Part1 Disease Prediction.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Disease Prediction: NBC, KNN, SVM, and Ensemble Learning\n",
    "\n",
    "<b>Problem Set</b>: \n",
    "\n",
    "In this project, naive Bayes classifier, K Nearest Neighbor, Support Vector Machine (with both linear and non-linear kernel functions), Random Forest and Gradient Boosting Classifier are used to build a disease diagnosis model. It is a binary classification problem to predict whether or not a patient has a certain unspecified disease.\n",
    "\n",
    "<b>Dataset</b>: \n",
    "\n",
    "Attributes’ information about the dataset (Disease Prediction Training.csv):\n",
    "- Age: in years\n",
    "- Gender: male/female\n",
    "- Height: in unit of cm\n",
    "- Weight: in unit of kg\n",
    "- Low Blood Pressure: lower bound of blood pressure measurement\n",
    "- High Blood Pressure: higher bound of blood pressure measurement\n",
    "- Cholesterol: three cholesteral levels\n",
    "- Glucose: three glucose levels\n",
    "- Smoke: 1/0 regarding if the patient smokes\n",
    "- Alcohol: 1/0 regarding if the patient drinks alcohol\n",
    "- Exercise: 1/0 regarding if the patient exercises regularly\n",
    "- Disease: The binary target variable. Does the patient have the disease?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Section 0:  Import Libraries and Load Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load packages and data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# import necessary libraries\n",
    "import pandas as pd\n",
    "from pandas import DataFrame, Series\n",
    "import numpy as np\n",
    "\n",
    "# data formating \n",
    "pd.set_option('display.max_columns',None)\n",
    "pd.set_option('display.max_rows',None)\n",
    "pd.set_option('max_colwidth',100)\n",
    "\n",
    "# data visualization\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "metadata": {},
   "outputs": [],
   "source": [
    "# read libaries for machine learning model \n",
    "from sklearn import metrics\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn import preprocessing\n",
    "from sklearn.model_selection import GridSearchCV, StratifiedKFold,cross_val_score, ShuffleSplit\n",
    "from sklearn.naive_bayes import GaussianNB, BernoulliNB, MultinomialNB\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from scipy.spatial.distance import cdist\n",
    "from sklearn.pipeline import make_pipeline\n",
    "from sklearn import model_selection \n",
    "from matplotlib.legend_handler import HandlerLine2D\n",
    "from sklearn.decomposition import PCA\n",
    "from sklearn.svm import SVC\n",
    "from sklearn import svm\n",
    "from sklearn.metrics import roc_curve"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 632,
   "metadata": {},
   "outputs": [],
   "source": [
    "# read data into a pandas dataframe\n",
    "test_df = pd.read_csv(\"Disease Prediction Testing.csv\",sep=',')\n",
    "train_df = pd.read_csv(\"Disease Prediction Training.csv\",sep=',')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Reading and Displaying Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Testing dataset has 21000 rows and 11 columns, aftering seting the ID column as index.\n",
    "- Column Labels: Each column label is unique \n",
    "- Empty Column: All columns are non-null\n",
    "- Type of Values: float64(1), int64(7), object(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 245,
   "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>Gender</th>\n",
       "      <th>Height</th>\n",
       "      <th>Weight</th>\n",
       "      <th>High Blood Pressure</th>\n",
       "      <th>Low Blood Pressure</th>\n",
       "      <th>Cholesterol</th>\n",
       "      <th>Glucose</th>\n",
       "      <th>Smoke</th>\n",
       "      <th>Alcohol</th>\n",
       "      <th>Exercise</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ID</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>44</td>\n",
       "      <td>female</td>\n",
       "      <td>160</td>\n",
       "      <td>59.0</td>\n",
       "      <td>100</td>\n",
       "      <td>80</td>\n",
       "      <td>high</td>\n",
       "      <td>normal</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>41</td>\n",
       "      <td>female</td>\n",
       "      <td>169</td>\n",
       "      <td>74.0</td>\n",
       "      <td>120</td>\n",
       "      <td>70</td>\n",
       "      <td>normal</td>\n",
       "      <td>normal</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>63</td>\n",
       "      <td>male</td>\n",
       "      <td>168</td>\n",
       "      <td>84.0</td>\n",
       "      <td>120</td>\n",
       "      <td>80</td>\n",
       "      <td>normal</td>\n",
       "      <td>high</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    Age  Gender  Height  Weight  High Blood Pressure  Low Blood Pressure  \\\n",
       "ID                                                                         \n",
       "0    44  female     160    59.0                  100                  80   \n",
       "1    41  female     169    74.0                  120                  70   \n",
       "2    63    male     168    84.0                  120                  80   \n",
       "\n",
       "   Cholesterol Glucose  Smoke  Alcohol  Exercise  \n",
       "ID                                                \n",
       "0         high  normal      0        0         1  \n",
       "1       normal  normal      0        0         1  \n",
       "2       normal    high      0        0         1  "
      ]
     },
     "execution_count": 245,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_df.set_index('ID', inplace=True)\n",
    "test_df.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 246,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 21000 entries, 0 to 20999\n",
      "Data columns (total 11 columns):\n",
      " #   Column               Non-Null Count  Dtype  \n",
      "---  ------               --------------  -----  \n",
      " 0   Age                  21000 non-null  int64  \n",
      " 1   Gender               21000 non-null  object \n",
      " 2   Height               21000 non-null  int64  \n",
      " 3   Weight               21000 non-null  float64\n",
      " 4   High Blood Pressure  21000 non-null  int64  \n",
      " 5   Low Blood Pressure   21000 non-null  int64  \n",
      " 6   Cholesterol          21000 non-null  object \n",
      " 7   Glucose              21000 non-null  object \n",
      " 8   Smoke                21000 non-null  int64  \n",
      " 9   Alcohol              21000 non-null  int64  \n",
      " 10  Exercise             21000 non-null  int64  \n",
      "dtypes: float64(1), int64(7), object(3)\n",
      "memory usage: 1.9+ MB\n"
     ]
    }
   ],
   "source": [
    "test_df.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Training Dataset has 49,000 rows and 12 columns.\n",
    "- Column Labels: Each column label is unique \n",
    "- Empty Column: All columns are non-null\n",
    "- Type of Values: float64(1), int64(8), object(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "scrolled": false
   },
   "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>Gender</th>\n",
       "      <th>Height</th>\n",
       "      <th>Weight</th>\n",
       "      <th>High Blood Pressure</th>\n",
       "      <th>Low Blood Pressure</th>\n",
       "      <th>Cholesterol</th>\n",
       "      <th>Glucose</th>\n",
       "      <th>Smoke</th>\n",
       "      <th>Alcohol</th>\n",
       "      <th>Exercise</th>\n",
       "      <th>Disease</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>59</td>\n",
       "      <td>female</td>\n",
       "      <td>167</td>\n",
       "      <td>88.0</td>\n",
       "      <td>130</td>\n",
       "      <td>68</td>\n",
       "      <td>normal</td>\n",
       "      <td>normal</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>64</td>\n",
       "      <td>female</td>\n",
       "      <td>150</td>\n",
       "      <td>71.0</td>\n",
       "      <td>140</td>\n",
       "      <td>100</td>\n",
       "      <td>normal</td>\n",
       "      <td>normal</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>41</td>\n",
       "      <td>female</td>\n",
       "      <td>166</td>\n",
       "      <td>83.0</td>\n",
       "      <td>100</td>\n",
       "      <td>70</td>\n",
       "      <td>normal</td>\n",
       "      <td>normal</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Age  Gender  Height  Weight  High Blood Pressure  Low Blood Pressure  \\\n",
       "0   59  female     167    88.0                  130                  68   \n",
       "1   64  female     150    71.0                  140                 100   \n",
       "2   41  female     166    83.0                  100                  70   \n",
       "\n",
       "  Cholesterol Glucose  Smoke  Alcohol  Exercise  Disease  \n",
       "0      normal  normal      0        0         1        0  \n",
       "1      normal  normal      0        0         0        1  \n",
       "2      normal  normal      0        1         1        0  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 49000 entries, 0 to 48999\n",
      "Data columns (total 12 columns):\n",
      " #   Column               Non-Null Count  Dtype  \n",
      "---  ------               --------------  -----  \n",
      " 0   Age                  49000 non-null  int64  \n",
      " 1   Gender               49000 non-null  object \n",
      " 2   Height               49000 non-null  int64  \n",
      " 3   Weight               49000 non-null  float64\n",
      " 4   High Blood Pressure  49000 non-null  int64  \n",
      " 5   Low Blood Pressure   49000 non-null  int64  \n",
      " 6   Cholesterol          49000 non-null  object \n",
      " 7   Glucose              49000 non-null  object \n",
      " 8   Smoke                49000 non-null  int64  \n",
      " 9   Alcohol              49000 non-null  int64  \n",
      " 10  Exercise             49000 non-null  int64  \n",
      " 11  Disease              49000 non-null  int64  \n",
      "dtypes: float64(1), int64(8), object(3)\n",
      "memory usage: 4.5+ MB\n"
     ]
    }
   ],
   "source": [
    "train_df.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Section1: Data Preparation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1.1 Derive Descriptive Statistics"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 1.1.1 Distribution of Numeric Variables \n",
    "- **The type of variables we have:** Age is a discrete numeric varaible. Height, Weight, High Blood Pressure, and Low Blood Pressure are continous numeric variables. The last four variables - Smoke, Alcohol, Exercise, and Disease have binary values.\n",
    "- **We have a balanced Dataset:** According to the distribution of the target variable \"Disease\", the number of cases of with disease and without disease are roughtly the same.\n",
    "- **Problem1**: Here we notice that blood pressure has negative values, indicating that there are error data in the High Blood Pressure column.\n",
    "- **Problem2**: The naming of \"High Blood Pressure\" and \"Low Blood Pressure\" have spaces."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Age</th>\n",
       "      <td>49000.0</td>\n",
       "      <td>52.853306</td>\n",
       "      <td>6.763065</td>\n",
       "      <td>29.0</td>\n",
       "      <td>48.0</td>\n",
       "      <td>53.0</td>\n",
       "      <td>58.0</td>\n",
       "      <td>64.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Height</th>\n",
       "      <td>49000.0</td>\n",
       "      <td>164.366878</td>\n",
       "      <td>8.216637</td>\n",
       "      <td>55.0</td>\n",
       "      <td>159.0</td>\n",
       "      <td>165.0</td>\n",
       "      <td>170.0</td>\n",
       "      <td>207.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Weight</th>\n",
       "      <td>49000.0</td>\n",
       "      <td>74.190527</td>\n",
       "      <td>14.329934</td>\n",
       "      <td>10.0</td>\n",
       "      <td>65.0</td>\n",
       "      <td>72.0</td>\n",
       "      <td>82.0</td>\n",
       "      <td>200.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>High Blood Pressure</th>\n",
       "      <td>49000.0</td>\n",
       "      <td>128.698939</td>\n",
       "      <td>147.624582</td>\n",
       "      <td>-150.0</td>\n",
       "      <td>120.0</td>\n",
       "      <td>120.0</td>\n",
       "      <td>140.0</td>\n",
       "      <td>14020.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Low Blood Pressure</th>\n",
       "      <td>49000.0</td>\n",
       "      <td>96.917367</td>\n",
       "      <td>200.368069</td>\n",
       "      <td>0.0</td>\n",
       "      <td>80.0</td>\n",
       "      <td>80.0</td>\n",
       "      <td>90.0</td>\n",
       "      <td>11000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Smoke</th>\n",
       "      <td>49000.0</td>\n",
       "      <td>0.088265</td>\n",
       "      <td>0.283683</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Alcohol</th>\n",
       "      <td>49000.0</td>\n",
       "      <td>0.054245</td>\n",
       "      <td>0.226503</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Exercise</th>\n",
       "      <td>49000.0</td>\n",
       "      <td>0.803204</td>\n",
       "      <td>0.397581</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Disease</th>\n",
       "      <td>49000.0</td>\n",
       "      <td>0.499959</td>\n",
       "      <td>0.500005</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                       count        mean         std    min    25%    50%  \\\n",
       "Age                  49000.0   52.853306    6.763065   29.0   48.0   53.0   \n",
       "Height               49000.0  164.366878    8.216637   55.0  159.0  165.0   \n",
       "Weight               49000.0   74.190527   14.329934   10.0   65.0   72.0   \n",
       "High Blood Pressure  49000.0  128.698939  147.624582 -150.0  120.0  120.0   \n",
       "Low Blood Pressure   49000.0   96.917367  200.368069    0.0   80.0   80.0   \n",
       "Smoke                49000.0    0.088265    0.283683    0.0    0.0    0.0   \n",
       "Alcohol              49000.0    0.054245    0.226503    0.0    0.0    0.0   \n",
       "Exercise             49000.0    0.803204    0.397581    0.0    1.0    1.0   \n",
       "Disease              49000.0    0.499959    0.500005    0.0    0.0    0.0   \n",
       "\n",
       "                       75%      max  \n",
       "Age                   58.0     64.0  \n",
       "Height               170.0    207.0  \n",
       "Weight                82.0    200.0  \n",
       "High Blood Pressure  140.0  14020.0  \n",
       "Low Blood Pressure    90.0  11000.0  \n",
       "Smoke                  0.0      1.0  \n",
       "Alcohol                0.0      1.0  \n",
       "Exercise               1.0      1.0  \n",
       "Disease                1.0      1.0  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.describe().T"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### We have a balance dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 636,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0    0.500041\n",
      "1    0.499959\n",
      "Name: Disease, dtype: float64\n"
     ]
    }
   ],
   "source": [
    "# target variable\n",
    "print(train_df['Disease'].value_counts(normalize=True, dropna=False))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 247,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Find all numeric columns \n",
    "numeric_cols = train_df.select_dtypes([np.number]).columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 155,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x1440 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# create histograms\n",
    "train_df[numeric_cols].hist(figsize=(15,20));"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 1.1.2 Distribution of Categorical Variables\n",
    "- Our 3 categorical variables - Gender, Cholesterol, and Glucose don't have 2, 3, and 3 unique values respectively.\n",
    "- Given the small number of unqie values, we can later apply one-hot encoding to transform them from categorical variables to numeric variables. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 248,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>count</th>\n",
       "      <th>unique</th>\n",
       "      <th>top</th>\n",
       "      <th>freq</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Gender</th>\n",
       "      <td>49000</td>\n",
       "      <td>2</td>\n",
       "      <td>female</td>\n",
       "      <td>31863</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Cholesterol</th>\n",
       "      <td>49000</td>\n",
       "      <td>3</td>\n",
       "      <td>normal</td>\n",
       "      <td>36676</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Glucose</th>\n",
       "      <td>49000</td>\n",
       "      <td>3</td>\n",
       "      <td>normal</td>\n",
       "      <td>41652</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             count unique     top   freq\n",
       "Gender       49000      2  female  31863\n",
       "Cholesterol  49000      3  normal  36676\n",
       "Glucose      49000      3  normal  41652"
      ]
     },
     "execution_count": 248,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "col_obj=[x for x in train_df.columns.to_list() if str(train_df[x].dtype) == 'object']\n",
    "train_df[col_obj].describe().T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 633,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "female    0.650265\n",
      "male      0.349735\n",
      "Name: Gender, dtype: float64\n"
     ]
    }
   ],
   "source": [
    "print(train_df['Gender'].value_counts(normalize=True, dropna=False))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 634,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "normal      0.748490\n",
      "high        0.136837\n",
      "too high    0.114673\n",
      "Name: Cholesterol, dtype: float64\n"
     ]
    }
   ],
   "source": [
    "print(train_df['Cholesterol'].value_counts(normalize=True, dropna=False))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 635,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "normal      0.850041\n",
      "too high    0.075939\n",
      "high        0.074020\n",
      "Name: Glucose, dtype: float64\n"
     ]
    }
   ],
   "source": [
    "print(train_df['Glucose'].value_counts(normalize=True, dropna=False))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1.2 Potential Data Issues "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 1.2.1 Column names\n",
    "- According to python's naming covention, column names with spaces is recommended be changed. \n",
    "- Replace the spaces in \"High Blood Pressure\" and \"Low Blood Pressure\" with \"_\"."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 616,
   "metadata": {},
   "outputs": [],
   "source": [
    "# rename the high and low blood pressure column \n",
    "train_df.rename(columns={'High Blood Pressure':'High_Blood_Pressure',\n",
    "                         'Low Blood Pressure':'Low_Blood_Pressure'},inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Age</th>\n",
       "      <th>Gender</th>\n",
       "      <th>Height</th>\n",
       "      <th>Weight</th>\n",
       "      <th>High_Blood_Pressure</th>\n",
       "      <th>Low_Blood_Pressure</th>\n",
       "      <th>Cholesterol</th>\n",
       "      <th>Glucose</th>\n",
       "      <th>Smoke</th>\n",
       "      <th>Alcohol</th>\n",
       "      <th>Exercise</th>\n",
       "      <th>Disease</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>59</td>\n",
       "      <td>female</td>\n",
       "      <td>167</td>\n",
       "      <td>88.0</td>\n",
       "      <td>130</td>\n",
       "      <td>68</td>\n",
       "      <td>normal</td>\n",
       "      <td>normal</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>64</td>\n",
       "      <td>female</td>\n",
       "      <td>150</td>\n",
       "      <td>71.0</td>\n",
       "      <td>140</td>\n",
       "      <td>100</td>\n",
       "      <td>normal</td>\n",
       "      <td>normal</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>41</td>\n",
       "      <td>female</td>\n",
       "      <td>166</td>\n",
       "      <td>83.0</td>\n",
       "      <td>100</td>\n",
       "      <td>70</td>\n",
       "      <td>normal</td>\n",
       "      <td>normal</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Age  Gender  Height  Weight  High_Blood_Pressure  Low_Blood_Pressure  \\\n",
       "0   59  female     167    88.0                  130                  68   \n",
       "1   64  female     150    71.0                  140                 100   \n",
       "2   41  female     166    83.0                  100                  70   \n",
       "\n",
       "  Cholesterol Glucose  Smoke  Alcohol  Exercise  Disease  \n",
       "0      normal  normal      0        0         1        0  \n",
       "1      normal  normal      0        0         0        1  \n",
       "2      normal  normal      0        1         1        0  "
      ]
     },
     "execution_count": 160,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 1.2.2 Missing Values\n",
    "- No missing values are found."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 161,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Age                    0\n",
       "Gender                 0\n",
       "Height                 0\n",
       "Weight                 0\n",
       "High_Blood_Pressure    0\n",
       "Low_Blood_Pressure     0\n",
       "Cholesterol            0\n",
       "Glucose                0\n",
       "Smoke                  0\n",
       "Alcohol                0\n",
       "Exercise               0\n",
       "Disease                0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 161,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.isnull().sum(axis=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 1.2.3 Duplicates\n",
    "- We delete duplicates becuase they don't provide us addtional information."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 617,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "There are duplicates rows in the dataset.\n"
     ]
    }
   ],
   "source": [
    "if train_df.duplicated().sum()==0:\n",
    "    print('No duplicated rows')\n",
    "else:\n",
    "    print('There are duplicates rows in the dataset.')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 618,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of rows after removing duplicates: 47248\n"
     ]
    }
   ],
   "source": [
    "train_df.drop_duplicates(inplace=True)\n",
    "print(\"Number of rows after removing duplicates:\", train_df.shape[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 619,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of duplicate rows removed: 1752\n"
     ]
    }
   ],
   "source": [
    "print(\"Number of duplicate rows removed:\", 49000-train_df.shape[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 1.2.3 Data Outliers\n",
    "- **Do we have outliers?** \n",
    "    - Columm High_Blood_Pressure and Low_Blood Pressure have siginificant number of outliers. \n",
    "- **Why do we need to take care of outliers here?** \n",
    "    - **Naive Bayes Algorthm:** Although Naive Bayes Algorthm is robust to noise and irrelevant attributes, in Gaussian Naive Bayes, outliers will affect the shape of the Gaussian distribution and have the usual effects on the mean etc. \n",
    "    - **KNN algorithm:** Classification accuracy of the KNN algorithm is found to be adversely affected by the presence of outliers in the experimental datasets. An outlier score based on rank difference can be assigned to the points in these datasets by taking into consideration the distance and density of their local neighborhood points.\n",
    "    - **Support Vector Machine:** Generally, SVM is not very robust to outliers. Presence of a few outliers can lead to very bad global misclassification.\n",
    "    - **Random Forest**: Outliers might affect the performance of random forest in two ways. 1. Whenever a decision tree is constructed, all of the points must be classified. This means that even outliers will get classified, and hence will affect the decision trees where they were selected during boosting. 2. Bootstrapping is a part of how a RandomForest does sub-sampling. Bootstrapping is susceptible to outliers.\n",
    "    - ** Gradient Boosting**: Outliers can be bad for boosting because boosting builds each tree on previous trees' residuals/errors. Outliers will have much larger residuals than non-outliers, so gradient boosting will focus a disproportionate amount of its attention on those points.\n",
    "- **To sum up, it makes sense to remove the outliers in this dataset.**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 251,
   "metadata": {},
   "outputs": [],
   "source": [
    "numeric_cols = train_df.select_dtypes([np.number]).columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 166,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x1440 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "train_df[numeric_cols].boxplot(figsize=(15,20));"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 620,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/jieerchen/opt/anaconda3/lib/python3.7/site-packages/pandas/core/indexing.py:671: SettingWithCopyWarning:\n",
      "\n",
      "\n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "\n"
     ]
    }
   ],
   "source": [
    "outliers_cols = ['Age','Height','Weight','High_Blood_Pressure','Low_Blood_Pressure']\n",
    "for target_col in outliers_cols:\n",
    "    '''\n",
    "    For each numeric column, we use IQR method to identify outliers.\n",
    "    Column values higher than the upper bound are set to upper bound values.\n",
    "    Column values lower thatn the lower bound value are set to lower bound values.\n",
    "    \n",
    "    '''\n",
    "    lower_quartile = train_df[target_col].quantile(0.25)\n",
    "    upper_quartile = train_df[target_col].quantile(0.75)\n",
    "    IQR = upper_quartile - lower_quartile\n",
    "    outlier_thresh = 1.5 * IQR\n",
    "    # get upper/lower bound\n",
    "    upper_bound = lower_quartile + outlier_thresh\n",
    "    lower_bound = lower_quartile - outlier_thresh\n",
    "    # winsorization\n",
    "    train_df[target_col].loc[train_df[target_col] > upper_bound] = upper_bound\n",
    "    train_df[target_col].loc[train_df[target_col] < lower_bound] = lower_bound"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Data Distribution After Altering Outliers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 253,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x1440 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "train_df[numeric_cols].boxplot(figsize=(15,20));"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 1.2.4 Error Data: negative values in High Blood Pressure Column \n",
    "- Based on medical knowledge, blood pressure shouldn't have negative values.\n",
    "- Previously we have 7 rows of negative blood pressure. \n",
    "- After taking care of the outliers, we now have no negative blood pressure values."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of negatives values in High Blood Pressure Column: 0\n"
     ]
    }
   ],
   "source": [
    "# number of rows with negative values \n",
    "negative_values = sum( n <0 for n in train_df['High_Blood_Pressure'].values.flatten())\n",
    "print(\"Number of negatives values in High Blood Pressure Column:\", negative_values)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1.4 Data pre-processing and selection"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 1.4.1 Feture Encoding "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Categorical Variable Gender**\n",
    "    - Male:35%, Female: 65%\n",
    "    - Male --> 0, Female --> 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 621,
   "metadata": {
    "scrolled": false
   },
   "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>Gender</th>\n",
       "      <th>Height</th>\n",
       "      <th>Weight</th>\n",
       "      <th>High_Blood_Pressure</th>\n",
       "      <th>Low_Blood_Pressure</th>\n",
       "      <th>Cholesterol</th>\n",
       "      <th>Glucose</th>\n",
       "      <th>Smoke</th>\n",
       "      <th>Alcohol</th>\n",
       "      <th>Exercise</th>\n",
       "      <th>Disease</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>59.0</td>\n",
       "      <td>1</td>\n",
       "      <td>167.0</td>\n",
       "      <td>88.0</td>\n",
       "      <td>130.0</td>\n",
       "      <td>68.0</td>\n",
       "      <td>normal</td>\n",
       "      <td>normal</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>63.0</td>\n",
       "      <td>1</td>\n",
       "      <td>150.0</td>\n",
       "      <td>71.0</td>\n",
       "      <td>140.0</td>\n",
       "      <td>95.0</td>\n",
       "      <td>normal</td>\n",
       "      <td>normal</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>41.0</td>\n",
       "      <td>1</td>\n",
       "      <td>166.0</td>\n",
       "      <td>83.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>70.0</td>\n",
       "      <td>normal</td>\n",
       "      <td>normal</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    Age  Gender  Height  Weight  High_Blood_Pressure  Low_Blood_Pressure  \\\n",
       "0  59.0       1   167.0    88.0                130.0                68.0   \n",
       "1  63.0       1   150.0    71.0                140.0                95.0   \n",
       "2  41.0       1   166.0    83.0                100.0                70.0   \n",
       "\n",
       "  Cholesterol Glucose  Smoke  Alcohol  Exercise  Disease  \n",
       "0      normal  normal      0        0         1        0  \n",
       "1      normal  normal      0        0         0        1  \n",
       "2      normal  normal      0        1         1        0  "
      ]
     },
     "execution_count": 621,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Gender: male & female \n",
    "train_df['Gender'] = train_df['Gender'].map({'male': 0, 'female': 1})\n",
    "train_df.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Categorical Variable Cholesterol**\n",
    "- normal: 74.8%, high:13.7%, too high: 11.5%"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 622,
   "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>Gender</th>\n",
       "      <th>Height</th>\n",
       "      <th>Weight</th>\n",
       "      <th>High_Blood_Pressure</th>\n",
       "      <th>Low_Blood_Pressure</th>\n",
       "      <th>Glucose</th>\n",
       "      <th>Smoke</th>\n",
       "      <th>Alcohol</th>\n",
       "      <th>Exercise</th>\n",
       "      <th>Disease</th>\n",
       "      <th>Cholesterol_high</th>\n",
       "      <th>Cholesterol_normal</th>\n",
       "      <th>Cholesterol_too high</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>59.0</td>\n",
       "      <td>1</td>\n",
       "      <td>167.0</td>\n",
       "      <td>88.0</td>\n",
       "      <td>130.0</td>\n",
       "      <td>68.0</td>\n",
       "      <td>normal</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>63.0</td>\n",
       "      <td>1</td>\n",
       "      <td>150.0</td>\n",
       "      <td>71.0</td>\n",
       "      <td>140.0</td>\n",
       "      <td>95.0</td>\n",
       "      <td>normal</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>41.0</td>\n",
       "      <td>1</td>\n",
       "      <td>166.0</td>\n",
       "      <td>83.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>70.0</td>\n",
       "      <td>normal</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    Age  Gender  Height  Weight  High_Blood_Pressure  Low_Blood_Pressure  \\\n",
       "0  59.0       1   167.0    88.0                130.0                68.0   \n",
       "1  63.0       1   150.0    71.0                140.0                95.0   \n",
       "2  41.0       1   166.0    83.0                100.0                70.0   \n",
       "\n",
       "  Glucose  Smoke  Alcohol  Exercise  Disease  Cholesterol_high  \\\n",
       "0  normal      0        0         1        0                 0   \n",
       "1  normal      0        0         0        1                 0   \n",
       "2  normal      0        1         1        0                 0   \n",
       "\n",
       "   Cholesterol_normal  Cholesterol_too high  \n",
       "0                   1                     0  \n",
       "1                   1                     0  \n",
       "2                   1                     0  "
      ]
     },
     "execution_count": 622,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# one hot encoding \n",
    "train_df = pd.get_dummies(train_df, columns=['Cholesterol'])\n",
    "train_df.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Categorical Variable Glucose**\n",
    "- normal: 85%, high: 7,4%, too high:7.59%"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 623,
   "metadata": {
    "scrolled": false
   },
   "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>Gender</th>\n",
       "      <th>Height</th>\n",
       "      <th>Weight</th>\n",
       "      <th>High_Blood_Pressure</th>\n",
       "      <th>Low_Blood_Pressure</th>\n",
       "      <th>Smoke</th>\n",
       "      <th>Alcohol</th>\n",
       "      <th>Exercise</th>\n",
       "      <th>Disease</th>\n",
       "      <th>Cholesterol_high</th>\n",
       "      <th>Cholesterol_normal</th>\n",
       "      <th>Cholesterol_too high</th>\n",
       "      <th>Glucose_high</th>\n",
       "      <th>Glucose_normal</th>\n",
       "      <th>Glucose_too high</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>59.0</td>\n",
       "      <td>1</td>\n",
       "      <td>167.0</td>\n",
       "      <td>88.0</td>\n",
       "      <td>130.0</td>\n",
       "      <td>68.0</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>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>63.0</td>\n",
       "      <td>1</td>\n",
       "      <td>150.0</td>\n",
       "      <td>71.0</td>\n",
       "      <td>140.0</td>\n",
       "      <td>95.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>41.0</td>\n",
       "      <td>1</td>\n",
       "      <td>166.0</td>\n",
       "      <td>83.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>70.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</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>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    Age  Gender  Height  Weight  High_Blood_Pressure  Low_Blood_Pressure  \\\n",
       "0  59.0       1   167.0    88.0                130.0                68.0   \n",
       "1  63.0       1   150.0    71.0                140.0                95.0   \n",
       "2  41.0       1   166.0    83.0                100.0                70.0   \n",
       "\n",
       "   Smoke  Alcohol  Exercise  Disease  Cholesterol_high  Cholesterol_normal  \\\n",
       "0      0        0         1        0                 0                   1   \n",
       "1      0        0         0        1                 0                   1   \n",
       "2      0        1         1        0                 0                   1   \n",
       "\n",
       "   Cholesterol_too high  Glucose_high  Glucose_normal  Glucose_too high  \n",
       "0                     0             0               1                 0  \n",
       "1                     0             0               1                 0  \n",
       "2                     0             0               1                 0  "
      ]
     },
     "execution_count": 623,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# one hot encoding \n",
    "train_df = pd.get_dummies(train_df, columns=['Glucose'])\n",
    "train_df.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 1.4.2 Select features for the modeling. \n",
    "- Also we change the target data type to be integer, as it is a requirement by the skitlearn algorithm."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 624,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    0\n",
       "1    1\n",
       "2    0\n",
       "Name: Disease, dtype: int64"
      ]
     },
     "execution_count": 624,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Get target variable\n",
    "y = train_df['Disease']\n",
    "y[0:3]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 625,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "50.956654249915346\n"
     ]
    }
   ],
   "source": [
    "# Target variable is balanced\n",
    "print(y.sum() / y.shape[0] * 100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 626,
   "metadata": {
    "scrolled": false
   },
   "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>Gender</th>\n",
       "      <th>Height</th>\n",
       "      <th>Weight</th>\n",
       "      <th>High_Blood_Pressure</th>\n",
       "      <th>Low_Blood_Pressure</th>\n",
       "      <th>Smoke</th>\n",
       "      <th>Alcohol</th>\n",
       "      <th>Exercise</th>\n",
       "      <th>Cholesterol_high</th>\n",
       "      <th>Cholesterol_normal</th>\n",
       "      <th>Cholesterol_too high</th>\n",
       "      <th>Glucose_high</th>\n",
       "      <th>Glucose_normal</th>\n",
       "      <th>Glucose_too high</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>59.0</td>\n",
       "      <td>1</td>\n",
       "      <td>167.0</td>\n",
       "      <td>88.0</td>\n",
       "      <td>130.0</td>\n",
       "      <td>68.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>63.0</td>\n",
       "      <td>1</td>\n",
       "      <td>150.0</td>\n",
       "      <td>71.0</td>\n",
       "      <td>140.0</td>\n",
       "      <td>95.0</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>41.0</td>\n",
       "      <td>1</td>\n",
       "      <td>166.0</td>\n",
       "      <td>83.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>70.0</td>\n",
       "      <td>0</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>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    Age  Gender  Height  Weight  High_Blood_Pressure  Low_Blood_Pressure  \\\n",
       "0  59.0       1   167.0    88.0                130.0                68.0   \n",
       "1  63.0       1   150.0    71.0                140.0                95.0   \n",
       "2  41.0       1   166.0    83.0                100.0                70.0   \n",
       "\n",
       "   Smoke  Alcohol  Exercise  Cholesterol_high  Cholesterol_normal  \\\n",
       "0      0        0         1                 0                   1   \n",
       "1      0        0         0                 0                   1   \n",
       "2      0        1         1                 0                   1   \n",
       "\n",
       "   Cholesterol_too high  Glucose_high  Glucose_normal  Glucose_too high  \n",
       "0                     0             0               1                 0  \n",
       "1                     0             0               1                 0  \n",
       "2                     0             0               1                 0  "
      ]
     },
     "execution_count": 626,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Get x variables\n",
    "X = train_df.drop('Disease',axis=1)\n",
    "X.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 1.4.3 Normalize the dataset\n",
    "**Choose MinMaxScaler over StandardScaler**: \n",
    "   - MultinomialNB assumes that features have multinomial distribution which is a generalization of the binomial distribution. Neither binomial nor multinomial distributions can contain negative values. If we use StandardScaler to process the X varaibles, the data will follow Gaussian with zero mean and unit variance, leading to the presence of negative values.\n",
    "   - Aftering applying MinMaxScaler, our data are range from 0 to 1."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 260,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.87, 1.  , 0.74, 0.95, 0.67, 0.1 , 0.  , 0.  , 1.  , 0.  , 1.  ,\n",
       "        0.  , 0.  , 1.  , 0.  ],\n",
       "       [1.  , 1.  , 0.23, 0.62, 0.83, 1.  , 0.  , 0.  , 0.  , 0.  , 1.  ,\n",
       "        0.  , 0.  , 1.  , 0.  ]])"
      ]
     },
     "execution_count": 260,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.preprocessing import MinMaxScaler\n",
    "X = preprocessing.MinMaxScaler().fit(X).transform(X)\n",
    "X[0:2]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 1.4.4 Create training and validation dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 261,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train set: (33073, 15) (33073,)\n",
      "Test set: (14175, 15) (14175,)\n"
     ]
    }
   ],
   "source": [
    "X_train, X_test, y_train, y_test =  train_test_split(X, y, test_size=0.3, random_state=16)\n",
    "print(\"Train set:\",X_train.shape, y_train.shape)\n",
    "print(\"Test set:\", X_test.shape, y_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1.5 Data Visualization"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 1.5.1 Correlation Matrix\n",
    "**Observation:** age(0.24), high blood pressure(0.45) and low blood pressure(0.36) have the highest correlation with the target variable Disease."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 262,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1080 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize = (20,15))\n",
    "Matrix = np.triu(train_df.corr())\n",
    "ax = sns.heatmap(train_df.corr(),  annot = True, fmt='.2g', mask=Matrix, cmap= 'coolwarm', linewidths=.2)\n",
    "bottom, top = ax.get_ylim()\n",
    "ax.set_ylim(bottom + 0.5, top - 0.5)\n",
    "plt.title('Correlation Matrix',y=-0.1, fontsize=20);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 1.5.2 Plot key columns to understand the data better "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Observatio:** It seems like people aged over 55 are more likely to be diagnosed with disease."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 263,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x216 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "bins = np.linspace(train_df.Age.min(), train_df.Age.max(), 10)\n",
    "g = sns.FacetGrid(train_df, col=\"Disease\", hue=\"Disease\", palette=\"Set1\", col_wrap=2)\n",
    "g.map(plt.hist, 'Age', bins=bins, ec=\"k\")\n",
    "\n",
    "g.axes[-1].legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Observation:** When looking at the effect of age and weight together, it looks like that people weight over 75 and age over 55 have a higher risks of being diagnosed with disease."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 264,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = train_df[train_df['Disease'] == 1][0:50].plot(kind='scatter', x='Weight', y='Age', color='DarkBlue', label='Disease=1');\n",
    "train_df[train_df['Disease'] == 0][0:50].plot(kind='scatter', x='Weight', y='Age', color='Yellow', label='Disease=0', ax=ax);\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Section 2: Build, tune and evaluate cluster analysis and decision tree models"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.1 Navie Bayes Algorithm \n",
    "- Scikit-learn implements three naive Bayes variants based on the same number of different probabilistic distributions: Gaussian, multinomial, and Bernoulli."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.1.1 Gaussian NB Model\n",
    "- Gaussian distribution is a continuous distribution characterized by its mean and variance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 540,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.6874\n",
      "Precision: 0.7367\n",
      "Recall: 0.6034\n",
      "F1-Score: 0.6634\n",
      "ROC AUC Score: 0.6892\n"
     ]
    }
   ],
   "source": [
    "gnb = GaussianNB()\n",
    "gnb.fit(X_train, y_train)\n",
    "gnb_pred = gnb.predict(X_test)\n",
    "print(f\"Accuracy: {round(metrics.accuracy_score(y_test, gnb_pred), 4)}\")\n",
    "print(f\"Precision: {round(metrics.precision_score(y_test, gnb_pred), 4)}\")\n",
    "print(f\"Recall: {round(metrics.recall_score(y_test, gnb_pred), 4)}\")\n",
    "print(f\"F1-Score: {round(metrics.f1_score(y_test, gnb_pred), 4)}\")\n",
    "print(f\"ROC AUC Score: {round(metrics.roc_auc_score(y_test, gnb_pred), 4)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 266,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.65      0.78      0.71      6938\n",
      "           1       0.74      0.60      0.66      7237\n",
      "\n",
      "    accuracy                           0.69     14175\n",
      "   macro avg       0.69      0.69      0.69     14175\n",
      "weighted avg       0.70      0.69      0.69     14175\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print (classification_report(y_test, gnb_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 267,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Use predict_proba to get the probability results of Gaussain NB Model\n",
    "y_pred_gnb = gnb.predict_proba(X_test)[:, 1]\n",
    "fpr_gnb, tpr_gnb, thresh = roc_curve(y_test, y_pred_gnb)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 534,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# ROC Curve\n",
    "plt.figure(1)\n",
    "plt.plot([0, 1], [0, 1], 'k--')\n",
    "plt.plot(fpr_gnb, tpr_gnb, label='Gaussain NB Model')\n",
    "plt.xlabel('False positive rate')\n",
    "plt.ylabel('True positive rate')\n",
    "plt.title('ROC curve - Gaussain NB Model Model')\n",
    "plt.legend(loc='best')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.1.2 Multinomial NB Model\n",
    "- Multinomial distribution is a discrete distribution used whenever a feature must be represented by a whole number."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 542,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.614\n",
      "Precision: 0.6851\n",
      "Recall: 0.4513\n",
      "F1-score: 0.5442\n",
      "ROC AUC Score: 0.6175\n"
     ]
    }
   ],
   "source": [
    "mnb = MultinomialNB(alpha=0.5)\n",
    "mnb.fit(X_train, y_train)\n",
    "mnb_pred = mnb.predict(X_test)\n",
    "print(f\"Accuracy: {round(metrics.accuracy_score(y_test, mnb_pred), 4)}\")\n",
    "print(f\"Precision: {round(metrics.precision_score(y_test, mnb_pred), 4)}\")\n",
    "print(f\"Recall: {round(metrics.recall_score(y_test, mnb_pred), 4)}\")\n",
    "print(f\"F1-score: {round(metrics.f1_score(y_test, mnb_pred), 4)}\")\n",
    "print(f\"ROC AUC Score: {round(metrics.roc_auc_score(y_test, mnb_pred), 4)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 271,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.58      0.78      0.67      6938\n",
      "           1       0.69      0.45      0.54      7237\n",
      "\n",
      "    accuracy                           0.61     14175\n",
      "   macro avg       0.63      0.62      0.60     14175\n",
      "weighted avg       0.63      0.61      0.60     14175\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print (classification_report(y_test, mnb_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 272,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Use predict_proba to get the probability results of Multinomial NB Model\n",
    "y_pred_mnb = mnb.predict_proba(X_test)[:, 1]\n",
    "fpr_mnb, tpr_mnb, thresh = roc_curve(y_test, y_pred_mnb)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 535,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# ROC Curve\n",
    "plt.figure(1)\n",
    "plt.plot([0, 1], [0, 1], 'k--')\n",
    "plt.plot(fpr_mnb, tpr_mnb, label='Multinomial NB Model')\n",
    "plt.xlabel('False positive rate')\n",
    "plt.ylabel('True positive rate')\n",
    "plt.title('ROC curve - Gaussain NB Model Model')\n",
    "plt.legend(loc='best')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.1.3 Bernoulli NB Model\n",
    "- Bernoulli distribution is a binary distribution useful when a feature can be present or absent. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 541,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.5898\n",
      "Precision: 0.6651\n",
      "Recall: 0.396\n",
      "F1-Score: 0.4964\n",
      "ROC AUC Score: 0.594\n"
     ]
    }
   ],
   "source": [
    "bnb = BernoulliNB()\n",
    "bnb.fit(X_train, y_train)\n",
    "bnb_pred = bnb.predict(X_test)\n",
    "print(f\"Accuracy: {round(metrics.accuracy_score(y_test, bnb_pred), 4)}\")\n",
    "print(f\"Precision: {round(metrics.precision_score(y_test, bnb_pred), 4)}\")\n",
    "print(f\"Recall: {round(metrics.recall_score(y_test, bnb_pred), 4)}\")\n",
    "print(f\"F1-Score: {round(metrics.f1_score(y_test, bnb_pred), 4)}\")\n",
    "print(f\"ROC AUC Score: {round(metrics.roc_auc_score(y_test, bnb_pred), 4)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 277,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.56      0.79      0.65      6938\n",
      "           1       0.67      0.40      0.50      7237\n",
      "\n",
      "    accuracy                           0.59     14175\n",
      "   macro avg       0.61      0.59      0.58     14175\n",
      "weighted avg       0.61      0.59      0.57     14175\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print (classification_report(y_test, bnb_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 278,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Use predict_proba to get the probability results of Bernoulli NB Model\n",
    "y_pred_bnb = bnb.predict_proba(X_test)[:, 1]\n",
    "fpr_bnb, tpr_bnb, thresh = roc_curve(y_test, y_pred_bnb)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 537,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# ROC Curve\n",
    "plt.figure(1)\n",
    "plt.plot([0, 1], [0, 1], 'k--')\n",
    "plt.plot(fpr_mnb, tpr_mnb, label='Bernoulli NB Model')\n",
    "plt.xlabel('False positive rate')\n",
    "plt.ylabel('True positive rate')\n",
    "plt.title('ROC curve - Gaussain NB Model Model')\n",
    "plt.legend(loc='best')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.1.4 Use 10-fold Cross Validation to help in identifying a case of overfitting\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 545,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model accuracy_mean of Gaussian NB Model is 0.6885\n",
      "Model accuracy_std of Gaussian NB Model is 0.0108\n",
      "Model accuracy_mean of Multinomial NB Model is 0.6109\n",
      "Model accuracy_std of Multinomial NB Model is 0.0111\n",
      "Model accuracy_mean of Bernoulli NB Model is 0.5882\n",
      "Model accuracy_std of Bernoulli NB Model is 0.01\n"
     ]
    }
   ],
   "source": [
    "model_names = ['Gaussian NB Model','Multinomial NB Model','Bernoulli NB Model']\n",
    "model_list = [gnb, mnb, bnb]\n",
    "count = 0\n",
    "\n",
    "for classifier in model_list:\n",
    "    cv_score = model_selection.cross_val_score(classifier, X_train, y_train, cv=10)\n",
    "    print('Model accuracy_mean of ' + model_names[count] + ' is ' + str(round(cv_score.mean(),4)))\n",
    "    print('Model accuracy_std of ' + model_names[count] + ' is ' + str(round(cv_score.std(),4)))\n",
    "    count += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.1.6 Summary of NB model performance: \n",
    "- Gussain NB Model has the best performance among the three models."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 543,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Metrics</th>\n",
       "      <th>Gussain</th>\n",
       "      <th>Multilnomial</th>\n",
       "      <th>Bernoulli</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Accuracy</td>\n",
       "      <td>0.6874</td>\n",
       "      <td>0.6140</td>\n",
       "      <td>0.5898</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Precision</td>\n",
       "      <td>0.7367</td>\n",
       "      <td>0.6851</td>\n",
       "      <td>0.6651</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Recall</td>\n",
       "      <td>0.6034</td>\n",
       "      <td>0.4513</td>\n",
       "      <td>0.3960</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>F1-Score</td>\n",
       "      <td>0.6634</td>\n",
       "      <td>0.5442</td>\n",
       "      <td>0.4964</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>CV_mean</td>\n",
       "      <td>0.6885</td>\n",
       "      <td>0.6109</td>\n",
       "      <td>0.5882</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>CV_std</td>\n",
       "      <td>0.0108</td>\n",
       "      <td>0.0111</td>\n",
       "      <td>0.0100</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>AUC</td>\n",
       "      <td>0.6892</td>\n",
       "      <td>0.6175</td>\n",
       "      <td>0.5940</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     Metrics  Gussain  Multilnomial  Bernoulli\n",
       "0   Accuracy   0.6874        0.6140     0.5898\n",
       "1  Precision   0.7367        0.6851     0.6651\n",
       "2     Recall   0.6034        0.4513     0.3960\n",
       "3   F1-Score   0.6634        0.5442     0.4964\n",
       "4    CV_mean   0.6885        0.6109     0.5882\n",
       "5     CV_std   0.0108        0.0111     0.0100\n",
       "6        AUC   0.6892        0.6175     0.5940"
      ]
     },
     "execution_count": 543,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nym_performance = {'Metrics': ['Accuracy','Precision','Recall','F1-Score','CV_mean','CV_std','AUC'], \n",
    "           'Gussain': [0.6874, 0.7367,0.6034,0.6634,0.6885,0.0108,0.6892],\n",
    "           'Multilnomial':[0.614,0.6851,0.4513,0.5442,0.6109,0.0111,0.6175],\n",
    "          'Bernoulli':[0.5898,0.6651,0.396,0.4964,0.5882,0.01, 0.594]}\n",
    "nym_performance = pd.DataFrame(data= nym_performance)\n",
    "nym_performance "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.2 K Nearest Neighbor (KNN)\n",
    "- The k-nearest neighbors (KNN) algorithm is a simple, easy-to-implement supervised machine learning algorithm that can be used to solve both classification and regression problems. It assumes that similar things exist in close proximity. In other words, similar things are near to each other."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.2.1 Base KNN Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 560,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy of base KNN: 69.22%\n"
     ]
    }
   ],
   "source": [
    "classifier = KNeighborsClassifier()\n",
    "classifier.fit(X_train, y_train)\n",
    "y_pred = classifier.predict(X_test)\n",
    "print(f\"Accuracy of base KNN: {round(metrics.accuracy_score(y_test, y_pred)*100, 2)}%\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.2.2 Sensitivity Analysis \n",
    "- Number of neighbors (K) is a hyper parameter of the KNN model. As we decrease the value of K to 1, our predictions become less stable. Inversely, as we increase the value of K, our predictions become more stable due to majority voting / averaging, and thus, more likely to make more accurate predictions (up to a certain point). Eventually, we begin to witness an increasing number of errors. It is at this point we know we have pushed the value of K too far."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Building the KNN models by assigning the value of K from 1 to 30"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 300,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.63, 0.63, 0.67, 0.68, 0.69, 0.7 , 0.7 , 0.71, 0.71, 0.71, 0.71,\n",
       "       0.71, 0.71, 0.72, 0.72, 0.72, 0.72, 0.72, 0.72, 0.72, 0.72, 0.72,\n",
       "       0.72, 0.72, 0.72, 0.72, 0.72, 0.72, 0.72, 0.72, 0.72, 0.72, 0.72,\n",
       "       0.72, 0.72, 0.72, 0.72, 0.72, 0.72, 0.72, 0.72, 0.72, 0.72, 0.72,\n",
       "       0.72, 0.72, 0.72, 0.72, 0.72])"
      ]
     },
     "execution_count": 300,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Importing the KNN model\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "\n",
    "Ks = 50\n",
    "mean_acc = np.zeros((Ks-1))\n",
    "std_acc = np.zeros((Ks-1))\n",
    "ConfustionMx = [];\n",
    "for n in range(1,Ks):\n",
    "    \n",
    "    #Train Model and Predict  \n",
    "    neigh = KNeighborsClassifier(n_neighbors = n).fit(X_train,y_train)\n",
    "    yhat=neigh.predict(X_test)\n",
    "    mean_acc[n-1] = metrics.accuracy_score(y_test, yhat)\n",
    "\n",
    "    \n",
    "    std_acc[n-1]=np.std(yhat==y_test)/np.sqrt(yhat.shape[0])\n",
    "\n",
    "mean_acc"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Plot model accuracy for different number of Neighbors¶\n",
    "- The accury of the model increase sharping as K increase from 0 to around 20\n",
    "- The variation of the accuracy is small after K reach around 20."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 301,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(range(1,Ks),mean_acc,'g')\n",
    "plt.fill_between(range(1,Ks),mean_acc - 1 * std_acc,mean_acc + 1 * std_acc, alpha=0.10)\n",
    "plt.legend(('Accuracy ', '+/- 3xstd'))\n",
    "plt.ylabel('Accuracy ')\n",
    "plt.xlabel('Number of Neighbors (K)')\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.2.3 More efficient parameter tuning using GridSearchCV: KNN\n",
    "- **Number of neighbors (K)** \n",
    "    - To select the K that’s right for our data, we run the KNN algorithm several times with different values of K and choose the K that reduces the number of errors we encounter while maintaining the algorithm’s ability to accurately make predictions when it’s given data it hasn’t seen before.\n",
    "    - If the number of neighbors is too large, the model will have high bias. If the number of neighors is too small, the model will have high variance (overfitting). We usually make K an odd number to have a tiebreaker.\n",
    "- **Distance functions**\n",
    "    - Euclidean: sqrt(sum((x - y)^2))\n",
    "    - Manhattan: sum(|x - y|)\n",
    "    - Minkowski: sum(|x - y|^p)^(1/p)\n",
    " \n",
    "- **Weight Options**\n",
    "    - uniform: all points in the neighborhood are weighted equally\n",
    "    - distance: weights closer neighbors more heavily than further neighbors\n",
    "    \n",
    "- **CV = 10 & scoring = 'accuracy'**\n",
    "    - Grid object is ready to do 10-fold cross validation on a KNN model using classification accuracy as the evaluation metric\n",
    "    - KNN model is being fit and predictions are being made 8 x 10 = 80 times"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 308,
   "metadata": {},
   "outputs": [],
   "source": [
    "# helper function for printing out grid search results \n",
    "def print_grid_search_metrics(gs):\n",
    "    print (\"Best score: \" + str(gs.best_score_))\n",
    "    print (\"Best parameters set\")\n",
    "    best_parameters = gs.best_params_\n",
    "    for param_name in sorted(best_parameters.keys()):\n",
    "        print(param_name,':',str(best_parameters[param_name]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 321,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=10, error_score=nan,\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='deprecated', n_jobs=None,\n",
       "             param_grid={'metric': ['minkowski', 'euclidean', 'manhattan'],\n",
       "                         'n_neighbors': [17, 19, 21, 23, 25, 27, 29, 31],\n",
       "                         'weights': ['uniform', 'distance']},\n",
       "             pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n",
       "             scoring='accuracy', verbose=0)"
      ]
     },
     "execution_count": 321,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Possible hyperparamter options for KNN\n",
    "# Choose k\n",
    "parameters = {\n",
    "    'n_neighbors':[17,19,21,23,25,27,29,31],\n",
    "    'metric':['minkowski','euclidean','manhattan'],\n",
    "    'weights':['uniform','distance']\n",
    "}\n",
    "Grid_KNN = GridSearchCV(KNeighborsClassifier(),parameters, cv=10,scoring='accuracy')\n",
    "Grid_KNN.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 322,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best score: 0.7245186367709987\n",
      "Best parameters set\n",
      "metric : manhattan\n",
      "n_neighbors : 29\n",
      "weights : uniform\n"
     ]
    }
   ],
   "source": [
    "# best k\n",
    "best_KNN_model = Grid_KNN.best_estimator_\n",
    "print_grid_search_metrics(Grid_KNN)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.2.4 Model Performance Evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 432,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred_KNN = best_KNN_model.predict(X_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Get Repeated Hold Out Accurary of optimal KNN Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- **Cross validation:** Use 10-fold Cross Validation to give a more accurate indication of how well the model generalizes to unseen data.\n",
    "- **Model Performance Evaluation:** The model accuracy of the KNN model is 72.43% "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 547,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cv_acc_mean 0.7243\n",
      "cv_acc_std 0.0064\n"
     ]
    }
   ],
   "source": [
    "knn_pipe = make_pipeline(best_KNN_model)\n",
    "knn_pipe.fit(X_train, y_train)\n",
    "pipe_pred = knn_pipe.predict(X_test)\n",
    "from sklearn.model_selection import KFold\n",
    "cv = KFold(n_splits=10, shuffle=True, random_state=16)\n",
    "print('cv_acc_mean',round(cross_val_score(knn_pipe, X_train, y_train, cv=cv).mean(),4))\n",
    "print('cv_acc_std',round(cross_val_score(knn_pipe, X_train, y_train, cv=cv).std(),4))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Confusion Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 357,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>col_0</th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Disease</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5202</td>\n",
       "      <td>1736</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2173</td>\n",
       "      <td>5064</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "col_0       0     1\n",
       "Disease            \n",
       "0        5202  1736\n",
       "1        2173  5064"
      ]
     },
     "execution_count": 357,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "knn_confusion = pd.crosstab(y_test, y_pred_KNN)\n",
    "knn_confusion"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 549,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.7242\n",
      "Precision: 0.7447\n",
      "Recall: 0.6997\n",
      "F1-Score: 0.7215\n",
      "ROC AUC Score: 0.7248\n"
     ]
    }
   ],
   "source": [
    "print(f\"Accuracy: {round(metrics.accuracy_score(y_test, y_pred_KNN), 4)}\")\n",
    "print(f\"Precision: {round(metrics.precision_score(y_test, y_pred_KNN), 4)}\")\n",
    "print(f\"Recall: {round(metrics.recall_score(y_test, y_pred_KNN), 4)}\")\n",
    "print(f\"F1-Score: {round(metrics.f1_score(y_test, y_pred_KNN), 4)}\")\n",
    "print(f\"ROC AUC Score: {round(metrics.roc_auc_score(y_test, y_pred_KNN), 4)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 358,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.71      0.75      0.73      6938\n",
      "           1       0.74      0.70      0.72      7237\n",
      "\n",
      "    accuracy                           0.72     14175\n",
      "   macro avg       0.73      0.72      0.72     14175\n",
      "weighted avg       0.73      0.72      0.72     14175\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(metrics.classification_report(y_test, y_pred_KNN))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### ROC Curve and AUC for KNN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 329,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Use predict_proba to get the probability results of KNN Model\n",
    "y_pred_knn = best_KNN_model.predict_proba(X_test)[:, 1]\n",
    "fpr_knn, tpr_knn, thresh = roc_curve(y_test, y_pred_knn)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 550,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# ROC Curve\n",
    "plt.figure(1)\n",
    "plt.plot([0, 1], [0, 1], 'k--')\n",
    "plt.plot(fpr_knn, tpr_knn, label='KNN Model')\n",
    "plt.xlabel('False positive rate')\n",
    "plt.ylabel('True positive rate')\n",
    "plt.title('ROC curve - KNN Model')\n",
    "plt.legend(loc='best')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.3 Support Vector Machine \n",
    "- **SVM works by mapping data to a high-dimensional feature space so that data points can be categorized, even when the data are not otherwise linearly separable.** A separator between the categories is found, then the data is transformed in such a way that the separator could be drawn as a hyperplane. Following this, characteristics of new data can be used to predict the group to which a new record should belong.\n",
    "- **In this section, we will use the Linear kernel and RBF kernel for the SVM classifier to identify the best kernel**: The SVM algorithm offers a choice of kernel functions for performing its processing. Basically, mapping data into a higher dimensional space is called kernelling. The mathematical function used for the transformation is known as the kernel function, and can be of different types, such as Linear, Polynomial, Radial basis function and Sigmoid."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.3.1 Linear SVM "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####  2.3.1.1 Linear SVM Base Model\n",
    "- A linear kernel can be used as normal dot product any two given observations. The product between two vectors is the sum of the multiplication of each pair of input values.\n",
    "- Formula: K(x, xi) = sum(x * xi)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 445,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import the library for SVM Classifier\n",
    "from sklearn import svm\n",
    "\n",
    "#Build a SVM Classifier with a Linear base Kernel\n",
    "svm_linear = svm.SVC(kernel='linear').fit(X_train, y_train)\n",
    "yhat_linear = svm_linear.predict(X_test)\n",
    "svm_linear = metrics.accuracy_score(y_test, yhat_linear)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 449,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy for base SVM_linear: 73.17%\n"
     ]
    }
   ],
   "source": [
    "print(f\"Accuracy for base SVM_linear: {round(svm_linear*100, 2)}%\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.3.1.2 Linear SVM Sensitivity Analysis \n",
    "- **Parameter C**: \n",
    "    - **C controls the trade off between the achieving a low training error and a low testing error that is the ability to generalize your classifier to unseen data.**\n",
    "    - In a SVM we are searching for two things - a hyperplane with the largest minimum margin and a hyperplane that correctly separates as many instances as possible. The problem is that we will not always be able to get both things. The C parameter determines how great our desire is for the latter. In other words, the C parameter tells the SVM optimization how much we want to avoid misclassifying each training example. \n",
    "- **The value of C**\n",
    "   - Large values of C result in a wider margin, at the cost of some misclassifications. The optimization will choose a smaller-margin hyperplane if that hyperplane does a better job of getting all the training points classified correctly.\n",
    "   - Conversely, small value of C will result in the Hard Margin classifier and tolerates zero constraint violation. It will cause the optimizer to look for a larger-margin separating hyperplane, even if that hyperplane misclassifies more points. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 458,
   "metadata": {},
   "outputs": [],
   "source": [
    "C = np.arange(0.05,1,0.05)\n",
    "train_results_svm =[]\n",
    "for n in C:\n",
    "    model = svm.LinearSVC(C=n)\n",
    "    model.fit(X_train, y_train)\n",
    "    acc = cross_val_score(model, X_train, y_train, cv=10).mean()*100\n",
    "    train_results_svm.append(acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 477,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(range(0,19),train_results_svm,'g')\n",
    "plt.ylabel('Accuracy ')\n",
    "plt.xlabel('Value of C')\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.3.1.3 Linear SVM Model Tuning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LinearSVC(C=1.0, class_weight=None, dual=True, fit_intercept=True,\n",
       "          intercept_scaling=1, loss='squared_hinge', max_iter=1000,\n",
       "          multi_class='ovr', penalty='l2', random_state=None, tol=0.0001,\n",
       "          verbose=0)"
      ]
     },
     "execution_count": 128,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lin_clf = svm.LinearSVC()\n",
    "lin_clf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=ShuffleSplit(n_splits=25, random_state=0, test_size=0.3, train_size=None),\n",
       "             error_score=nan,\n",
       "             estimator=LinearSVC(C=1.0, class_weight=None, dual=True,\n",
       "                                 fit_intercept=True, intercept_scaling=1,\n",
       "                                 loss='squared_hinge', max_iter=1000,\n",
       "                                 multi_class='ovr', penalty='l2',\n",
       "                                 random_state=None, tol=0.0001, verbose=0),\n",
       "             iid='deprecated', n_jobs=None,\n",
       "             param_grid={'C': [0.25, 0.5, 0.75, 1], 'penalty': ['l2']},\n",
       "             pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n",
       "             scoring=None, verbose=0)"
      ]
     },
     "execution_count": 130,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bs = model_selection.ShuffleSplit(n_splits=25,test_size=0.3,random_state=0)\n",
    "param_grid = {'C':[0.25,0.5,0.75,1],'penalty':['l2']}\n",
    "\n",
    "gridbs = GridSearchCV(lin_clf,param_grid,cv=bs)\n",
    "gridbs.fit(X_train,y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 567,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best hyperparameter is C: 0.25\n",
      "Best hyperparameter is penalty: l2\n"
     ]
    }
   ],
   "source": [
    "for key, val in gridbs.best_params_.items():\n",
    "  print(f\"Best hyperparameter is {key}: {val}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 497,
   "metadata": {},
   "outputs": [],
   "source": [
    "best_svm_linear = gridbs.best_estimator_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.3.1.4 Linear SVM Performance Evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 556,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred_svml = best_svm_linear.predict(X_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Get Repeated Hold Out Accurary of Optimal Linear SVM "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- **Cross validation:** Use 10-fold Cross Validation to give a more accurate indication of how well the model generalizes to unseen data.\n",
    "- **Model Performance Evaluation:** The model accuracy of the optimal SVM model is 72.43% "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 551,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cv_acc_mean 0.7267\n",
      "cv_acc_std 0.0058\n"
     ]
    }
   ],
   "source": [
    "svml_pipe = make_pipeline(best_svm_linear)\n",
    "svml_pipe.fit(X_train, y_train)\n",
    "pipe_pred = svml_pipe.predict(X_test)\n",
    "from sklearn.model_selection import KFold\n",
    "cv = KFold(n_splits=10, shuffle=True, random_state=16)\n",
    "print('cv_acc_mean',round(cross_val_score(svml_pipe, X_train, y_train, cv=cv).mean(),4))\n",
    "print('cv_acc_std',round(cross_val_score(svml_pipe, X_train, y_train, cv=cv).std(),4))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Confusion Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 465,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>col_0</th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Disease</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5389</td>\n",
       "      <td>1549</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2251</td>\n",
       "      <td>4986</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "col_0       0     1\n",
       "Disease            \n",
       "0        5389  1549\n",
       "1        2251  4986"
      ]
     },
     "execution_count": 465,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "svml_confusion = pd.crosstab(y_test, y_pred_svml)\n",
    "svml_confusion"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 582,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.7307\n",
      "Precision: 0.7592\n",
      "Recall: 0.6919\n",
      "F1-Score: 0.724\n",
      "ROC AUC Score: 0.7315\n"
     ]
    }
   ],
   "source": [
    "print(f\"Accuracy: {round(metrics.accuracy_score(y_test, y_pred_svml), 4)}\")\n",
    "print(f\"Precision: {round(metrics.precision_score(y_test, y_pred_svml), 4)}\")\n",
    "print(f\"Recall: {round(metrics.recall_score(y_test, y_pred_svml), 4)}\")\n",
    "print(f\"F1-Score: {round(metrics.f1_score(y_test, y_pred_svml), 4)}\")\n",
    "print(f\"ROC AUC Score: {round(metrics.roc_auc_score(y_test, y_pred_svml), 4)}\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 466,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.71      0.78      0.74      6938\n",
      "           1       0.76      0.69      0.72      7237\n",
      "\n",
      "    accuracy                           0.73     14175\n",
      "   macro avg       0.73      0.73      0.73     14175\n",
      "weighted avg       0.73      0.73      0.73     14175\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(metrics.classification_report(y_test, y_pred_svml))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### ROC Curve and AUC for Linear SVM "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 469,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Use predict_proba to get the probability results of Linear SVM Model\n",
    "y_pred_svml = best_svm_linear._predict_proba_lr(X_test)[:, 1]\n",
    "fpr_svml, tpr_svml, thresh = roc_curve(y_test, y_pred_svml)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 559,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# ROC Curve\n",
    "plt.figure(1)\n",
    "plt.plot([0, 1], [0, 1], 'k--')\n",
    "plt.plot(fpr_svml, tpr_svml, label='SVM Liner Model')\n",
    "plt.xlabel('False positive rate')\n",
    "plt.ylabel('True positive rate')\n",
    "plt.title('ROC curve - SVM Liner Model')\n",
    "plt.legend(loc='best')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.3.2 RBF SVM  \n",
    "- The Radial basis function kernel is a popular kernel function commonly used in support vector machine classification. RBF can map an input space in infinite dimensional space.\n",
    "- Formula: K(x,xi) = 1 + sum(x * xi)^d"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### RBF SVM Base Modele"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 446,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Build a SVM Classifier with a Radial base Function Kernel\n",
    "svm_rbf = svm.SVC(kernel='rbf').fit(X_train, y_train)\n",
    "yhat_rbf = svm_rbf.predict(X_test)\n",
    "svm_rbf = metrics.accuracy_score(y_test, yhat_rbf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 450,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy for base SVM_linear: 73.25%\n"
     ]
    }
   ],
   "source": [
    "print(f\"Accuracy for base SVM_linear: {round(svm_rbf*100, 2)}%\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### RBF SVM Model Tuning \n",
    "- **Parameter**: In RBF, gamma is a parameter, which ranges from 0 to 1.  The value of gamma needs to be manually specified in the learning algorithm.\n",
    "- **Overfitting:** A higher value of gamma will perfectly fit the training dataset, which causes over-fitting. Gamma=0.1 is considered to be a good default value."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.3.3 SVM with RBF Kernel Function: γ"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 494,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 120 candidates, totalling 600 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.\n",
      "[Parallel(n_jobs=1)]: Done 600 out of 600 | elapsed: 294.7min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.7312613227261807\n",
      "SVC(C=0.7, break_ties=False, cache_size=200, class_weight=None, coef0=0.0,\n",
      "    decision_function_shape='ovr', degree=3, gamma=0.5, kernel='rbf',\n",
      "    max_iter=-1, probability=False, random_state=None, shrinking=True,\n",
      "    tol=0.001, verbose=False)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "hyper = {'C':[0.05,0.1,0.2,0.3,0.25,0.4,0.5,0.6,0.7,0.8,0.9,1],\n",
    "         'gamma':[0.1,0.2,0.3,0.4,0.5,0.6,0.7,0.8,0.9,1.0],\n",
    "         'kernel':['rbf']\n",
    "        }\n",
    "\n",
    "gd=GridSearchCV(estimator=svm.SVC(),param_grid=hyper,verbose=True)\n",
    "\n",
    "gd.fit(X_train,y_train)\n",
    "print(gd.best_score_)\n",
    "print(gd.best_estimator_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 568,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best hyperparameter is C: 0.7\n",
      "Best hyperparameter is gamma: 0.5\n",
      "Best hyperparameter is kernel: rbf\n"
     ]
    }
   ],
   "source": [
    "for key, val in gd.best_params_.items():\n",
    "  print(f\"Best hyperparameter is {key}: {val}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 498,
   "metadata": {},
   "outputs": [],
   "source": [
    "best_svm_rbf = gd.best_estimator_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.3.1.4 RBF SVM Performance Evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 499,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred_svmr = best_svm_rbf.predict(X_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Get Repeated Hold Out Accurary of optimal RBF RBF Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- **Cross validation:** Use 10-fold Cross Validation to give a more accurate indication of how well the model generalizes to unseen data.\n",
    "- **Model Performance Evaluation:** The model accuracy of the KNN model is 72.43% "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Confusion Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 561,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cv_acc_mean 0.7311\n",
      "cv_acc_std 0.0057\n"
     ]
    }
   ],
   "source": [
    "svmr_pipe = make_pipeline(best_svm_rbf)\n",
    "svmr_pipe.fit(X_train, y_train)\n",
    "pipe_pred = svmr_pipe.predict(X_test)\n",
    "cv = KFold(n_splits=10, shuffle=True, random_state=16)\n",
    "cross_val_score(svmr_pipe, X_train, y_train, cv=cv).mean()\n",
    "print('cv_acc_mean',round(cross_val_score(svmr_pipe, X_train, y_train, cv=cv).mean(),4))\n",
    "print('cv_acc_std',round(cross_val_score(svmr_pipe, X_train, y_train, cv=cv).std(),4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 502,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>col_0</th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Disease</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5454</td>\n",
       "      <td>1484</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2302</td>\n",
       "      <td>4935</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "col_0       0     1\n",
       "Disease            \n",
       "0        5454  1484\n",
       "1        2302  4935"
      ]
     },
     "execution_count": 502,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "svml_confusion = pd.crosstab(y_test, y_pred_svmr)\n",
    "svml_confusion"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 569,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.7329\n",
      "Precision: 0.7688\n",
      "Recall: 0.6819\n",
      "F1-Score: 0.7228\n",
      "ROC AUC Score: 0.734\n"
     ]
    }
   ],
   "source": [
    "print(f\"Accuracy: {round(metrics.accuracy_score(y_test, y_pred_svmr), 4)}\")\n",
    "print(f\"Precision: {round(metrics.precision_score(y_test, y_pred_svmr), 4)}\")\n",
    "print(f\"Recall: {round(metrics.recall_score(y_test, y_pred_svmr), 4)}\")\n",
    "print(f\"F1-Score: {round(metrics.f1_score(y_test,y_pred_svmr), 4)}\")\n",
    "print(f\"ROC AUC Score: {round(metrics.roc_auc_score(y_test, y_pred_svmr), 4)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 503,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.70      0.79      0.74      6938\n",
      "           1       0.77      0.68      0.72      7237\n",
      "\n",
      "    accuracy                           0.73     14175\n",
      "   macro avg       0.74      0.73      0.73     14175\n",
      "weighted avg       0.74      0.73      0.73     14175\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(metrics.classification_report(y_test, y_pred_svmr))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### ROC Curve and AUC for RBF SVM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 509,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Use predict_proba to get the probability results of KNN Model\n",
    "fpr_svmr, tpr_svmr, thresh = roc_curve(y_test, y_pred_svmr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 570,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# ROC Curve\n",
    "plt.figure(1)\n",
    "plt.plot([0, 1], [0, 1], 'k--')\n",
    "plt.plot(fpr_svmr, tpr_svmr, label='SVM RBF Model')\n",
    "plt.xlabel('False positive rate')\n",
    "plt.ylabel('True positive rate')\n",
    "plt.title('ROC curve - SVM Liner Model')\n",
    "plt.legend(loc='best')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.4 Random Forest \n",
    "- The random forest is a classification algorithm consisting of many decisions trees. \n",
    "- It uses **bagging and feature randomness** when building each individual tree to try to create an uncorrelated forest of trees whose prediction by committee is more accurate than that of any individual tree.\n",
    "-  It fits a number of decision tree classifiers on various sub-samples of the dataset and **uses averaging to improve the predictive accuracy and control over-fitting**. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.4.1 Base Model: Random Forest\n",
    "- We use 10-fold Cross Validation to get the accuracy for RF. \n",
    "- The base model has an accuracy of 69.83%."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 364,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0.71 0.7  0.69 0.69 0.68 0.7  0.71 0.69 0.71 0.7 ]\n",
      "Model accuracy of Random Forest is 0.6982735405349175\n"
     ]
    }
   ],
   "source": [
    "# libraries\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "# Random Forest\n",
    "classifier_RF = RandomForestClassifier()\n",
    "\n",
    "# Use 10-fold Cross Validation to get the accuracy for RF\n",
    "cv_score = model_selection.cross_val_score(classifier_RF, X_train, y_train, cv=10)\n",
    "print(cv_score)\n",
    "print('Model accuracy of Random Forest is', str(cv_score.mean()))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.4.2 Find Optimal Hyperparameter: Random Forest"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Max_depth: train the model with different max depth ranging from 1 to 10 and calculate the accuracy of each model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 374,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.71, 0.73, 0.73, 0.73, 0.73, 0.73, 0.73, 0.73, 0.74, 0.73, 0.74,\n",
       "       0.74, 0.73, 0.73, 0.73])"
      ]
     },
     "execution_count": 374,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "md = 16\n",
    "mean_acc = np.zeros((md-1))\n",
    "std_acc = np.zeros((md-1))\n",
    "for n in range(1,md):\n",
    "    \n",
    "    #Train Model and Predict  \n",
    "    RF = RandomForestClassifier(criterion=\"entropy\", max_depth = n).fit(X_train,y_train)\n",
    "    yhat_RF =RF.predict(X_test)\n",
    "    mean_acc[n-1] = metrics.accuracy_score(y_test, yhat_RF)\n",
    "\n",
    "    \n",
    "    std_acc[n-1]=np.std(yhat_RF==y_test)/np.sqrt(yhat_RF.shape[0])\n",
    "\n",
    "mean_acc"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Plot model accuracy for different number of max depth¶\n",
    "- **Disucssion of overfitting and underfitting**: \n",
    "    - If we set max_depth too high, then the decision tree might simply overfit the training data without capturing useful patterns as we would like; this will cause testing error to increase. \n",
    "    - If we set it too low, that is not good as well; then we might be giving the decision tree too little flexibility to capture the patterns and interactions in the training data. This will also cause the testing error to increase.\n",
    "    - There are some nice golden spots in between the extremes of too-high and too-low. As shown below, the optimal max_depth might be around 3 and 9."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 375,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(range(1,md),mean_acc,'r')\n",
    "plt.fill_between(range(1,md),mean_acc - 1 * std_acc,mean_acc + 1 * std_acc, alpha=0.10)\n",
    "plt.legend(('Accuracy ', '+/- 3xstd'))\n",
    "plt.ylabel('Accuracy ')\n",
    "plt.xlabel('Number of Max Depth')\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### More efficient parameter tuning using GridSearchCV: RF\n",
    "- **n_estimators:** The number of trees in the forest. The default value of n_estimators is 100.\n",
    "- **criterion:** The function to measure the quality of a split. Supported criteria are “gini” for the Gini impurity and “entropy” for the information gain.\n",
    "- **max_depth**: The maximum depth of the tree. If None, then nodes are expanded until all leaves are pure or until all leaves contain less than min_samples_split samples. \n",
    "- **max_features:** Max number of features considered for splitting a node\n",
    "- **bootstrap:** method for sampling data points (with or without replacement) By default, it bootstrap = True"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 388,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=5, error_score=nan,\n",
       "             estimator=RandomForestClassifier(bootstrap=True, ccp_alpha=0.0,\n",
       "                                              class_weight=None,\n",
       "                                              criterion='gini', max_depth=None,\n",
       "                                              max_features='auto',\n",
       "                                              max_leaf_nodes=None,\n",
       "                                              max_samples=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",
       "                                              n_estimators=100, n_jobs=None,\n",
       "                                              oob_score=False,\n",
       "                                              random_state=None, verbose=0,\n",
       "                                              warm_start=False),\n",
       "             iid='deprecated', n_jobs=None,\n",
       "             param_grid={'criterion': ['gini', 'entropy'],\n",
       "                         'max_depth': [8, 9, 10],\n",
       "                         'max_features': ['auto', 'sqrt', 'log2'],\n",
       "                         'n_estimators': [27, 28, 29, 30, 31]},\n",
       "             pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n",
       "             scoring=None, verbose=0)"
      ]
     },
     "execution_count": 388,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Possible hyperparamter options for Random Forest\n",
    "# Choose the number of trees\n",
    "parameters = {\n",
    "    'n_estimators' : [27,28,29,30,31],\n",
    "    'criterion':['gini','entropy'],\n",
    "    'max_depth': [ 8, 9, 10],\n",
    "    'max_features':[\"auto\",\"sqrt\",\"log2\"]\n",
    "}\n",
    "Grid_RF = GridSearchCV(RandomForestClassifier(),parameters, cv=5)\n",
    "Grid_RF.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 389,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best score: 0.7326824224297116\n",
      "Best parameters set\n",
      "criterion : entropy\n",
      "max_depth : 10\n",
      "max_features : sqrt\n",
      "n_estimators : 30\n"
     ]
    }
   ],
   "source": [
    "# best number of tress\n",
    "print_grid_search_metrics(Grid_RF)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 390,
   "metadata": {},
   "outputs": [],
   "source": [
    "# best random forest\n",
    "best_RF_model = Grid_RF.best_estimator_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 391,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(bootstrap=True, ccp_alpha=0.0, class_weight=None,\n",
       "                       criterion='entropy', max_depth=10, max_features='sqrt',\n",
       "                       max_leaf_nodes=None, max_samples=None,\n",
       "                       min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "                       min_samples_leaf=1, min_samples_split=2,\n",
       "                       min_weight_fraction_leaf=0.0, n_estimators=30,\n",
       "                       n_jobs=None, oob_score=False, random_state=None,\n",
       "                       verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 391,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "best_RF_model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.4.3 Model Performance Evaluation: RF"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Get Repeated Hold Out Accurary of Optimal RF Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- **Cross validation:** Use 10-fold Cross Validation to give a more accurate indication of how well the model generalizes to unseen data.\n",
    "- **Model Performance Evaluation:** The model accuracy of the RF model is 73.11% "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 564,
   "metadata": {},
   "outputs": [],
   "source": [
    "rf_pipe = make_pipeline(best_RF_model)\n",
    "rf_pipe.fit(X_train, y_train)\n",
    "pipe_pred = rf_pipe.predict(X_test)\n",
    "from sklearn.model_selection import KFold\n",
    "cv = KFold(n_splits=10, shuffle=True, random_state=16)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 563,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cv_acc_mean 0.7296\n"
     ]
    }
   ],
   "source": [
    "print('cv_acc_mean',round(cross_val_score(rf_pipe, X_train, y_train, cv=cv).mean(),4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 562,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cv_acc_std 0.006\n"
     ]
    }
   ],
   "source": [
    "print('cv_acc_std',round(cross_val_score(rf_pipe, X_train, y_train, cv=cv).std(),4))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Confusion Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 565,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred_rf = best_RF_model.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 566,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.7356\n",
      "Precision: 0.7717\n",
      "Recall: 0.6847\n",
      "F1-Score: 0.7256\n",
      "ROC AUC Score: 0.7367\n"
     ]
    }
   ],
   "source": [
    "print(f\"Accuracy: {round(metrics.accuracy_score(y_test, y_pred_rf), 4)}\")\n",
    "print(f\"Precision: {round(metrics.precision_score(y_test, y_pred_rf), 4)}\")\n",
    "print(f\"Recall: {round(metrics.recall_score(y_test, y_pred_rf), 4)}\")\n",
    "print(f\"F1-Score: {round(metrics.f1_score(y_test, y_pred_rf), 4)}\")\n",
    "print(f\"ROC AUC Score: {round(metrics.roc_auc_score(y_test, y_pred_rf), 4)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 392,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.70      0.79      0.74      6938\n",
      "           1       0.77      0.68      0.72      7237\n",
      "\n",
      "    accuracy                           0.73     14175\n",
      "   macro avg       0.74      0.74      0.73     14175\n",
      "weighted avg       0.74      0.73      0.73     14175\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(metrics.classification_report(y_test,y_pred_rf))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### ROC and AUC for RF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 393,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Use predict_proba to get the probability results of KNN Model\n",
    "y_pred_rf = best_RF_model.predict_proba(X_test)[:, 1]\n",
    "fpr_rf, tpr_rf, thresh = roc_curve(y_test, y_pred_rf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 394,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8003074707411661"
      ]
     },
     "execution_count": 394,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# AUC score\n",
    "metrics.auc(fpr_rf,tpr_rf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 572,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# ROC Curve\n",
    "plt.figure(1)\n",
    "plt.plot([0, 1], [0, 1], 'k--')\n",
    "plt.plot(fpr_rf, tpr_rf, label='Random Forest Model')\n",
    "plt.xlabel('False positive rate')\n",
    "plt.ylabel('True positive rate')\n",
    "plt.title('ROC curve - Random Forest Model')\n",
    "plt.legend(loc='best')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.5 Gradient Boosting Machine "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.5.1 Base Model: GBM\n",
    "- We use 10-fold Cross Validation to get the accuracy for GBM Base Model. \n",
    "- The base model has an accuracy of 73.40%."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 438,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy for GBM: 73.34%\n"
     ]
    }
   ],
   "source": [
    "from sklearn.ensemble import GradientBoostingClassifier as gbm\n",
    "num_trees = 100\n",
    "gbm_model = gbm(n_estimators=num_trees, random_state=16)\n",
    "cv = 10 \n",
    "results = cross_val_score(gbm_model, X_train, y_train, cv=cv)\n",
    "print(f\"Accuracy for GBM: {round(results.mean()*100, 2)}%\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.5.2 Model Tuning"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Discussion on Overfitting for Gradient Boosting Machine:\n",
    "- **We can generally address the overfitting for a gradient boosted machine in by**\n",
    "    - Reducing the number of dependent features by removing all non-significant and correlated features from the data set.\n",
    "    - Increasing the number of training examples.\n",
    "- **At this stage it might not be possible to reduce features or increase size of training dataset. In such case we use regularisation to smoothen the cost function and reduce high variance and increase bias.**\n",
    "   - Regularization via shrinkage (learning_rate < 1.0) improves performance considerably. \n",
    "   - In combination with shrinkage, stochastic gradient boosting (subsample < 1.0) can produce more accurate models by reducing the variance via bagging. \n",
    "   - Therefore, we experiment with subsample and learning rate."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### More efficient parameter tuning using GridSearchCV: RF"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- **loss**:loss function to be optimized. ‘deviance’ refers to deviance (= logistic regression) for classification with probabilistic outputs. For loss ‘exponential’ gradient boosting recovers the AdaBoost algorithm.\n",
    "-  **learning_rate**: learning rate shrinks the contribution of each tree by learning_rate. There is a trade-off between learning_rate and n_estimators. \n",
    "- **n_estimators**: The number of boosting stages to perform. Gradient boosting is fairly robust to over-fitting so a large number usually results in better performance.\n",
    "-  **max_depth(default=3)**: maximum depth of the individual regression estimators. The maximum depth limits the number of nodes in the tree. Tune this parameter for best performance; the best value depends on the interaction of the input variables.\n",
    "-  random_state=0\n",
    "-  **subsample(default=1.0)**: The fraction of samples to be used for fitting the individual base learners. If smaller than 1.0 this results in Stochastic Gradient Boosting. subsample interacts with the parameter n_estimators. Choosing subsample < 1.0 leads to a reduction of variance and an increase in bias."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 400,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=None, error_score=nan,\n",
       "             estimator=GradientBoostingClassifier(ccp_alpha=0.0,\n",
       "                                                  criterion='friedman_mse',\n",
       "                                                  init=None, learning_rate=0.1,\n",
       "                                                  loss='deviance', max_depth=3,\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",
       "                                                  n_estimators=100,\n",
       "                                                  n_iter_n...\n",
       "                                                  subsample=1.0, tol=0.0001,\n",
       "                                                  validation_fraction=0.1,\n",
       "                                                  verbose=0, warm_start=False),\n",
       "             iid='deprecated', n_jobs=None,\n",
       "             param_grid={'learning_rate': array([0.02, 0.04, 0.06, 0.08]),\n",
       "                         'loss': ['deviance', 'exponential'],\n",
       "                         'max_depth': range(2, 10),\n",
       "                         'n_estimators': range(60, 160, 50),\n",
       "                         'subsample': [1, 0.5]},\n",
       "             pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n",
       "             scoring=None, verbose=0)"
      ]
     },
     "execution_count": 400,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.model_selection import GridSearchCV\n",
    "param_grid = {'learning_rate': np.arange(0.02, 0.1, 0.02),\n",
    "              'n_estimators': range(60, 160, 50),\n",
    "              'max_depth': range(2, 10),\n",
    "              'loss':['deviance','exponential'],\n",
    "              'subsample':[1,0.5]}\n",
    "clf = GridSearchCV(gbm(), param_grid)\n",
    "clf.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 401,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best hyperparameter is learning_rate: 0.06\n",
      "Best hyperparameter is loss: deviance\n",
      "Best hyperparameter is max_depth: 4\n",
      "Best hyperparameter is n_estimators: 110\n",
      "Best hyperparameter is subsample: 0.5\n"
     ]
    }
   ],
   "source": [
    "for key, val in clf.best_params_.items():\n",
    "  print(f\"Best hyperparameter is {key}: {val}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 439,
   "metadata": {},
   "outputs": [],
   "source": [
    "# best gradient boosting machine \n",
    "best_GBM_model = clf.best_estimator_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 403,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GradientBoostingClassifier(ccp_alpha=0.0, criterion='friedman_mse', init=None,\n",
       "                           learning_rate=0.06, loss='deviance', max_depth=4,\n",
       "                           max_features=None, max_leaf_nodes=None,\n",
       "                           min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "                           min_samples_leaf=1, min_samples_split=2,\n",
       "                           min_weight_fraction_leaf=0.0, n_estimators=110,\n",
       "                           n_iter_no_change=None, presort='deprecated',\n",
       "                           random_state=None, subsample=0.5, tol=0.0001,\n",
       "                           validation_fraction=0.1, verbose=0,\n",
       "                           warm_start=False)"
      ]
     },
     "execution_count": 403,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "best_GBM_model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2.5.3 Model Performance Evluation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Get Repeated Hold Out Accurary of Optimal GBM Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- **Cross validation:** Use 10-fold Cross Validation to give a more accurate indication of how well the model generalizes to unseen data.\n",
    "- **Model Performance Evaluation:** The model accuracy of the RF model is 73.42% "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 575,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7332868811129081"
      ]
     },
     "execution_count": 575,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gbm_pipe = make_pipeline(best_GBM_model)\n",
    "gbm_pipe.fit(X_train, y_train)\n",
    "pipe_pred = gbm_pipe.predict(X_test)\n",
    "from sklearn.model_selection import KFold\n",
    "cv = KFold(n_splits=10, shuffle=True, random_state=16)\n",
    "cross_val_score(gbm_pipe, X_train, y_train, cv=cv).mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 577,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cv_acc_mean 0.7333\n"
     ]
    }
   ],
   "source": [
    "print('cv_acc_mean',round(cross_val_score(gbm_pipe, X_train, y_train, cv=cv).mean(),4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 578,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cv_acc_std 0.0061\n"
     ]
    }
   ],
   "source": [
    "print('cv_acc_std',round(cross_val_score(gbm_pipe, X_train, y_train, cv=cv).std(),4))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Confusion Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 573,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred_gbm = best_GBM_model.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 574,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.7369\n",
      "Precision: 0.7579\n",
      "Recall: 0.712\n",
      "F1-Score: 0.7343\n",
      "ROC AUC Score: 0.7374\n"
     ]
    }
   ],
   "source": [
    "print(f\"Accuracy: {round(metrics.accuracy_score(y_test, y_pred_gbm), 4)}\")\n",
    "print(f\"Precision: {round(metrics.precision_score(y_test, y_pred_gbm), 4)}\")\n",
    "print(f\"Recall: {round(metrics.recall_score(y_test, y_pred_gbm), 4)}\")\n",
    "print(f\"F1-Score: {round(metrics.f1_score(y_test,y_pred_gbm), 4)}\")\n",
    "print(f\"ROC AUC Score: {round(metrics.roc_auc_score(y_test, y_pred_gbm), 4)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 407,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.72      0.76      0.74      6938\n",
      "           1       0.76      0.71      0.73      7237\n",
      "\n",
      "    accuracy                           0.74     14175\n",
      "   macro avg       0.74      0.74      0.74     14175\n",
      "weighted avg       0.74      0.74      0.74     14175\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(metrics.classification_report(y_test,y_pred_gbm))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### ROC and AUC for GBM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 408,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Use predict_proba to get the probability results of KNN Model\n",
    "y_pred_gbm = best_GBM_model.predict_proba(X_test)[:, 1]\n",
    "fpr_gbm, tpr_gbm, thresh = roc_curve(y_test, y_pred_gbm)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 409,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8026573867126005"
      ]
     },
     "execution_count": 409,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# AUC score\n",
    "metrics.auc(fpr_gbm,tpr_gbm)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 580,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# ROC Curve\n",
    "plt.figure(1)\n",
    "plt.plot([0, 1], [0, 1], 'k--')\n",
    "plt.plot(fpr_gbm, tpr_gbm, label='GBM Model')\n",
    "plt.xlabel('False positive rate')\n",
    "plt.ylabel('True positive rate')\n",
    "plt.title('ROC curve - GBM Model')\n",
    "plt.legend(loc='best')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Section 3: Predcition and Interpretation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.1 Load and Process Testing Data\n",
    "- The training and testing data should undergo the same data preparation steps or the predictive model will not make sense."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Test_df Basic Info"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 414,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of rows: 21000\n",
      "Number of columns: 11\n"
     ]
    },
    {
     "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>Gender</th>\n",
       "      <th>Height</th>\n",
       "      <th>Weight</th>\n",
       "      <th>High_Blood_Pressure</th>\n",
       "      <th>Low_Blood_Pressure</th>\n",
       "      <th>Cholesterol</th>\n",
       "      <th>Glucose</th>\n",
       "      <th>Smoke</th>\n",
       "      <th>Alcohol</th>\n",
       "      <th>Exercise</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ID</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>44</td>\n",
       "      <td>female</td>\n",
       "      <td>160</td>\n",
       "      <td>59.0</td>\n",
       "      <td>100</td>\n",
       "      <td>80</td>\n",
       "      <td>high</td>\n",
       "      <td>normal</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>41</td>\n",
       "      <td>female</td>\n",
       "      <td>169</td>\n",
       "      <td>74.0</td>\n",
       "      <td>120</td>\n",
       "      <td>70</td>\n",
       "      <td>normal</td>\n",
       "      <td>normal</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>63</td>\n",
       "      <td>male</td>\n",
       "      <td>168</td>\n",
       "      <td>84.0</td>\n",
       "      <td>120</td>\n",
       "      <td>80</td>\n",
       "      <td>normal</td>\n",
       "      <td>high</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    Age  Gender  Height  Weight  High_Blood_Pressure  Low_Blood_Pressure  \\\n",
       "ID                                                                         \n",
       "0    44  female     160    59.0                  100                  80   \n",
       "1    41  female     169    74.0                  120                  70   \n",
       "2    63    male     168    84.0                  120                  80   \n",
       "\n",
       "   Cholesterol Glucose  Smoke  Alcohol  Exercise  \n",
       "ID                                                \n",
       "0         high  normal      0        0         1  \n",
       "1       normal  normal      0        0         1  \n",
       "2       normal    high      0        0         1  "
      ]
     },
     "execution_count": 414,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# test_df\n",
    "print('Number of rows:', test_df.shape[0])\n",
    "print('Number of columns:', test_df.shape[1])\n",
    "\n",
    "# rename the high and low blood pressure columns\n",
    "test_df.rename(columns={'High Blood Pressure':'High_Blood_Pressure',\n",
    "                         'Low Blood Pressure':'Low_Blood_Pressure'},inplace=True)\n",
    "# print test_df\n",
    "test_df.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Missing values "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 415,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Age                    0\n",
       "Gender                 0\n",
       "Height                 0\n",
       "Weight                 0\n",
       "High_Blood_Pressure    0\n",
       "Low_Blood_Pressure     0\n",
       "Cholesterol            0\n",
       "Glucose                0\n",
       "Smoke                  0\n",
       "Alcohol                0\n",
       "Exercise               0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 415,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# check for missing values \n",
    "test_df.isnull().sum(axis=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Duplicates"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 417,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "No duplicated rows\n",
      "Number of rows after removing duplicates: 20434\n",
      "Number of duplicate rows removed: 566\n"
     ]
    }
   ],
   "source": [
    "# check for duplicates \n",
    "if test_df.duplicated().sum()==0:\n",
    "    print('No duplicated rows')\n",
    "else:\n",
    "    print('There are duplicates rows in the dataset.')\n",
    "test_df.drop_duplicates(inplace=True)\n",
    "print(\"Number of rows after removing duplicates:\", test_df.shape[0])\n",
    "print(\"Number of duplicate rows removed:\", 21000-test_df.shape[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Outliers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 418,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/jieerchen/opt/anaconda3/lib/python3.7/site-packages/pandas/core/indexing.py:671: SettingWithCopyWarning:\n",
      "\n",
      "\n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# preprocess outliers \n",
    "outliers_cols = ['Age','Height','Weight','High_Blood_Pressure','Low_Blood_Pressure']\n",
    "for target_col in outliers_cols:\n",
    "    '''\n",
    "    For each numeric column, we use IQR method to identify outliers.\n",
    "    Column values higher than the upper bound are set to upper bound values.\n",
    "    Column values lower thatn the lower bound value are set to lower bound values.\n",
    "    \n",
    "    '''\n",
    "    lower_quartile = test_df[target_col].quantile(0.25)\n",
    "    upper_quartile = test_df[target_col].quantile(0.75)\n",
    "    IQR = upper_quartile - lower_quartile\n",
    "    outlier_thresh = 1.5 * IQR\n",
    "    # get upper/lower bound\n",
    "    upper_bound = lower_quartile + outlier_thresh\n",
    "    lower_bound = lower_quartile - outlier_thresh\n",
    "    # winsorization\n",
    "    test_df[target_col].loc[test_df[target_col] > upper_bound] = upper_bound\n",
    "    test_df[target_col].loc[test_df[target_col] < lower_bound] = lower_bound\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Negative values "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 419,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of negatives values in High Blood Pressure Column: 0\n"
     ]
    }
   ],
   "source": [
    "# number of rows with negative values in High_Blood_Pressure\n",
    "negative_values = sum( n <0 for n in test_df['High_Blood_Pressure'].values.flatten())\n",
    "print(\"Number of negatives values in High Blood Pressure Column:\", negative_values)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Feature Encoding "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 420,
   "metadata": {},
   "outputs": [],
   "source": [
    "# categorical variables\n",
    "test_df['Gender'] = test_df['Gender'].map({'male': 0, 'female': 1})\n",
    "test_df = pd.get_dummies(test_df, columns=['Cholesterol'])\n",
    "test_df = pd.get_dummies(test_df, columns=['Glucose'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Ready for Prediction Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 421,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of rows: 20434\n",
      "Number of columns: 15\n"
     ]
    },
    {
     "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>Gender</th>\n",
       "      <th>Height</th>\n",
       "      <th>Weight</th>\n",
       "      <th>High_Blood_Pressure</th>\n",
       "      <th>Low_Blood_Pressure</th>\n",
       "      <th>Smoke</th>\n",
       "      <th>Alcohol</th>\n",
       "      <th>Exercise</th>\n",
       "      <th>Cholesterol_high</th>\n",
       "      <th>Cholesterol_normal</th>\n",
       "      <th>Cholesterol_too high</th>\n",
       "      <th>Glucose_high</th>\n",
       "      <th>Glucose_normal</th>\n",
       "      <th>Glucose_too high</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ID</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>44.0</td>\n",
       "      <td>1</td>\n",
       "      <td>160.0</td>\n",
       "      <td>59.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>80.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>41.0</td>\n",
       "      <td>1</td>\n",
       "      <td>169.0</td>\n",
       "      <td>74.0</td>\n",
       "      <td>120.0</td>\n",
       "      <td>70.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>63.0</td>\n",
       "      <td>0</td>\n",
       "      <td>168.0</td>\n",
       "      <td>84.0</td>\n",
       "      <td>120.0</td>\n",
       "      <td>80.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     Age  Gender  Height  Weight  High_Blood_Pressure  Low_Blood_Pressure  \\\n",
       "ID                                                                          \n",
       "0   44.0       1   160.0    59.0                100.0                80.0   \n",
       "1   41.0       1   169.0    74.0                120.0                70.0   \n",
       "2   63.0       0   168.0    84.0                120.0                80.0   \n",
       "\n",
       "    Smoke  Alcohol  Exercise  Cholesterol_high  Cholesterol_normal  \\\n",
       "ID                                                                   \n",
       "0       0        0         1                 1                   0   \n",
       "1       0        0         1                 0                   1   \n",
       "2       0        0         1                 0                   1   \n",
       "\n",
       "    Cholesterol_too high  Glucose_high  Glucose_normal  Glucose_too high  \n",
       "ID                                                                        \n",
       "0                      0             0               1                 0  \n",
       "1                      0             0               1                 0  \n",
       "2                      0             1               0                 0  "
      ]
     },
     "execution_count": 421,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# test_df\n",
    "print('Number of rows:', test_df.shape[0])\n",
    "print('Number of columns:', test_df.shape[1])\n",
    "test_df.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Data Scaling "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 425,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.37, 1.  , 0.53, 0.38, 0.17, 0.5 , 0.  , 0.  , 1.  , 1.  , 0.  ,\n",
       "        0.  , 0.  , 1.  , 0.  ],\n",
       "       [0.27, 1.  , 0.8 , 0.68, 0.5 , 0.17, 0.  , 0.  , 1.  , 0.  , 1.  ,\n",
       "        0.  , 0.  , 1.  , 0.  ]])"
      ]
     },
     "execution_count": 425,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.preprocessing import MinMaxScaler\n",
    "X_testset = test_df\n",
    "X_testset = preprocessing.MinMaxScaler().fit(X_testset).transform(X_testset)\n",
    "X_testset[0:2]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.3 Make Prediction "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### NBC: Guassian Navie Bayes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 602,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 0, 1, 1, 0, 0, 0, 1, 1, 1])"
      ]
     },
     "execution_count": 602,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "NBC = gnb.predict(X_testset)\n",
    "NBC[0:10]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### KNN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 603,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 0, 0, 1, 0, 1, 0, 1, 1, 1])"
      ]
     },
     "execution_count": 603,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "KNN = best_KNN_model.predict(X_testset)\n",
    "KNN[0:10]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### SVM-Linear"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 604,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 0, 1, 1, 0, 1, 0, 1, 1, 1])"
      ]
     },
     "execution_count": 604,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "SVM_Linear = best_svm_linear.predict(X_testset)\n",
    "SVM_Linear[0:10]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### SVM-RBF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 606,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 0, 0, 1, 0, 1, 0, 1, 1, 1])"
      ]
     },
     "execution_count": 606,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "SVM_RBF = best_svm_rbf.predict(X_testset)\n",
    "SVM_RBF[0:10]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### RF: Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 434,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 0, 1, 1, 0, 0, 0, 1, 1, 1])"
      ]
     },
     "execution_count": 434,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "RF = best_RF_model.predict(X_testset)\n",
    "RF[0:10]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### GBM: Gradient Boosting Machine"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 607,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 0, 1, 1, 0, 1, 0, 1, 1, 1])"
      ]
     },
     "execution_count": 607,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "GBM = best_GBM_model.predict(X_testset)\n",
    "GBM[0:10]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Output prediction result as a CSV file¶"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 608,
   "metadata": {},
   "outputs": [],
   "source": [
    "output_df = pd.read_csv('Disease Prediction Testing.csv')\n",
    "prediction = pd.DataFrame(list(zip(output_df['ID'], NBC,KNN,SVM_Linear,SVM_RBF,RF,GBM)),columns=['ID','NBC','KNN','SVM-Linear','SVM-RBF','RF','GBM'])\n",
    "prediction.to_csv('Disease_Prediction.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 609,
   "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>ID</th>\n",
       "      <th>NBC</th>\n",
       "      <th>KNN</th>\n",
       "      <th>SVM-Linear</th>\n",
       "      <th>SVM-RBF</th>\n",
       "      <th>RF</th>\n",
       "      <th>GBM</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>6</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>7</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>9</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   ID  NBC  KNN  SVM-Linear  SVM-RBF  RF  GBM\n",
       "0   0    0    0           0        0   0    0\n",
       "1   1    0    0           0        0   0    0\n",
       "2   2    1    0           1        0   1    1\n",
       "3   3    1    1           1        1   1    1\n",
       "4   4    0    0           0        0   0    0\n",
       "5   5    0    1           1        1   0    1\n",
       "6   6    0    0           0        0   0    0\n",
       "7   7    1    1           1        1   1    1\n",
       "8   8    1    1           1        1   1    1\n",
       "9   9    1    1           1        1   1    1"
      ]
     },
     "execution_count": 609,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "prediction.head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.3 Model Performance Summary and Interpretation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 631,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_performance = {'Metrics': ['Accuracy','Precision','Recall','F1-Score','CV_mean','CV_std','ROC_AUC','Parameters'], \n",
    "           'Gussain': [0.6874, 0.7367,0.6034,0.6634,0.6885,0.0108,0.6892,'None'],\n",
    "           'KNN':[0.7307,0.7592,0.6919,0.724,0.7243,0.0064,0.7248,'metric : manhattan,n_neighbors : 29,weights : uniform'],\n",
    "           'SVM_Linear':[0.7307,0.7592,0.6919,0.724,0.7267,0.0058,0.7315,'penalty: l2,C: 0.25'],\n",
    "           'SVM_RBF':[0.7329,0.7688,0.6819,0.7228,0.7311,0.0057,0.7340,'C: 0.7,gamma: 0.5'],  \n",
    "           'RF':[0.7356,0.7717,0.6847,0.7256,0.7296,0.006,0.7367,'criterion : entropy,max_depth : 10,max_features : sqrt,n_estimators : 30'],\n",
    "           'GBM':[0.7369, 0.7579,0.712,0.7343,0.7333,0.0061, 0.7374,'learning_rate: 0.06,loss: deviance,max_depth: 4,n_estimators: 110,subsample: 0.5']}\n",
    "                  \n",
    "model_performance = pd.DataFrame(data= model_performance)\n",
    "model_performance.set_index('Metrics',inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 3.3.2 Summary and Interpretation of each Individual Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 1. Summary and Interpretation of each Individual Model: Navie Bayes Algorithm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 593,
   "metadata": {
    "scrolled": false
   },
   "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>Metrics</th>\n",
       "      <th>Gussain</th>\n",
       "      <th>Multilnomial</th>\n",
       "      <th>Bernoulli</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Accuracy</td>\n",
       "      <td>0.6874</td>\n",
       "      <td>0.6140</td>\n",
       "      <td>0.5898</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Precision</td>\n",
       "      <td>0.7367</td>\n",
       "      <td>0.6851</td>\n",
       "      <td>0.6651</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Recall</td>\n",
       "      <td>0.6034</td>\n",
       "      <td>0.4513</td>\n",
       "      <td>0.3960</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>F1-Score</td>\n",
       "      <td>0.6634</td>\n",
       "      <td>0.5442</td>\n",
       "      <td>0.4964</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>CV_mean</td>\n",
       "      <td>0.6885</td>\n",
       "      <td>0.6109</td>\n",
       "      <td>0.5882</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>CV_std</td>\n",
       "      <td>0.0108</td>\n",
       "      <td>0.0111</td>\n",
       "      <td>0.0100</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>AUC</td>\n",
       "      <td>0.6892</td>\n",
       "      <td>0.6175</td>\n",
       "      <td>0.5940</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     Metrics  Gussain  Multilnomial  Bernoulli\n",
       "0   Accuracy   0.6874        0.6140     0.5898\n",
       "1  Precision   0.7367        0.6851     0.6651\n",
       "2     Recall   0.6034        0.4513     0.3960\n",
       "3   F1-Score   0.6634        0.5442     0.4964\n",
       "4    CV_mean   0.6885        0.6109     0.5882\n",
       "5     CV_std   0.0108        0.0111     0.0100\n",
       "6        AUC   0.6892        0.6175     0.5940"
      ]
     },
     "execution_count": 593,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nym_performance "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- **Gussain NB model has outperformed the other two model given that it has the highest values for Accuracy(0.6874), Precision(0.7367), Recall(0.6034) and largest AUC (0.6892).** \n",
    "     - **Accuracy**: 0.74% of the predictions made by the model are correct.\n",
    "     - **Precision** (0.60 for disease = 1): 60% of the cases are correctly predicted among all the cases that are predicted as having disease.\n",
    "     - **Recall** (0.66 for disease =1): 66% of the cases are correctly identified by the model among all the cases are diagnoised with disease \n",
    "     - **F-1 score 0.66**: F-1 score is 66%. It takes into consideration of both Precision and Recall.\n",
    "     - **AUC**: Gussain NB model has the largest area under the curve: 0.76, indicating that Gussain NB model is the most useful among all.\n",
    "     - **Cross validation accuracy (mean & std)**: we also observe that the variations of model accuracy for each model is small and that they have relatively unbaised performance and low variances.\n",
    "- **Adtional Comments on the NB model Performance**\n",
    "    - Naive Bayes classifier assumes that the presence of a particular feature in a class is unrelated to the presence of any other feature. However, in real life, it is almost impossible that we get a set of predictors which are completely independent. \n",
    "    - According to our previous correlation matrix, we see that many of our variables are correlated with others. This could be one of the reason that we are not achieving higher accuracy that we've got."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2. Summary and Interpretation: KNN "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 601,
   "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>KNN</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Metrics</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Accuracy</th>\n",
       "      <td>0.7307</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Precision</th>\n",
       "      <td>0.7592</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Recall</th>\n",
       "      <td>0.6919</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>F1-Score</th>\n",
       "      <td>0.724</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CV_mean</th>\n",
       "      <td>0.7243</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CV_std</th>\n",
       "      <td>0.0064</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ROC_AUC</th>\n",
       "      <td>0.7248</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Parameters</th>\n",
       "      <td>metric : manhattan,n_neighbors : 29,weights : uniform</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                              KNN\n",
       "Metrics                                                          \n",
       "Accuracy                                                   0.7307\n",
       "Precision                                                  0.7592\n",
       "Recall                                                     0.6919\n",
       "F1-Score                                                    0.724\n",
       "CV_mean                                                    0.7243\n",
       "CV_std                                                     0.0064\n",
       "ROC_AUC                                                    0.7248\n",
       "Parameters  metric : manhattan,n_neighbors : 29,weights : uniform"
      ]
     },
     "execution_count": 601,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_performance[['KNN']]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- **Confusion Matrices**\n",
    "     - **Accuracy**: around 73.07% of the predictions made by the model are correct.\n",
    "     - **Precision** (0.7592 for disease = 1): around 75.92% of the cases are correctly predicted among all the cases that are predicted as having disease.\n",
    "     - **Recall** (0.6919 for disease =1): around 69.19% of the cases are correctly identified by the model among all the cases are diagnoised with disease \n",
    "     - **F-1 score 0.72**: F-1 score is 0.724. It takes into consideration of both Precision and Recall.\n",
    "- **ROC curve and AUC score**\n",
    "    - **Result**: The AUC of the best KNN model is around 0.0064.\n",
    "- **Cross-validation to help in identifying a case of overfitting**\n",
    "    - **Result:** The average accuracy is around 0.7248. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 3. Summary and Interpretation: SVM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 600,
   "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>SVM_Linear</th>\n",
       "      <th>SVM_RBF</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Metrics</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Accuracy</th>\n",
       "      <td>0.7307</td>\n",
       "      <td>0.7329</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Precision</th>\n",
       "      <td>0.7592</td>\n",
       "      <td>0.7688</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Recall</th>\n",
       "      <td>0.6919</td>\n",
       "      <td>0.6819</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>F1-Score</th>\n",
       "      <td>0.724</td>\n",
       "      <td>0.7228</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CV_mean</th>\n",
       "      <td>0.7267</td>\n",
       "      <td>0.7311</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CV_std</th>\n",
       "      <td>0.0058</td>\n",
       "      <td>0.0057</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ROC_AUC</th>\n",
       "      <td>0.7315</td>\n",
       "      <td>0.734</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Parameters</th>\n",
       "      <td>penalty: l2,C: 0.25</td>\n",
       "      <td>C: 0.7,gamma: 0.5</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                     SVM_Linear            SVM_RBF\n",
       "Metrics                                           \n",
       "Accuracy                 0.7307             0.7329\n",
       "Precision                0.7592             0.7688\n",
       "Recall                   0.6919             0.6819\n",
       "F1-Score                  0.724             0.7228\n",
       "CV_mean                  0.7267             0.7311\n",
       "CV_std                   0.0058             0.0057\n",
       "ROC_AUC                  0.7315              0.734\n",
       "Parameters  penalty: l2,C: 0.25  C: 0.7,gamma: 0.5"
      ]
     },
     "execution_count": 600,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_performance[['SVM_Linear','SVM_RBF']]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### SVM Linear and SVM RBF has very close performance. If we give more weight to cross validation accuracy and AUC score, SVM RBF perform slightly better than SVM Linear. \n",
    "- **Confusion Matrix**\n",
    "    - SCM Linear has higher precission while SVM RBF has higher recall. Therefore, we would look at the F1-score which takes into consideration of both precision and recall. As a reseult, SVM Linear has sightly better performance.\n",
    "- **Cross-validation to help in identifying a case of overfitting**\n",
    "    - **Result:** SVM RBF has a higher cross valiation mean accuarcy score (0.7311 compared to 0.0058) and lower cross validation standard deviation accuracy score(0,0057 compared to 0.7267). \n",
    "- **ROC curve and AUC score**\n",
    "    - **Result**: The AUC of the best SVM model is around 0.734."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 4 Summary and Interpretation: RF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 611,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>RF</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Metrics</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Accuracy</th>\n",
       "      <td>0.7356</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Precision</th>\n",
       "      <td>0.7717</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Recall</th>\n",
       "      <td>0.6847</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>F1-Score</th>\n",
       "      <td>0.7256</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CV_mean</th>\n",
       "      <td>0.7296</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CV_std</th>\n",
       "      <td>0.006</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ROC_AUC</th>\n",
       "      <td>0.7367</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Parameters</th>\n",
       "      <td>criterion : entropy,max_depth : 10,max_features : sqrt,n_estimators : 30</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                                  RF\n",
       "Metrics                                                                             \n",
       "Accuracy                                                                      0.7356\n",
       "Precision                                                                     0.7717\n",
       "Recall                                                                        0.6847\n",
       "F1-Score                                                                      0.7256\n",
       "CV_mean                                                                       0.7296\n",
       "CV_std                                                                         0.006\n",
       "ROC_AUC                                                                       0.7367\n",
       "Parameters  criterion : entropy,max_depth : 10,max_features : sqrt,n_estimators : 30"
      ]
     },
     "execution_count": 611,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_performance[['RF']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 398,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f9b1e4bf950>"
      ]
     },
     "execution_count": 398,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "importances = best_RF_model.feature_importances_\n",
    "X_index = train_df.drop('Disease',axis=1)\n",
    "feat_importances = pd.Series(best_RF_model.feature_importances_, index=X_index.columns)\n",
    "feat_importances.nlargest(12).plot(kind='barh',color=\"salmon\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 397,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Feature ranking:\n",
      "1. feature 4 (0.396965)\n",
      "2. feature 5 (0.210927)\n",
      "3. feature 0 (0.130146)\n",
      "4. feature 3 (0.071704)\n",
      "5. feature 11 (0.049116)\n",
      "6. feature 2 (0.045143)\n",
      "7. feature 10 (0.029585)\n",
      "8. feature 8 (0.010802)\n",
      "9. feature 9 (0.008827)\n",
      "10. feature 6 (0.008689)\n",
      "11. feature 14 (0.008522)\n",
      "12. feature 1 (0.008306)\n",
      "13. feature 7 (0.007921)\n",
      "14. feature 13 (0.007527)\n",
      "15. feature 12 (0.005820)\n"
     ]
    }
   ],
   "source": [
    "std = np.std([best_RF_model.feature_importances_ for tree in best_RF_model.estimators_],\n",
    "             axis=0)\n",
    "indices = np.argsort(importances)[::-1]\n",
    "\n",
    "# Print the feature ranking\n",
    "print(\"Feature ranking:\")\n",
    "\n",
    "for f in range(X.shape[1]):\n",
    "    print(\"%d. feature %d (%f)\" % (f + 1, indices[f], importances[indices[f]]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Performance Summary**\n",
    "- **Confusion Matrices**\n",
    "     - **Accuracy**: 0.73.56% of the predictions made by the model are correct.\n",
    "     - **Precision** (0.77.17 for disease = 1): 77.17% of the cases are correctly predicted among all the cases that are predicted as having disease.\n",
    "     - **Recall** (0.6847 for disease =1): 68.47% of the cases are correctly identified by the model among all the cases are diagnoised with disease \n",
    "     - **F-1 score 0.7256**: F-1 score is 72%. It takes into consideration of both Precision and Recall.\n",
    "- **ROC curve and AUC score**\n",
    "    - **Result**: The AUC of the best RF model is around 0.7367.\n",
    "- **Cross-validation to help in identifying a case of overfitting**\n",
    "    - **Result:** The average accuracy is 0.7296. \n",
    "- **RF suggests that 6 features are informative, while the remaining are not. These features and their relative importance are**\n",
    "    1. **High_Blood_Pressure**: 0.396965\n",
    "    2. **Low_Blood_PressureWeight**: 0.210927\n",
    "    3. **Age**: 0.130146\n",
    "    4. **Weight**: 0.071704\n",
    "    5. **Cholesterol_too high**: 0.049116\n",
    "    6. **Height**: 0.045143"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 3.3.5  Summary and Interpretation: GBM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 613,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>GBM</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Metrics</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Accuracy</th>\n",
       "      <td>0.7369</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Precision</th>\n",
       "      <td>0.7579</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Recall</th>\n",
       "      <td>0.712</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>F1-Score</th>\n",
       "      <td>0.7343</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CV_mean</th>\n",
       "      <td>0.7333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CV_std</th>\n",
       "      <td>0.0061</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ROC_AUC</th>\n",
       "      <td>0.7374</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Parameters</th>\n",
       "      <td>learning_rate: 0.06,loss: deviance,max_depth: 4,n_estimators: 110,subsample: 0.5</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                                         GBM\n",
       "Metrics                                                                                     \n",
       "Accuracy                                                                              0.7369\n",
       "Precision                                                                             0.7579\n",
       "Recall                                                                                 0.712\n",
       "F1-Score                                                                              0.7343\n",
       "CV_mean                                                                               0.7333\n",
       "CV_std                                                                                0.0061\n",
       "ROC_AUC                                                                               0.7374\n",
       "Parameters  learning_rate: 0.06,loss: deviance,max_depth: 4,n_estimators: 110,subsample: 0.5"
      ]
     },
     "execution_count": 613,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_performance[['GBM']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 630,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f9b071b8e10>"
      ]
     },
     "execution_count": 630,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "importances_gbm = best_GBM_model.feature_importances_\n",
    "X_index = train_df.drop('Disease',axis=1)\n",
    "feat_importances_gbm = pd.Series(best_GBM_model.feature_importances_, index=X_index.columns)\n",
    "feat_importances_gbm.nlargest(12).plot(kind='barh',color=\"lightblue\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 629,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Feature ranking:\n",
      "1. feature 4 (0.673581)\n",
      "2. feature 0 (0.130931)\n",
      "3. feature 11 (0.048845)\n",
      "4. feature 3 (0.043179)\n",
      "5. feature 5 (0.028067)\n",
      "6. feature 10 (0.021341)\n",
      "7. feature 2 (0.020415)\n",
      "8. feature 8 (0.007134)\n",
      "9. feature 14 (0.006256)\n",
      "10. feature 6 (0.005230)\n",
      "11. feature 7 (0.004718)\n",
      "12. feature 13 (0.003354)\n",
      "13. feature 12 (0.003182)\n",
      "14. feature 1 (0.002442)\n",
      "15. feature 9 (0.001326)\n"
     ]
    }
   ],
   "source": [
    "std = np.std([best_GBM_model.feature_importances_ for tree in best_GBM_model.estimators_],\n",
    "             axis=0)\n",
    "indices = np.argsort(importances_gbm)[::-1]\n",
    "\n",
    "# Print the feature ranking\n",
    "print(\"Feature ranking:\")\n",
    "\n",
    "for f in range(X.shape[1]):\n",
    "    print(\"%d. feature %d (%f)\" % (f + 1, indices[f], importances_gbm[indices[f]]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- **Confusion Matrices**\n",
    "     - **Accuracy**: 0.74% of the predictions made by the model are correct.\n",
    "     - **Precision** (0.76 for disease = 1): 70% of the cases are correctly predicted among all the cases that are predicted as having disease.\n",
    "     - **Recall** (0.71 for disease =1): 72% of the cases are correctly identified by the model among all the cases are diagnoised with disease \n",
    "     - **F-1 score 0.73**: F-1 score is 72%. It takes into consideration of both Precision and Recall.\n",
    "- **ROC curve and AUC score**\n",
    "    - **Result**: The AUC of the best KNN model is around 0.8027.\n",
    "- **cross-validation to help in identifying a case of overfitting**\n",
    "    - **Result:** The average accuracy is 0.7341. \n",
    "- **GMB suggests that 4 features are informative, while the remaining are not. These features and their relative importance are**\n",
    "    1. **High_Blood_Pressure**: 0.673581\n",
    "    2. **Age**: 0.130931\n",
    "    3. **Cholesterol_too high**: 0.048845\n",
    "    4. **Weight**: 0.043179"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 3.3.1 Table Summary of Model Preformance: best NB, KNN, SVM, RF and GMB"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Table Summary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 592,
   "metadata": {
    "scrolled": false
   },
   "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>Gussain</th>\n",
       "      <th>KNN</th>\n",
       "      <th>SVM_Linear</th>\n",
       "      <th>SVM_RBF</th>\n",
       "      <th>RF</th>\n",
       "      <th>GBM</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Metrics</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Accuracy</th>\n",
       "      <td>0.6874</td>\n",
       "      <td>0.7307</td>\n",
       "      <td>0.7307</td>\n",
       "      <td>0.7329</td>\n",
       "      <td>0.7356</td>\n",
       "      <td>0.7369</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Precision</th>\n",
       "      <td>0.7367</td>\n",
       "      <td>0.7592</td>\n",
       "      <td>0.7592</td>\n",
       "      <td>0.7688</td>\n",
       "      <td>0.7717</td>\n",
       "      <td>0.7579</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Recall</th>\n",
       "      <td>0.6034</td>\n",
       "      <td>0.6919</td>\n",
       "      <td>0.6919</td>\n",
       "      <td>0.6819</td>\n",
       "      <td>0.6847</td>\n",
       "      <td>0.712</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>F1-Score</th>\n",
       "      <td>0.6634</td>\n",
       "      <td>0.724</td>\n",
       "      <td>0.724</td>\n",
       "      <td>0.7228</td>\n",
       "      <td>0.7256</td>\n",
       "      <td>0.7343</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CV_mean</th>\n",
       "      <td>0.6885</td>\n",
       "      <td>0.7243</td>\n",
       "      <td>0.7267</td>\n",
       "      <td>0.7311</td>\n",
       "      <td>0.7296</td>\n",
       "      <td>0.7333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CV_std</th>\n",
       "      <td>0.0108</td>\n",
       "      <td>0.0064</td>\n",
       "      <td>0.0058</td>\n",
       "      <td>0.0057</td>\n",
       "      <td>0.006</td>\n",
       "      <td>0.0061</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ROC_AUC</th>\n",
       "      <td>0.6892</td>\n",
       "      <td>0.7248</td>\n",
       "      <td>0.7315</td>\n",
       "      <td>0.734</td>\n",
       "      <td>0.7367</td>\n",
       "      <td>0.7374</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Parameters</th>\n",
       "      <td>None</td>\n",
       "      <td>metric : manhattan,n_neighbors : 29,weights : uniform</td>\n",
       "      <td>penalty: l2,C: 0.25</td>\n",
       "      <td>C: 0.7,gamma: 0.5</td>\n",
       "      <td>criterion : entropy,max_depth : 10,max_features : sqrt,n_estimators : 30</td>\n",
       "      <td>learning_rate: 0.06,loss: deviance,max_depth: 4,n_estimators: 110,subsample: 0.5</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           Gussain                                                    KNN  \\\n",
       "Metrics                                                                     \n",
       "Accuracy    0.6874                                                 0.7307   \n",
       "Precision   0.7367                                                 0.7592   \n",
       "Recall      0.6034                                                 0.6919   \n",
       "F1-Score    0.6634                                                  0.724   \n",
       "CV_mean     0.6885                                                 0.7243   \n",
       "CV_std      0.0108                                                 0.0064   \n",
       "ROC_AUC     0.6892                                                 0.7248   \n",
       "Parameters    None  metric : manhattan,n_neighbors : 29,weights : uniform   \n",
       "\n",
       "                     SVM_Linear            SVM_RBF  \\\n",
       "Metrics                                              \n",
       "Accuracy                 0.7307             0.7329   \n",
       "Precision                0.7592             0.7688   \n",
       "Recall                   0.6919             0.6819   \n",
       "F1-Score                  0.724             0.7228   \n",
       "CV_mean                  0.7267             0.7311   \n",
       "CV_std                   0.0058             0.0057   \n",
       "ROC_AUC                  0.7315              0.734   \n",
       "Parameters  penalty: l2,C: 0.25  C: 0.7,gamma: 0.5   \n",
       "\n",
       "                                                                                  RF  \\\n",
       "Metrics                                                                                \n",
       "Accuracy                                                                      0.7356   \n",
       "Precision                                                                     0.7717   \n",
       "Recall                                                                        0.6847   \n",
       "F1-Score                                                                      0.7256   \n",
       "CV_mean                                                                       0.7296   \n",
       "CV_std                                                                         0.006   \n",
       "ROC_AUC                                                                       0.7367   \n",
       "Parameters  criterion : entropy,max_depth : 10,max_features : sqrt,n_estimators : 30   \n",
       "\n",
       "                                                                                         GBM  \n",
       "Metrics                                                                                       \n",
       "Accuracy                                                                              0.7369  \n",
       "Precision                                                                             0.7579  \n",
       "Recall                                                                                 0.712  \n",
       "F1-Score                                                                              0.7343  \n",
       "CV_mean                                                                               0.7333  \n",
       "CV_std                                                                                0.0061  \n",
       "ROC_AUC                                                                               0.7374  \n",
       "Parameters  learning_rate: 0.06,loss: deviance,max_depth: 4,n_estimators: 110,subsample: 0.5  "
      ]
     },
     "execution_count": 592,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_performance"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Interpretation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- **We have used Confusion Matrix to gain a comprehensive understanding of model performance. It gives direct comparisons of values like True Positives, False Positives, True Negatives and False Negatives.**\n",
    "    - In this disease prediction case, we have a balanced training dataset given that Disease = 1 and Disease = 0 accounts for around 50% of the cases respectively. This means that looking at Accuracy score is a fair way to evaluate model performance. Although GBM has the highest accuract score(0.7369), here all models have very close accuracy scores so we can't really tell them apart. \n",
    "    - Therefore, we need to also look at the recall score given that doctors want to correctly diagnosed people with disease as having disease as much as possible in reality. We see that GBM has the highest Recall (0.712) while the recall of other models are below 0.7. Addtionaly, GBM have the highest F1-score(0.7256). \n",
    "    - Till now, we can carefully conclude that GBM has the best model performance under the confusion matrix evluation method.\n",
    "- **We have used cross-validation to help us in identifying a case of overfitting.**\n",
    "    - Cross validation is to build an estimator against different cross sections of your data to gain an aggregate understanding of performance across all sections. It estimate the model’s predictive performance on unseen data by repeating the experiment multiple times and using all the different parts of the training set as validation sets. This gives a more accurate indication of how well the model generalizes to unseen data.\n",
    "    - All our models seems to have unbasied and low variance performance with a cross validation mean accuracy of around 0.7 and a cross validation standard deviation of around 0.006. \n",
    "    - GBM algorithm has the highest mean accuracy of 0.7333\n",
    "- **We haved used ROC curve and AUC score to compare the usefulness of our models.**\n",
    "    - ROC curves are frequently used to show in a graphical way the connection/trade-off between clinical sensitivity and specificity for every possible cut-off for a test or a combination of tests. As for AUV, AUC is a measure of the usefulness of a test in general, where a greater area means a more useful test, the areas under ROC curves are used to compare the usefulness of tests.\n",
    "    - GBM has the highest ROC_AUC score 0.7374, closely followed by that of RF 0.7367.\n",
    "- **Overall, we can conculde that the GBM model has the higest performance among all algorithms.**"
   ]
  }
 ],
 "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": 4
}