",
      "[Parallel(n_jobs=-2)]: Done 100 out of 100 | elapsed:    1.1s finished\n"
     ]
    }
   ],
   "source": [
    "# fitting the NB model\n",
    "gs_NB.fit(X_train_transformed, y_train);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'var_smoothing': 6.280291441834253}"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# displaying the best parameters for NB\n",
    "gs_NB.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7389447370017501"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# displaying the AUC score of the NB model\n",
    "gs_NB.best_score_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The optimal Naive Bayes model achieves an AUC score of 0.739.\n",
    "\n",
    "Now, visualizing the search results of NB model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "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>mean_score</th>\n",
       "      <th>std_score</th>\n",
       "      <th>max_score</th>\n",
       "      <th>min_score</th>\n",
       "      <th>var_smoothing</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.738944</td>\n",
       "      <td>0.016005</td>\n",
       "      <td>0.769200</td>\n",
       "      <td>0.726272</td>\n",
       "      <td>6.280291</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.738324</td>\n",
       "      <td>0.015968</td>\n",
       "      <td>0.768452</td>\n",
       "      <td>0.724093</td>\n",
       "      <td>10.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.736037</td>\n",
       "      <td>0.017456</td>\n",
       "      <td>0.768747</td>\n",
       "      <td>0.717565</td>\n",
       "      <td>0.977010</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.733570</td>\n",
       "      <td>0.018435</td>\n",
       "      <td>0.768158</td>\n",
       "      <td>0.713596</td>\n",
       "      <td>0.151991</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.733470</td>\n",
       "      <td>0.018467</td>\n",
       "      <td>0.768136</td>\n",
       "      <td>0.713504</td>\n",
       "      <td>0.120450</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    mean_score  std_score  max_score  min_score  var_smoothing\n",
       "16    0.738944   0.016005   0.769200   0.726272       6.280291\n",
       "0     0.738324   0.015968   0.768452   0.724093      10.000000\n",
       "10    0.736037   0.017456   0.768747   0.717565       0.977010\n",
       "9     0.733570   0.018435   0.768158   0.713596       0.151991\n",
       "4     0.733470   0.018467   0.768136   0.713504       0.120450"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# converting search results to dataframe and display dataframe\n",
    "results_NB = get_search_results(gs_NB)\n",
    "results_NB.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 281,
       "width": 400
      },
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# visualizing the hyper-parameter tuning of NB\n",
    "\n",
    "plt.plot(results_NB['var_smoothing'], results_NB['mean_score'], marker = '.', label = i)    \n",
    "plt.xlabel('Var. Smoothing')\n",
    "plt.ylabel(\"AUC Score\")\n",
    "plt.title(\"NB Performance Comparison\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "From this graph we could see that the AUC score of NB increases slightly on increasing the var_smoothing."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Random Forest Classifier"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In order to fit and tune Random forest classifier we considered max_features = 'sqrt' and bootstrap = True, we implemented Randomized search for hyper-parameter tuning along with 5-fold stratified cross-validation for the purpose of this assignment to save the run time. We considered following hyper-parameters:\n",
    "* max_depth\n",
    "* min_samples_split\n",
    "* min_samples_leaf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 20 candidates, totalling 100 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 4 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done  76 tasks      | elapsed:    2.3s\n",
      "[Parallel(n_jobs=-1)]: Done 100 out of 100 | elapsed:    3.1s finished\n"
     ]
    }
   ],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "params_RF = {'max_depth': [3, 4, 5, 10],\n",
    "             'min_samples_split': [2, 5, 10],\n",
    "             'min_samples_leaf': [1, 2, 4]}\n",
    "\n",
    "gs_RF = RandomizedSearchCV(estimator=RandomForestClassifier(max_features = 'sqrt', criterion='gini', bootstrap = True, random_state = seed_value), \n",
    "                          param_distributions=params_RF, \n",
    "                          cv=cv_method,\n",
    "                          refit=True,\n",
    "                          n_jobs=-1,\n",
    "                          scoring='roc_auc',\n",
    "                          n_iter=20,\n",
    "                          verbose=1) \n",
    "\n",
    "# fitting the model\n",
    "gs_RF.fit(X_train, y_train);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'min_samples_split': 10, 'min_samples_leaf': 2, 'max_depth': 10}"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# display the best hyperparameters for Random forest classifier\n",
    "gs_RF.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7732882080919651"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# display the AUC score for the optimal Random forest classifer\n",
    "gs_RF.best_score_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "On performing hyper-parameter tuning to the Random forest classifier we achieved the best AUC score of 0.773 which is better than the above KNN, DT and NB models. However, for further model comparison we may need to compare their performances in the test dataset using other evaluation techniques such as paired t-tests. This is done in the model evaluation section of this report.\n",
    "\n",
    "Now, visualizing the search results of Random forest classifier."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "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>mean_score</th>\n",
       "      <th>std_score</th>\n",
       "      <th>max_score</th>\n",
       "      <th>min_score</th>\n",
       "      <th>min_samples_split</th>\n",
       "      <th>min_samples_leaf</th>\n",
       "      <th>max_depth</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.773286</td>\n",
       "      <td>0.010388</td>\n",
       "      <td>0.793726</td>\n",
       "      <td>0.765549</td>\n",
       "      <td>10.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>10.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.772115</td>\n",
       "      <td>0.009355</td>\n",
       "      <td>0.790339</td>\n",
       "      <td>0.765144</td>\n",
       "      <td>5.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>10.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.772115</td>\n",
       "      <td>0.009355</td>\n",
       "      <td>0.790339</td>\n",
       "      <td>0.765144</td>\n",
       "      <td>2.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>10.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.770070</td>\n",
       "      <td>0.013865</td>\n",
       "      <td>0.796159</td>\n",
       "      <td>0.757037</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>10.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.770055</td>\n",
       "      <td>0.012138</td>\n",
       "      <td>0.792369</td>\n",
       "      <td>0.757480</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>10.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    mean_score  std_score  max_score  min_score  min_samples_split  \\\n",
       "6     0.773286   0.010388   0.793726   0.765549               10.0   \n",
       "8     0.772115   0.009355   0.790339   0.765144                5.0   \n",
       "12    0.772115   0.009355   0.790339   0.765144                2.0   \n",
       "3     0.770070   0.013865   0.796159   0.757037                5.0   \n",
       "10    0.770055   0.012138   0.792369   0.757480                2.0   \n",
       "\n",
       "    min_samples_leaf  max_depth  \n",
       "6                2.0       10.0  \n",
       "8                4.0       10.0  \n",
       "12               4.0       10.0  \n",
       "3                1.0       10.0  \n",
       "10               1.0       10.0  "
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# converting the search results into dataframe\n",
    "results_RF = get_search_results(gs_RF)\n",
    "\n",
    "# display the dataframe\n",
    "results_RF.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 281,
       "width": 394
      },
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "for i in results_RF['max_depth'].unique():\n",
    "    temp = results_RF[results_RF['max_depth'] == i]\n",
    "    plt.plot(temp['min_samples_split'], temp['mean_score'], marker = '.', label = i)\n",
    "    \n",
    "plt.legend(title = \"Max Depth\")\n",
    "plt.xlabel('Min Samples for Split')\n",
    "plt.ylabel(\"AUC Score\")\n",
    "plt.title(\"RF Performance Comparison\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The AUC score of RF with max_depth = 10 is higher that with max_depth = 3. However, there is no change in score on increasing only the min_sample_split."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Support Vector Classification"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For support vector classification tuning we considered the below hyper-parameters using grid search\n",
    "* C\n",
    "* degree"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn import svm\n",
    "\n",
    "\n",
    "params_SVC = {'C': [1.0, 1.5, 2.0] ,\n",
    "             'degree': [1, 2, 3]}\n",
    "\n",
    "gs_SVC = GridSearchCV(estimator=svm.SVC(), \n",
    "                           param_grid=params_SVC, \n",
    "                           cv=cv_method,\n",
    "                           refit=True,\n",
    "                           n_jobs=-2,\n",
    "                           scoring='roc_auc',\n",
    "                           verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 9 candidates, totalling 45 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-2)]: Using backend LokyBackend with 3 concurrent workers.\n",
      "[Parallel(n_jobs=-2)]: Done  45 out of  45 | elapsed:  2.1min finished\n"
     ]
    }
   ],
   "source": [
    "# fitting SVC model\n",
    "gs_SVC.fit(X_train, y_train);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'C': 2.0, 'degree': 1}"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# display the best parameters of optimal SVC model\n",
    "gs_SVC.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7622966578534639"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# display the AUC score of optimal SVC model\n",
    "gs_SVC.best_score_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The AUC score of the optimal SVC model is 0.762.\n",
    "\n",
    "Visualizing the grid search output below."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>mean_score</th>\n",
       "      <th>std_score</th>\n",
       "      <th>max_score</th>\n",
       "      <th>min_score</th>\n",
       "      <th>C</th>\n",
       "      <th>degree</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.762295</td>\n",
       "      <td>0.009735</td>\n",
       "      <td>0.778045</td>\n",
       "      <td>0.747745</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.762295</td>\n",
       "      <td>0.009735</td>\n",
       "      <td>0.778045</td>\n",
       "      <td>0.747745</td>\n",
       "      <td>2.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.762295</td>\n",
       "      <td>0.009735</td>\n",
       "      <td>0.778045</td>\n",
       "      <td>0.747745</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.761828</td>\n",
       "      <td>0.010259</td>\n",
       "      <td>0.779145</td>\n",
       "      <td>0.747660</td>\n",
       "      <td>1.5</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.761828</td>\n",
       "      <td>0.010259</td>\n",
       "      <td>0.779145</td>\n",
       "      <td>0.747660</td>\n",
       "      <td>1.5</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   mean_score  std_score  max_score  min_score    C  degree\n",
       "6    0.762295   0.009735   0.778045   0.747745  2.0     1.0\n",
       "7    0.762295   0.009735   0.778045   0.747745  2.0     2.0\n",
       "8    0.762295   0.009735   0.778045   0.747745  2.0     3.0\n",
       "3    0.761828   0.010259   0.779145   0.747660  1.5     1.0\n",
       "4    0.761828   0.010259   0.779145   0.747660  1.5     2.0"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# format grid search output into dataframe and then display \n",
    "results_SVC = get_search_results(gs_SVC)\n",
    "results_SVC.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 281,
       "width": 413
      },
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# visualizing the combination of C and degree during hypertuning of SVC model\n",
    "for i in results_SVC['C'].unique():\n",
    "    temp = results_SVC[results_SVC['C'] == i]\n",
    "    plt.plot(temp['degree'], temp['mean_score'], marker = '.', label = i)\n",
    "    \n",
    "plt.legend(title = \"C value\")\n",
    "plt.xlabel('Degree')\n",
    "plt.ylabel(\"AUC Score\")\n",
    "plt.title(\"SVC Performance Comparison\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The above graph indicates that their is no change in AUC score on increasing the degree. However, the accuracy increases slightly on increasing the value of C. So, the AUC score of SVC is high for C value 2 than 1."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Model Comparison <a id='mc'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The above KNN, DT, NB, RFC and SVC models were trained and tuned with the training dataset using 5-fold stratified cross-validation method.Now, we would like to compare the performances of those tuned models using paired t-test. In order to do that  we will perform 5-fold cross-validation to those 5 tuned models with our test dataset, then we will perform paired t-tests among the combination of models to determine if the performance of one model is better than the other and prove it statistically significant. During the process used AUC score to compare the model and also we are supposed to use the same seed for cross-validation for all the 5 models."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import cross_val_score\n",
    "\n",
    "# defining cross-validation method\n",
    "cv_method_ttest = StratifiedKFold(n_splits=5, random_state=seed_value)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7489279524037676"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cv_results_KNN = cross_val_score(estimator=gs_KNN.best_estimator_,\n",
    "                                 X=X_test,\n",
    "                                 y=y_test, \n",
    "                                 cv=cv_method_ttest, \n",
    "                                 n_jobs=-2,\n",
    "                                 scoring='roc_auc')\n",
    "cv_results_KNN.mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The cross validation score for the tuned KNN model on our test set is 0.749 while using our train set was 0.753."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7090517313036963"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cv_results_DT = cross_val_score(estimator=gs_DT.best_estimator_,\n",
    "                                X=X_test,\n",
    "                                y=y_test, \n",
    "                                cv=cv_method_ttest, \n",
    "                                n_jobs=-2,\n",
    "                                scoring='roc_auc')\n",
    "cv_results_DT.mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The cross validation score for the tuned DT model on our test set is 0.709 while using our train set was 0.720."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.6164754157818573"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cv_results_NB = cross_val_score(estimator=gs_NB.best_estimator_,\n",
    "                                X=X_test,\n",
    "                                y=y_test, \n",
    "                                cv=cv_method_ttest, \n",
    "                                n_jobs=-2,\n",
    "                                scoring='roc_auc')\n",
    "cv_results_NB.mean()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The cross validation score for the tuned NB model on our test set is 0.616 while using our train set was 0.739. Our tune Naive bayes model is found to be over-fitting, so we simply ignore NB model for the model comparison for the purpose of our assignment. However, this might need to retrain and tune the model using over-fitting preventing techniques which has not been covered in this report."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7728593613857612"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cv_results_RF = cross_val_score(estimator=gs_RF.best_estimator_,\n",
    "                                X=X_test,\n",
    "                                y=y_test, \n",
    "                                cv=cv_method_ttest, \n",
    "                                n_jobs=-2,\n",
    "                                scoring='roc_auc')\n",
    "cv_results_RF.mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The cross validation score for the tuned Random forest classifier on our test set is 0.773 while using our train set was 0.773."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7582933405817631"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cv_results_SVC = cross_val_score(estimator=gs_SVC.best_estimator_,\n",
    "                                X=X_test,\n",
    "                                y=y_test, \n",
    "                                cv=cv_method_ttest, \n",
    "                                n_jobs=-2,\n",
    "                                scoring='roc_auc')\n",
    "cv_results_SVC.mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The cross validation score for the tuned SVC on our test set is 0.758 while using our train set was 0.762.\n",
    "\n",
    "Now, we will compare the performances of KNN, DT, RF and SVC (note: we ignored NB because of over-fitting) using paired t-tests with stats.ttest_rel function from the SciPy module."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ttest_relResult(statistic=4.597509114851691, pvalue=0.010049741937848836)\n"
     ]
    }
   ],
   "source": [
    "from scipy import stats\n",
    "\n",
    "# paired t-test between KNN and DT\n",
    "print(stats.ttest_rel(cv_results_KNN, cv_results_DT))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Cross-validation score on test set for KNN was 0.749 and for DT it was 0.709 and the paired t-test between these 2 model indicates p-value < 0.05. So we can say that KNN is statistically better than the DT.\n",
    "\n",
    "The cross-validation score of Random forest is 0.773 on test set which is better than that of KNN. So let's check whether the difference is statistically significant."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ttest_relResult(statistic=-3.8273027348307234, pvalue=0.018661146491933196)\n"
     ]
    }
   ],
   "source": [
    "print(stats.ttest_rel(cv_results_KNN, cv_results_RF))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As the p-value is really really less than 0.05 so it shows the performance of Random forest is far better than KNN."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ttest_relResult(statistic=1.4556949592854929, pvalue=0.21918219680151954)\n"
     ]
    }
   ],
   "source": [
    "print(stats.ttest_rel(cv_results_RF, cv_results_SVC))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The cross validation score of SVC is 0.758 which is little less than than that of Random forest and the paired t-test indictes p-value greater than 0.05 so we can say that the AUC score of Random forest is not significantly better than SVC.\n",
    "\n",
    "From all the above results, we can conclude that Random forest and SVC perform equally well and they are statistically better than the other models."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For the purpose of this assignment we were supposed to compare the models using paired-t test, but there are other ways to evaluate models based on classification report (metrics such as Accuracy, Precision, Recall, F-1 Score), confusion matrix, ROC curves etc. \n",
    "\n",
    "Below is the classification report for all the above 5 models:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "# predicting using the test data\n",
    "pred_KNN = gs_KNN.predict(X_test)\n",
    "pred_DT = gs_DT.predict(X_test)\n",
    "pred_NB = gs_NB.predict(X_test)\n",
    "pred_RF = gs_RF.predict(X_test)\n",
    "pred_SVC = gs_SVC.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Classification report for K-Nearest Neighbor\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.69      0.82      0.75      1772\n",
      "           1       0.74      0.59      0.66      1577\n",
      "\n",
      "    accuracy                           0.71      3349\n",
      "   macro avg       0.72      0.70      0.70      3349\n",
      "weighted avg       0.72      0.71      0.71      3349\n",
      "\n",
      "\n",
      "Classification report for Decision Tree\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.66      0.83      0.73      1772\n",
      "           1       0.73      0.52      0.61      1577\n",
      "\n",
      "    accuracy                           0.68      3349\n",
      "   macro avg       0.69      0.67      0.67      3349\n",
      "weighted avg       0.69      0.68      0.67      3349\n",
      "\n",
      "\n",
      "Classification report for Naive Bayes\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.59      0.84      0.69      1772\n",
      "           1       0.66      0.36      0.46      1577\n",
      "\n",
      "    accuracy                           0.61      3349\n",
      "   macro avg       0.63      0.60      0.58      3349\n",
      "weighted avg       0.63      0.61      0.59      3349\n",
      "\n",
      "\n",
      "Classification report for Random Forest\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.70      0.84      0.76      1772\n",
      "           1       0.76      0.59      0.67      1577\n",
      "\n",
      "    accuracy                           0.72      3349\n",
      "   macro avg       0.73      0.71      0.71      3349\n",
      "weighted avg       0.73      0.72      0.72      3349\n",
      "\n",
      "\n",
      "Classification report for SVC\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.70      0.84      0.76      1772\n",
      "           1       0.77      0.59      0.67      1577\n",
      "\n",
      "    accuracy                           0.72      3349\n",
      "   macro avg       0.73      0.71      0.71      3349\n",
      "weighted avg       0.73      0.72      0.72      3349\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn import metrics\n",
    "\n",
    "# display classification report for all the 5 models\n",
    "print(\"\\nClassification report for K-Nearest Neighbor\") \n",
    "print(metrics.classification_report(y_test, pred_KNN))\n",
    "print(\"\\nClassification report for Decision Tree\") \n",
    "print(metrics.classification_report(y_test, pred_DT))\n",
    "print(\"\\nClassification report for Naive Bayes\") \n",
    "print(metrics.classification_report(y_test, pred_NB))\n",
    "print(\"\\nClassification report for Random Forest\") \n",
    "print(metrics.classification_report(y_test, pred_RF))\n",
    "print(\"\\nClassification report for SVC\") \n",
    "print(metrics.classification_report(y_test, pred_SVC))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Looking at the precision and recall for both positive and negative classes of all the models we can say Random Forest classifier and Support Vector classifier performed equally well which supports our earlier model evaluation using paired t-tests."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## CRITIQUE & LIMITATIONS <a id='cl'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The important thing to note is there is no over-fitting with the Random Forest Classifier built which is also one of the model that performed better than the other models. Therefore, we can further improve the performance of Random Forest classifier with other hyper-parameter searching and tuning the model.\n",
    "\n",
    "We used whole set of features during predictive modelling, but in large dataset with lots of features this may not be an appropriate solution. Too many features is sometime bad it results in over-fitting. Our approach needs deeper analysis of feature selection and hyper-parameter tuning. The AUC score of Naive Byes decreases on the test dataset this might also need further investigation which is not discussed in this report. The other limitation is we are using the old dataset, the owner of the dataset has come with additional features to improve the performance of the classifier."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## SUMMARY & CONCLUSIONS <a id='sc'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Project Summary <a id='ps'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The main goal of the whole project was to explore and build a predictive model to predict the customers who are more likely to subscribe for the term deposit in the bank. In Phase 1 we focused more on the data preparation and exploration.All the cleansing of the data were done in this phase. The insights from data explorations were mentioned in Phase 1 report. \n",
    "\n",
    "In phase two, we used the already cleansed dataset to build predictive models. Further data transformations to suit the functionalities of models, such as scaling and encoding were done which was followed by the model building. We built 5 models \n",
    "KNN, DT, NB, RFC and SVC. Necessary hyperparameter tuning for the models were done as well and all these models were evaluated using comparison paired t-tests."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Summary of Findings <a id='sof'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The cross-validation AUC score of Random Forest classifier on both train and test data is found to be the best among other models. However, on performing paired t-tests there was not statistically significant differences between Random Forest classifier and Support Vector Classification this is also supported by comparing the precision and recall metrics from the classification report of all the models. We also found that there is no over-fitting in the Random Forest classifier but for some reason there is a huge over-fitting in the Naive Bayes classifier."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Conclusions <a id='c'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "At present, businesses have started looking towards automation and AI to improve their business processes. Specially banks who has been more keen on AI and Data Science. As our goal of the project was to predict whether the customer will subscribe for the term deposit or not during the bank marketing campaigns we can conclude that using the built Random Forest Classifier can help achieve the goal. However, the model is to predict the outcome of the marketing campaigns done on a phone call. For further improvements on the marketing strategies, the banks would have to do necessary data explorations."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## REFERENCES <a id='ref'></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Aksakalli, V., Yenice, Z., Wong, Y., Ture, I., & Malekipirbazari, M. (2021). Feature Selection and Ranking in Machine Learning | www.featureranking.com. www.featureranking.com. Retrieved 28 May 2021, from https://www.featureranking.com/.\n",
    "\n",
    "Janio Martinez, B. (2021). Bank Marketing Dataset. Kaggle.com. Retrieved 5 April 2021, from https://www.kaggle.com/janiobachmann/bank-marketing-dataset.\n",
    "\n",
    "S. Moro, P. Cortez, & P.Rita (2021). UCI Machine Learning Repository: Bank Marketing Data Set. Archive.ics.uci.edu. Retrieved 5 April 2021, from https://archive.ics.uci.edu/ml/datasets/Bank+Marketing.\n",
    "\n",
    "S. Moro, P. Cortez & P. Rita. A Data-Driven Approach to Predict the Success of Bank Telemarketing. Decision Support Systems, Elsevier, 62:22-31, June 2014.\n",
    "\n",
    "scikit-learn: machine learning in Python — scikit-learn 0.16.1 documentation. Scikit-learn.org. (2021). Retrieved 31 May 2021, from https://scikit-learn.org."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}