03 Amazon Fine Food Reviews Analysis_KNN.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Amazon Fine Food Reviews Analysis\n",
    "\n",
    "\n",
    "Data Source: https://www.kaggle.com/snap/amazon-fine-food-reviews <br>\n",
    "\n",
    "EDA: https://nycdatascience.com/blog/student-works/amazon-fine-foods-visualization/\n",
    "\n",
    "\n",
    "The Amazon Fine Food Reviews dataset consists of reviews of fine foods from Amazon.<br>\n",
    "\n",
    "Number of reviews: 568,454<br>\n",
    "Number of users: 256,059<br>\n",
    "Number of products: 74,258<br>\n",
    "Timespan: Oct 1999 - Oct 2012<br>\n",
    "Number of Attributes/Columns in data: 10 \n",
    "\n",
    "Attribute Information:\n",
    "\n",
    "1. Id\n",
    "2. ProductId - unique identifier for the product\n",
    "3. UserId - unqiue identifier for the user\n",
    "4. ProfileName\n",
    "5. HelpfulnessNumerator - number of users who found the review helpful\n",
    "6. HelpfulnessDenominator - number of users who indicated whether they found the review helpful or not\n",
    "7. Score - rating between 1 and 5\n",
    "8. Time - timestamp for the review\n",
    "9. Summary - brief summary of the review\n",
    "10. Text - text of the review\n",
    "\n",
    "\n",
    "#### Objective:\n",
    "Given a review, determine whether the review is positive (rating of 4 or 5) or negative (rating of 1 or 2).\n",
    "\n",
    "<br>\n",
    "[Q] How to determine if a review is positive or negative?<br>\n",
    "<br> \n",
    "[Ans] We could use Score/Rating. A rating of 4 or 5 can be cosnidered as a positive review. A rating of 1 or 2 can be considered as negative one. A review of rating 3 is considered nuetral and such reviews are ignored from our analysis. This is an approximate and proxy way of determining the polarity (positivity/negativity) of a review.\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# [1]. Reading Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## [1.1] Loading the data\n",
    "\n",
    "The dataset is available in two forms\n",
    "1. .csv file\n",
    "2. SQLite Database\n",
    "\n",
    "In order to load the data, We have used the SQLITE dataset as it is easier to query the data and visualise the data efficiently.\n",
    "<br> \n",
    "\n",
    "Here as we only want to get the global sentiment of the recommendations (positive or negative), we will purposefully ignore all Scores equal to 3. If the score is above 3, then the recommendation wil be set to \"positive\". Otherwise, it will be set to \"negative\"."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "\n",
    "\n",
    "import sqlite3\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import nltk\n",
    "import string\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from sklearn.feature_extraction.text import TfidfTransformer\n",
    "from sklearn.feature_extraction.text import TfidfVectorizer\n",
    "\n",
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn import metrics\n",
    "from sklearn.metrics import roc_curve, auc\n",
    "from nltk.stem.porter import PorterStemmer\n",
    "\n",
    "import re\n",
    "# Tutorial about Python regular expressions: https://pymotw.com/2/re/\n",
    "import string\n",
    "from nltk.corpus import stopwords\n",
    "from nltk.stem import PorterStemmer\n",
    "from nltk.stem.wordnet import WordNetLemmatizer\n",
    "\n",
    "from gensim.models import Word2Vec\n",
    "from gensim.models import KeyedVectors\n",
    "import pickle\n",
    "\n",
    "from tqdm import tqdm\n",
    "import os"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of data points in our data (50000, 10)\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>Id</th>\n",
       "      <th>ProductId</th>\n",
       "      <th>UserId</th>\n",
       "      <th>ProfileName</th>\n",
       "      <th>HelpfulnessNumerator</th>\n",
       "      <th>HelpfulnessDenominator</th>\n",
       "      <th>Score</th>\n",
       "      <th>Time</th>\n",
       "      <th>Summary</th>\n",
       "      <th>Text</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>B001E4KFG0</td>\n",
       "      <td>A3SGXH7AUHU8GW</td>\n",
       "      <td>delmartian</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1303862400</td>\n",
       "      <td>Good Quality Dog Food</td>\n",
       "      <td>I have bought several of the Vitality canned d...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>B00813GRG4</td>\n",
       "      <td>A1D87F6ZCVE5NK</td>\n",
       "      <td>dll pa</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1346976000</td>\n",
       "      <td>Not as Advertised</td>\n",
       "      <td>Product arrived labeled as Jumbo Salted Peanut...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>B000LQOCH0</td>\n",
       "      <td>ABXLMWJIXXAIN</td>\n",
       "      <td>Natalia Corres \"Natalia Corres\"</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1219017600</td>\n",
       "      <td>\"Delight\" says it all</td>\n",
       "      <td>This is a confection that has been around a fe...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Id   ProductId          UserId                      ProfileName  \\\n",
       "0   1  B001E4KFG0  A3SGXH7AUHU8GW                       delmartian   \n",
       "1   2  B00813GRG4  A1D87F6ZCVE5NK                           dll pa   \n",
       "2   3  B000LQOCH0   ABXLMWJIXXAIN  Natalia Corres \"Natalia Corres\"   \n",
       "\n",
       "   HelpfulnessNumerator  HelpfulnessDenominator  Score        Time  \\\n",
       "0                     1                       1      1  1303862400   \n",
       "1                     0                       0      0  1346976000   \n",
       "2                     1                       1      1  1219017600   \n",
       "\n",
       "                 Summary                                               Text  \n",
       "0  Good Quality Dog Food  I have bought several of the Vitality canned d...  \n",
       "1      Not as Advertised  Product arrived labeled as Jumbo Salted Peanut...  \n",
       "2  \"Delight\" says it all  This is a confection that has been around a fe...  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# using SQLite Table to read data.\n",
    "con = sqlite3.connect('database.sqlite') \n",
    "\n",
    "# filtering only positive and negative reviews i.e. \n",
    "# not taking into consideration those reviews with Score=3\n",
    "# SELECT * FROM Reviews WHERE Score != 3 LIMIT 500000, will give top 500000 data points\n",
    "# you can change the number to any other number based on your computing power\n",
    "\n",
    "# filtered_data = pd.read_sql_query(\"\"\" SELECT * FROM Reviews WHERE Score != 3 LIMIT 500000\"\"\", con) \n",
    "# for tsne assignment you can take 5k data points\n",
    "\n",
    "filtered_data = pd.read_sql_query(\"\"\" SELECT * FROM Reviews WHERE Score != 3 LIMIT 50000\"\"\", con) \n",
    "\n",
    "# Give reviews with Score>3 a positive rating(1), and reviews with a score<3 a negative rating(0).\n",
    "def partition(x):\n",
    "    if x < 3:\n",
    "        return 0\n",
    "    return 1\n",
    "\n",
    "#changing reviews with score less than 3 to be positive and vice-versa\n",
    "actualScore = filtered_data['Score']\n",
    "positiveNegative = actualScore.map(partition) \n",
    "filtered_data['Score'] = positiveNegative\n",
    "print(\"Number of data points in our data\", filtered_data.shape)\n",
    "filtered_data.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "display = pd.read_sql_query(\"\"\"\n",
    "SELECT UserId, ProductId, ProfileName, Time, Score, Text, COUNT(*)\n",
    "FROM Reviews\n",
    "GROUP BY UserId\n",
    "HAVING COUNT(*)>1\n",
    "\"\"\", con)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(80668, 7)\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>UserId</th>\n",
       "      <th>ProductId</th>\n",
       "      <th>ProfileName</th>\n",
       "      <th>Time</th>\n",
       "      <th>Score</th>\n",
       "      <th>Text</th>\n",
       "      <th>COUNT(*)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>#oc-R115TNMSPFT9I7</td>\n",
       "      <td>B005ZBZLT4</td>\n",
       "      <td>Breyton</td>\n",
       "      <td>1331510400</td>\n",
       "      <td>2</td>\n",
       "      <td>Overall its just OK when considering the price...</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>#oc-R11D9D7SHXIJB9</td>\n",
       "      <td>B005HG9ESG</td>\n",
       "      <td>Louis E. Emory \"hoppy\"</td>\n",
       "      <td>1342396800</td>\n",
       "      <td>5</td>\n",
       "      <td>My wife has recurring extreme muscle spasms, u...</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>#oc-R11DNU2NBKQ23Z</td>\n",
       "      <td>B005ZBZLT4</td>\n",
       "      <td>Kim Cieszykowski</td>\n",
       "      <td>1348531200</td>\n",
       "      <td>1</td>\n",
       "      <td>This coffee is horrible and unfortunately not ...</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>#oc-R11O5J5ZVQE25C</td>\n",
       "      <td>B005HG9ESG</td>\n",
       "      <td>Penguin Chick</td>\n",
       "      <td>1346889600</td>\n",
       "      <td>5</td>\n",
       "      <td>This will be the bottle that you grab from the...</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>#oc-R12KPBODL2B5ZD</td>\n",
       "      <td>B007OSBEV0</td>\n",
       "      <td>Christopher P. Presta</td>\n",
       "      <td>1348617600</td>\n",
       "      <td>1</td>\n",
       "      <td>I didnt like this coffee. Instead of telling y...</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               UserId   ProductId             ProfileName        Time  Score  \\\n",
       "0  #oc-R115TNMSPFT9I7  B005ZBZLT4                 Breyton  1331510400      2   \n",
       "1  #oc-R11D9D7SHXIJB9  B005HG9ESG  Louis E. Emory \"hoppy\"  1342396800      5   \n",
       "2  #oc-R11DNU2NBKQ23Z  B005ZBZLT4        Kim Cieszykowski  1348531200      1   \n",
       "3  #oc-R11O5J5ZVQE25C  B005HG9ESG           Penguin Chick  1346889600      5   \n",
       "4  #oc-R12KPBODL2B5ZD  B007OSBEV0   Christopher P. Presta  1348617600      1   \n",
       "\n",
       "                                                Text  COUNT(*)  \n",
       "0  Overall its just OK when considering the price...         2  \n",
       "1  My wife has recurring extreme muscle spasms, u...         3  \n",
       "2  This coffee is horrible and unfortunately not ...         2  \n",
       "3  This will be the bottle that you grab from the...         3  \n",
       "4  I didnt like this coffee. Instead of telling y...         2  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(display.shape)\n",
    "display.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "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>UserId</th>\n",
       "      <th>ProductId</th>\n",
       "      <th>ProfileName</th>\n",
       "      <th>Time</th>\n",
       "      <th>Score</th>\n",
       "      <th>Text</th>\n",
       "      <th>COUNT(*)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>80638</th>\n",
       "      <td>AZY10LLTJ71NX</td>\n",
       "      <td>B001ATMQK2</td>\n",
       "      <td>undertheshrine \"undertheshrine\"</td>\n",
       "      <td>1296691200</td>\n",
       "      <td>5</td>\n",
       "      <td>I bought this 6 pack because for the price tha...</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              UserId   ProductId                      ProfileName        Time  \\\n",
       "80638  AZY10LLTJ71NX  B001ATMQK2  undertheshrine \"undertheshrine\"  1296691200   \n",
       "\n",
       "       Score                                               Text  COUNT(*)  \n",
       "80638      5  I bought this 6 pack because for the price tha...         5  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "display[display['UserId']=='AZY10LLTJ71NX']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "393063"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "display['COUNT(*)'].sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#  [2] Exploratory Data Analysis"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## [2.1] Data Cleaning: Deduplication\n",
    "\n",
    "It is observed (as shown in the table below) that the reviews data had many duplicate entries. Hence it was necessary to remove duplicates in order to get unbiased results for the analysis of the data.  Following is an example:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "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>Id</th>\n",
       "      <th>ProductId</th>\n",
       "      <th>UserId</th>\n",
       "      <th>ProfileName</th>\n",
       "      <th>HelpfulnessNumerator</th>\n",
       "      <th>HelpfulnessDenominator</th>\n",
       "      <th>Score</th>\n",
       "      <th>Time</th>\n",
       "      <th>Summary</th>\n",
       "      <th>Text</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>78445</td>\n",
       "      <td>B000HDL1RQ</td>\n",
       "      <td>AR5J8UI46CURR</td>\n",
       "      <td>Geetha Krishnan</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>1199577600</td>\n",
       "      <td>LOACKER QUADRATINI VANILLA WAFERS</td>\n",
       "      <td>DELICIOUS WAFERS. I FIND THAT EUROPEAN WAFERS ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>138317</td>\n",
       "      <td>B000HDOPYC</td>\n",
       "      <td>AR5J8UI46CURR</td>\n",
       "      <td>Geetha Krishnan</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>1199577600</td>\n",
       "      <td>LOACKER QUADRATINI VANILLA WAFERS</td>\n",
       "      <td>DELICIOUS WAFERS. I FIND THAT EUROPEAN WAFERS ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>138277</td>\n",
       "      <td>B000HDOPYM</td>\n",
       "      <td>AR5J8UI46CURR</td>\n",
       "      <td>Geetha Krishnan</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>1199577600</td>\n",
       "      <td>LOACKER QUADRATINI VANILLA WAFERS</td>\n",
       "      <td>DELICIOUS WAFERS. I FIND THAT EUROPEAN WAFERS ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>73791</td>\n",
       "      <td>B000HDOPZG</td>\n",
       "      <td>AR5J8UI46CURR</td>\n",
       "      <td>Geetha Krishnan</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>1199577600</td>\n",
       "      <td>LOACKER QUADRATINI VANILLA WAFERS</td>\n",
       "      <td>DELICIOUS WAFERS. I FIND THAT EUROPEAN WAFERS ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>155049</td>\n",
       "      <td>B000PAQ75C</td>\n",
       "      <td>AR5J8UI46CURR</td>\n",
       "      <td>Geetha Krishnan</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>1199577600</td>\n",
       "      <td>LOACKER QUADRATINI VANILLA WAFERS</td>\n",
       "      <td>DELICIOUS WAFERS. I FIND THAT EUROPEAN WAFERS ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Id   ProductId         UserId      ProfileName  HelpfulnessNumerator  \\\n",
       "0   78445  B000HDL1RQ  AR5J8UI46CURR  Geetha Krishnan                     2   \n",
       "1  138317  B000HDOPYC  AR5J8UI46CURR  Geetha Krishnan                     2   \n",
       "2  138277  B000HDOPYM  AR5J8UI46CURR  Geetha Krishnan                     2   \n",
       "3   73791  B000HDOPZG  AR5J8UI46CURR  Geetha Krishnan                     2   \n",
       "4  155049  B000PAQ75C  AR5J8UI46CURR  Geetha Krishnan                     2   \n",
       "\n",
       "   HelpfulnessDenominator  Score        Time  \\\n",
       "0                       2      5  1199577600   \n",
       "1                       2      5  1199577600   \n",
       "2                       2      5  1199577600   \n",
       "3                       2      5  1199577600   \n",
       "4                       2      5  1199577600   \n",
       "\n",
       "                             Summary  \\\n",
       "0  LOACKER QUADRATINI VANILLA WAFERS   \n",
       "1  LOACKER QUADRATINI VANILLA WAFERS   \n",
       "2  LOACKER QUADRATINI VANILLA WAFERS   \n",
       "3  LOACKER QUADRATINI VANILLA WAFERS   \n",
       "4  LOACKER QUADRATINI VANILLA WAFERS   \n",
       "\n",
       "                                                Text  \n",
       "0  DELICIOUS WAFERS. I FIND THAT EUROPEAN WAFERS ...  \n",
       "1  DELICIOUS WAFERS. I FIND THAT EUROPEAN WAFERS ...  \n",
       "2  DELICIOUS WAFERS. I FIND THAT EUROPEAN WAFERS ...  \n",
       "3  DELICIOUS WAFERS. I FIND THAT EUROPEAN WAFERS ...  \n",
       "4  DELICIOUS WAFERS. I FIND THAT EUROPEAN WAFERS ...  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "display= pd.read_sql_query(\"\"\"\n",
    "SELECT *\n",
    "FROM Reviews\n",
    "WHERE Score != 3 AND UserId=\"AR5J8UI46CURR\"\n",
    "ORDER BY ProductID\n",
    "\"\"\", con)\n",
    "display.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As it can be seen above that same user has multiple reviews with same values for HelpfulnessNumerator, HelpfulnessDenominator, Score, Time, Summary and Text and on doing analysis it was found that <br>\n",
    "<br> \n",
    "ProductId=B000HDOPZG was Loacker Quadratini Vanilla Wafer Cookies, 8.82-Ounce Packages (Pack of 8)<br>\n",
    "<br> \n",
    "ProductId=B000HDL1RQ was Loacker Quadratini Lemon Wafer Cookies, 8.82-Ounce Packages (Pack of 8) and so on<br>\n",
    "\n",
    "It was inferred after analysis that reviews with same parameters other than ProductId belonged to the same product just having different flavour or quantity. Hence in order to reduce redundancy it was decided to eliminate the rows having same parameters.<br>\n",
    "\n",
    "The method used for the same was that we first sort the data according to ProductId and then just keep the first similar product review and delelte the others. for eg. in the above just the review for ProductId=B000HDL1RQ remains. This method ensures that there is only one representative for each product and deduplication without sorting would lead to possibility of different representatives still existing for the same product."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Sorting data according to ProductId in ascending order\n",
    "sorted_data=filtered_data.sort_values('ProductId', axis=0, ascending=True, inplace=False, kind='quicksort', na_position='last')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(46072, 10)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Deduplication of entries\n",
    "final=sorted_data.drop_duplicates(subset={\"UserId\",\"ProfileName\",\"Time\",\"Text\"}, keep='first', inplace=False)\n",
    "final.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "92.144"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Checking to see how much % of data still remains\n",
    "(final['Id'].size*1.0)/(filtered_data['Id'].size*1.0)*100"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<b>Observation:-</b> It was also seen that in two rows given below the value of HelpfulnessNumerator is greater than HelpfulnessDenominator which is not practically possible hence these two rows too are removed from calcualtions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "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>ProductId</th>\n",
       "      <th>UserId</th>\n",
       "      <th>ProfileName</th>\n",
       "      <th>HelpfulnessNumerator</th>\n",
       "      <th>HelpfulnessDenominator</th>\n",
       "      <th>Score</th>\n",
       "      <th>Time</th>\n",
       "      <th>Summary</th>\n",
       "      <th>Text</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>64422</td>\n",
       "      <td>B000MIDROQ</td>\n",
       "      <td>A161DK06JJMCYF</td>\n",
       "      <td>J. E. Stephens \"Jeanne\"</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>1224892800</td>\n",
       "      <td>Bought This for My Son at College</td>\n",
       "      <td>My son loves spaghetti so I didn't hesitate or...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>44737</td>\n",
       "      <td>B001EQ55RW</td>\n",
       "      <td>A2V0I904FH7ABY</td>\n",
       "      <td>Ram</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1212883200</td>\n",
       "      <td>Pure cocoa taste with crunchy almonds inside</td>\n",
       "      <td>It was almost a 'love at first bite' - the per...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      Id   ProductId          UserId              ProfileName  \\\n",
       "0  64422  B000MIDROQ  A161DK06JJMCYF  J. E. Stephens \"Jeanne\"   \n",
       "1  44737  B001EQ55RW  A2V0I904FH7ABY                      Ram   \n",
       "\n",
       "   HelpfulnessNumerator  HelpfulnessDenominator  Score        Time  \\\n",
       "0                     3                       1      5  1224892800   \n",
       "1                     3                       2      4  1212883200   \n",
       "\n",
       "                                        Summary  \\\n",
       "0             Bought This for My Son at College   \n",
       "1  Pure cocoa taste with crunchy almonds inside   \n",
       "\n",
       "                                                Text  \n",
       "0  My son loves spaghetti so I didn't hesitate or...  \n",
       "1  It was almost a 'love at first bite' - the per...  "
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "display= pd.read_sql_query(\"\"\"\n",
    "SELECT *\n",
    "FROM Reviews\n",
    "WHERE Score != 3 AND Id=44737 OR Id=64422\n",
    "ORDER BY ProductID\n",
    "\"\"\", con)\n",
    "\n",
    "display.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "final=final[final.HelpfulnessNumerator<=final.HelpfulnessDenominator]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(46071, 10)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "1    38479\n",
       "0     7592\n",
       "Name: Score, dtype: int64"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Before starting the next phase of preprocessing lets see the number of entries left\n",
    "print(final.shape)\n",
    "\n",
    "#How many positive and negative reviews are present in our dataset?\n",
    "final['Score'].value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#  [3] Preprocessing"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## [3.1].  Preprocessing Review Text\n",
    "\n",
    "Now that we have finished deduplication our data requires some preprocessing before we go on further with analysis and making the prediction model.\n",
    "\n",
    "Hence in the Preprocessing phase we do the following in the order below:-\n",
    "\n",
    "1. Begin by removing the html tags\n",
    "2. Remove any punctuations or limited set of special characters like , or . or # etc.\n",
    "3. Check if the word is made up of english letters and is not alpha-numeric\n",
    "4. Check to see if the length of the word is greater than 2 (as it was researched that there is no adjective in 2-letters)\n",
    "5. Convert the word to lowercase\n",
    "6. Remove Stopwords\n",
    "7. Finally Snowball Stemming the word (it was obsereved to be better than Porter Stemming)<br>\n",
    "\n",
    "After which we collect the words used to describe positive and negative reviews"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "My dogs loves this chicken but its a product from China, so we wont be buying it anymore.  Its very hard to find any chicken products made in the USA but they are out there, but this one isnt.  Its too bad too because its a good product but I wont take any chances till they know what is going on with the china imports.\n",
      "==================================================\n",
      "this is yummy, easy and unusual. it makes a quick, delicous pie, crisp or cobbler. home made is better, but a heck of a lot more work. this is great to have on hand for last minute dessert needs where you really want to impress wih your creativity in cooking! recommended.\n",
      "==================================================\n",
      "Great flavor, low in calories, high in nutrients, high in protein! Usually protein powders are high priced and high in calories, this one is a great bargain and tastes great, I highly recommend for the lady gym rats, probably not \"macho\" enough for guys since it is soy based...\n",
      "==================================================\n",
      "For those of you wanting a high-quality, yet affordable green tea, you should definitely give this one a try. Let me first start by saying that everyone is looking for something different for their ideal tea, and I will attempt to briefly highlight what makes this tea attractive to a wide range of tea drinkers (whether you are a beginner or long-time tea enthusiast).  I have gone through over 12 boxes of this tea myself, and highly recommend it for the following reasons:<br /><br />-Quality:  First, this tea offers a smooth quality without any harsh or bitter after tones, which often turns people off from many green teas.  I've found my ideal brewing time to be between 3-5 minutes, giving you a light but flavorful cup of tea.  However, if you get distracted or forget about your tea and leave it brewing for 20+ minutes like I sometimes do, the quality of this tea is such that you still get a smooth but deeper flavor without the bad after taste.  The leaves themselves are whole leaves (not powdered stems, branches, etc commonly found in other brands), and the high-quality nylon bags also include chunks of tropical fruit and other discernible ingredients.  This isn't your standard cheap paper bag with a mix of unknown ingredients that have been ground down to a fine powder, leaving you to wonder what it is you are actually drinking.<br /><br />-Taste:  This tea offers notes of real pineapple and other hints of tropical fruits, yet isn't sweet or artificially flavored.  You have the foundation of a high-quality young hyson green tea for those true \"tea flavor\" lovers, yet the subtle hints of fruit make this a truly unique tea that I believe most will enjoy.  If you want it sweet, you can add sugar, splenda, etc but this really is not necessary as this tea offers an inherent warmth of flavor through it's ingredients.<br /><br />-Price:  This tea offers an excellent product at an exceptional price (especially when purchased at the prices Amazon offers).  Compared to other brands which I believe to be of similar quality (Mighty Leaf, Rishi, Two Leaves, etc.), Revolution offers a superior product at an outstanding price.  I have been purchasing this through Amazon for less per box than I would be paying at my local grocery store for Lipton, etc.<br /><br />Overall, this is a wonderful tea that is comparable, and even better than, other teas that are priced much higher.  It offers a well-balanced cup of green tea that I believe many will enjoy.  In terms of taste, quality, and price, I would argue you won't find a better combination that that offered by Revolution's Tropical Green Tea.\n",
      "==================================================\n"
     ]
    }
   ],
   "source": [
    "# printing some random reviews\n",
    "sent_0 = final['Text'].values[0]\n",
    "print(sent_0)\n",
    "print(\"=\"*50)\n",
    "\n",
    "sent_1000 = final['Text'].values[1000]\n",
    "print(sent_1000)\n",
    "print(\"=\"*50)\n",
    "\n",
    "sent_1500 = final['Text'].values[1500]\n",
    "print(sent_1500)\n",
    "print(\"=\"*50)\n",
    "\n",
    "sent_4900 = final['Text'].values[4900]\n",
    "print(sent_4900)\n",
    "print(\"=\"*50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "My dogs loves this chicken but its a product from China, so we wont be buying it anymore.  Its very hard to find any chicken products made in the USA but they are out there, but this one isnt.  Its too bad too because its a good product but I wont take any chances till they know what is going on with the china imports.\n"
     ]
    }
   ],
   "source": [
    "# remove urls from text python: https://stackoverflow.com/a/40823105/4084039\n",
    "sent_0 = re.sub(r\"http\\S+\", \"\", sent_0)\n",
    "sent_1000 = re.sub(r\"http\\S+\", \"\", sent_1000)\n",
    "sent_150 = re.sub(r\"http\\S+\", \"\", sent_1500)\n",
    "sent_4900 = re.sub(r\"http\\S+\", \"\", sent_4900)\n",
    "\n",
    "print(sent_0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "My dogs loves this chicken but its a product from China, so we wont be buying it anymore.  Its very hard to find any chicken products made in the USA but they are out there, but this one isnt.  Its too bad too because its a good product but I wont take any chances till they know what is going on with the china imports.\n",
      "==================================================\n",
      "this is yummy, easy and unusual. it makes a quick, delicous pie, crisp or cobbler. home made is better, but a heck of a lot more work. this is great to have on hand for last minute dessert needs where you really want to impress wih your creativity in cooking! recommended.\n",
      "==================================================\n",
      "Great flavor, low in calories, high in nutrients, high in protein! Usually protein powders are high priced and high in calories, this one is a great bargain and tastes great, I highly recommend for the lady gym rats, probably not \"macho\" enough for guys since it is soy based...\n",
      "==================================================\n",
      "For those of you wanting a high-quality, yet affordable green tea, you should definitely give this one a try. Let me first start by saying that everyone is looking for something different for their ideal tea, and I will attempt to briefly highlight what makes this tea attractive to a wide range of tea drinkers (whether you are a beginner or long-time tea enthusiast).  I have gone through over 12 boxes of this tea myself, and highly recommend it for the following reasons:-Quality:  First, this tea offers a smooth quality without any harsh or bitter after tones, which often turns people off from many green teas.  I've found my ideal brewing time to be between 3-5 minutes, giving you a light but flavorful cup of tea.  However, if you get distracted or forget about your tea and leave it brewing for 20+ minutes like I sometimes do, the quality of this tea is such that you still get a smooth but deeper flavor without the bad after taste.  The leaves themselves are whole leaves (not powdered stems, branches, etc commonly found in other brands), and the high-quality nylon bags also include chunks of tropical fruit and other discernible ingredients.  This isn't your standard cheap paper bag with a mix of unknown ingredients that have been ground down to a fine powder, leaving you to wonder what it is you are actually drinking.-Taste:  This tea offers notes of real pineapple and other hints of tropical fruits, yet isn't sweet or artificially flavored.  You have the foundation of a high-quality young hyson green tea for those true \"tea flavor\" lovers, yet the subtle hints of fruit make this a truly unique tea that I believe most will enjoy.  If you want it sweet, you can add sugar, splenda, etc but this really is not necessary as this tea offers an inherent warmth of flavor through it's ingredients.-Price:  This tea offers an excellent product at an exceptional price (especially when purchased at the prices Amazon offers).  Compared to other brands which I believe to be of similar quality (Mighty Leaf, Rishi, Two Leaves, etc.), Revolution offers a superior product at an outstanding price.  I have been purchasing this through Amazon for less per box than I would be paying at my local grocery store for Lipton, etc.Overall, this is a wonderful tea that is comparable, and even better than, other teas that are priced much higher.  It offers a well-balanced cup of green tea that I believe many will enjoy.  In terms of taste, quality, and price, I would argue you won't find a better combination that that offered by Revolution's Tropical Green Tea.\n"
     ]
    }
   ],
   "source": [
    "# https://stackoverflow.com/questions/16206380/python-beautifulsoup-how-to-remove-all-tags-from-an-element\n",
    "from bs4 import BeautifulSoup\n",
    "\n",
    "soup = BeautifulSoup(sent_0, 'lxml')\n",
    "text = soup.get_text()\n",
    "print(text)\n",
    "print(\"=\"*50)\n",
    "\n",
    "soup = BeautifulSoup(sent_1000, 'lxml')\n",
    "text = soup.get_text()\n",
    "print(text)\n",
    "print(\"=\"*50)\n",
    "\n",
    "soup = BeautifulSoup(sent_1500, 'lxml')\n",
    "text = soup.get_text()\n",
    "print(text)\n",
    "print(\"=\"*50)\n",
    "\n",
    "soup = BeautifulSoup(sent_4900, 'lxml')\n",
    "text = soup.get_text()\n",
    "print(text)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "# https://stackoverflow.com/a/47091490/4084039\n",
    "import re\n",
    "\n",
    "def decontracted(phrase):\n",
    "    # specific\n",
    "    phrase = re.sub(r\"won't\", \"will not\", phrase)\n",
    "    phrase = re.sub(r\"can\\'t\", \"can not\", phrase)\n",
    "\n",
    "    # general\n",
    "    phrase = re.sub(r\"n\\'t\", \" not\", phrase)\n",
    "    phrase = re.sub(r\"\\'re\", \" are\", phrase)\n",
    "    phrase = re.sub(r\"\\'s\", \" is\", phrase)\n",
    "    phrase = re.sub(r\"\\'d\", \" would\", phrase)\n",
    "    phrase = re.sub(r\"\\'ll\", \" will\", phrase)\n",
    "    phrase = re.sub(r\"\\'t\", \" not\", phrase)\n",
    "    phrase = re.sub(r\"\\'ve\", \" have\", phrase)\n",
    "    phrase = re.sub(r\"\\'m\", \" am\", phrase)\n",
    "    return phrase"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Great flavor, low in calories, high in nutrients, high in protein! Usually protein powders are high priced and high in calories, this one is a great bargain and tastes great, I highly recommend for the lady gym rats, probably not \"macho\" enough for guys since it is soy based...\n",
      "==================================================\n"
     ]
    }
   ],
   "source": [
    "sent_1500 = decontracted(sent_1500)\n",
    "print(sent_1500)\n",
    "print(\"=\"*50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "My dogs loves this chicken but its a product from China, so we wont be buying it anymore.  Its very hard to find any chicken products made in the USA but they are out there, but this one isnt.  Its too bad too because its a good product but I wont take any chances till they know what is going on with the china imports.\n"
     ]
    }
   ],
   "source": [
    "#remove words with numbers python: https://stackoverflow.com/a/18082370/4084039\n",
    "sent_0 = re.sub(\"\\S*\\d\\S*\", \"\", sent_0).strip()\n",
    "print(sent_0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Great flavor low in calories high in nutrients high in protein Usually protein powders are high priced and high in calories this one is a great bargain and tastes great I highly recommend for the lady gym rats probably not macho enough for guys since it is soy based \n"
     ]
    }
   ],
   "source": [
    "#remove spacial character: https://stackoverflow.com/a/5843547/4084039\n",
    "sent_1500 = re.sub('[^A-Za-z0-9]+', ' ', sent_1500)\n",
    "print(sent_1500)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "# https://gist.github.com/sebleier/554280\n",
    "# we are removing the words from the stop words list: 'no', 'nor', 'not'\n",
    "# <br /><br /> ==> after the above steps, we are getting \"br br\"\n",
    "# we are including them into stop words list\n",
    "# instead of <br /> if we have <br/> these tags would have revmoved in the 1st step\n",
    "\n",
    "stopwords= set(['br', 'the', 'i', 'me', 'my', 'myself', 'we', 'our', 'ours', 'ourselves', 'you', \"you're\", \"you've\",\\\n",
    "            \"you'll\", \"you'd\", 'your', 'yours', 'yourself', 'yourselves', 'he', 'him', 'his', 'himself', \\\n",
    "            'she', \"she's\", 'her', 'hers', 'herself', 'it', \"it's\", 'its', 'itself', 'they', 'them', 'their',\\\n",
    "            'theirs', 'themselves', 'what', 'which', 'who', 'whom', 'this', 'that', \"that'll\", 'these', 'those', \\\n",
    "            'am', 'is', 'are', 'was', 'were', 'be', 'been', 'being', 'have', 'has', 'had', 'having', 'do', 'does', \\\n",
    "            'did', 'doing', 'a', 'an', 'the', 'and', 'but', 'if', 'or', 'because', 'as', 'until', 'while', 'of', \\\n",
    "            'at', 'by', 'for', 'with', 'about', 'against', 'between', 'into', 'through', 'during', 'before', 'after',\\\n",
    "            'above', 'below', 'to', 'from', 'up', 'down', 'in', 'out', 'on', 'off', 'over', 'under', 'again', 'further',\\\n",
    "            'then', 'once', 'here', 'there', 'when', 'where', 'why', 'how', 'all', 'any', 'both', 'each', 'few', 'more',\\\n",
    "            'most', 'other', 'some', 'such', 'only', 'own', 'same', 'so', 'than', 'too', 'very', \\\n",
    "            's', 't', 'can', 'will', 'just', 'don', \"don't\", 'should', \"should've\", 'now', 'd', 'll', 'm', 'o', 're', \\\n",
    "            've', 'y', 'ain', 'aren', \"aren't\", 'couldn', \"couldn't\", 'didn', \"didn't\", 'doesn', \"doesn't\", 'hadn',\\\n",
    "            \"hadn't\", 'hasn', \"hasn't\", 'haven', \"haven't\", 'isn', \"isn't\", 'ma', 'mightn', \"mightn't\", 'mustn',\\\n",
    "            \"mustn't\", 'needn', \"needn't\", 'shan', \"shan't\", 'shouldn', \"shouldn't\", 'wasn', \"wasn't\", 'weren', \"weren't\", \\\n",
    "            'won', \"won't\", 'wouldn', \"wouldn't\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████████████████████████████████████████| 46071/46071 [00:16<00:00, 2718.78it/s]\n"
     ]
    }
   ],
   "source": [
    "# Combining all the above stundents \n",
    "from tqdm import tqdm\n",
    "preprocessed_reviews = []\n",
    "# tqdm is for printing the status bar\n",
    "for sentance in tqdm(final['Text'].values):\n",
    "    sentance = re.sub(r\"http\\S+\", \"\", sentance)\n",
    "    sentance = BeautifulSoup(sentance, 'lxml').get_text()\n",
    "    sentance = decontracted(sentance)\n",
    "    sentance = re.sub(\"\\S*\\d\\S*\", \"\", sentance).strip()\n",
    "    sentance = re.sub('[^A-Za-z]+', ' ', sentance)\n",
    "    # https://gist.github.com/sebleier/554280\n",
    "    sentance = ' '.join(e.lower() for e in sentance.split() if e.lower() not in stopwords)\n",
    "    preprocessed_reviews.append(sentance.strip())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'great flavor low calories high nutrients high protein usually protein powders high priced high calories one great bargain tastes great highly recommend lady gym rats probably not macho enough guys since soy based'"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preprocessed_reviews[1500]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2><font color='red'>[3.2] Preprocessing Review Summary</font></h2>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Similartly you can do preprocessing for review summary also."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Importing all relevant libraries\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.metrics import roc_auc_score\n",
    "from sklearn.metrics import confusion_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Data Splitting\n",
    "# Train Data :  60%\n",
    "# CV Data    :  20%\n",
    "# Test Data  :  20%\n",
    "\n",
    "X=preprocessed_reviews\n",
    "y=np.array(final['Score'])\n",
    "\n",
    "#####  SPLIT  #######\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=4)\n",
    "X_train, X_cv, y_train, y_cv = train_test_split(X_train, y_train, test_size=0.25, random_state=4)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# [4] Featurization"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## [4.1] BAG OF WORDS"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "some feature names  ['aa', 'aaa', 'aaaa', 'aaaaa', 'aaaaaaaaaaaaaaa', 'aaaaaaahhhhhh', 'aaaaaawwwwwwwwww', 'aaaaah', 'aaaand', 'aaah']\n",
      "==================================================\n",
      "the type of count vectorizer  <class 'scipy.sparse.csr.csr_matrix'>\n",
      "the shape of out text BOW vectorizer  (27642, 31346)\n",
      "the number of unique words  31346\n"
     ]
    }
   ],
   "source": [
    "#BoW\n",
    "X_train_bow, X_test_bow, X_cv_bow = X_train, X_test, X_cv\n",
    "y_train_bow, y_test_bow, y_cv_bow = y_train, y_test, y_cv\n",
    "\n",
    "count_vect = CountVectorizer() #in scikit-learn\n",
    "count_vect.fit(X_train_bow)\n",
    "print(\"some feature names \", count_vect.get_feature_names()[:10])\n",
    "print('='*50)\n",
    "\n",
    "bow_counts_train = count_vect.transform(X_train_bow)\n",
    "bow_counts_cv = count_vect.transform(X_cv_bow)\n",
    "bow_counts_test = count_vect.transform(X_test_bow)\n",
    "print(\"the type of count vectorizer \",type(bow_counts_train))\n",
    "print(\"the shape of out text BOW vectorizer \",bow_counts_train.get_shape())\n",
    "print(\"the number of unique words \", bow_counts_train.get_shape()[1])\n",
    "# final_counts_test=final_counts_test.toarray()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## [4.2] Bi-Grams and n-Grams."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "the type of count vectorizer  <class 'scipy.sparse.csr.csr_matrix'>\n",
      "the shape of out text BOW vectorizer  (46071, 5000)\n",
      "the number of unique words including both unigrams and bigrams  5000\n"
     ]
    }
   ],
   "source": [
    "#bi-gram, tri-gram and n-gram\n",
    "\n",
    "#removing stop words like \"not\" should be avoided before building n-grams\n",
    "# count_vect = CountVectorizer(ngram_range=(1,2))\n",
    "# please do read the CountVectorizer documentation http://scikit-learn.org/stable/modules/generated/sklearn.feature_extraction.text.CountVectorizer.html\n",
    "\n",
    "# you can choose these numebrs min_df=10, max_features=5000, of your choice\n",
    "count_vect = CountVectorizer(ngram_range=(1,2), min_df=10, max_features=5000)\n",
    "final_bigram_counts = count_vect.fit_transform(preprocessed_reviews)\n",
    "print(\"the type of count vectorizer \",type(final_bigram_counts))\n",
    "print(\"the shape of out text BOW vectorizer \",final_bigram_counts.get_shape())\n",
    "print(\"the number of unique words including both unigrams and bigrams \", final_bigram_counts.get_shape()[1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## [4.3] TF-IDF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "some sample features(unique words in the corpus) ['ability', 'able', 'able buy', 'able drink', 'able eat', 'able enjoy', 'able find', 'able get', 'able make', 'able order']\n",
      "==================================================\n",
      "the type of count vectorizer  <class 'scipy.sparse.csr.csr_matrix'>\n",
      "the shape of out text TFIDF vectorizer  (27642, 16605)\n",
      "the number of unique words including both unigrams and bigrams  16605\n"
     ]
    }
   ],
   "source": [
    "X_train_tf, X_test_tf, X_cv_tf = X_train, X_test, X_cv\n",
    "y_train_tf, y_test_tf, y_cv_tf = y_train, y_test, y_cv\n",
    "\n",
    "tf_idf_vect = TfidfVectorizer(ngram_range=(1,2), min_df=10)\n",
    "final_tf_idf_train = tf_idf_vect.fit_transform(X_train_tf)\n",
    "print(\"some sample features(unique words in the corpus)\",tf_idf_vect.get_feature_names()[0:10])\n",
    "print('='*50)\n",
    "\n",
    "\n",
    "final_tf_idf_cv = tf_idf_vect.transform(X_cv_tf)\n",
    "final_tf_idf_test = tf_idf_vect.transform(X_test_tf)\n",
    "print(\"the type of count vectorizer \",type(final_tf_idf_train))\n",
    "print(\"the shape of out text TFIDF vectorizer \",final_tf_idf_train.get_shape())\n",
    "print(\"the number of unique words including both unigrams and bigrams \", final_tf_idf_train.get_shape()[1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## [4.4] Word2Vec"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Train your own Word2Vec model using your own text corpus\n",
    "\n",
    "X_train_w2v, X_test_w2v, X_cv_w2v = X_train, X_test, X_cv\n",
    "y_train_w2v, y_test_w2v, y_cv_w2v = y_train, y_test, y_cv\n",
    "\n",
    "\n",
    "i=0\n",
    "list_of_sentance_test=[]\n",
    "for sentance in X_test_w2v:\n",
    "    list_of_sentance_test.append(sentance.split())\n",
    "\n",
    "i=0\n",
    "list_of_sentance_cv=[]\n",
    "for sentance in X_cv_w2v:\n",
    "    list_of_sentance_cv.append(sentance.split())\n",
    "\n",
    "i=0\n",
    "list_of_sentance=[]\n",
    "for sentance in X_train_w2v:\n",
    "    list_of_sentance.append(sentance.split())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[('awesome', 0.8652348518371582), ('fantastic', 0.8356021046638489), ('good', 0.8076390027999878), ('wonderful', 0.8012189865112305), ('amazing', 0.7997298836708069), ('excellent', 0.7938966751098633), ('terrific', 0.7613211870193481), ('perfect', 0.72663813829422), ('decent', 0.7050527334213257), ('outstanding', 0.6983649730682373)]\n",
      "==================================================\n",
      "[('best', 0.7431176900863647), ('experienced', 0.7216763496398926), ('tastiest', 0.7138723134994507), ('greatest', 0.6920297741889954), ('ever', 0.6838415861129761), ('hooked', 0.6724706292152405), ('eaten', 0.657467246055603), ('horrible', 0.6574298143386841), ('closest', 0.6569807529449463), ('freaked', 0.6451006531715393)]\n"
     ]
    }
   ],
   "source": [
    "# Using Google News Word2Vectors\n",
    "\n",
    "# in this project we are using a pretrained model by google\n",
    "# its 3.3G file, once you load this into your memory \n",
    "# it occupies ~9Gb, so please do this step only if you have >12G of ram\n",
    "# we will provide a pickle file wich contains a dict , \n",
    "# and it contains all our courpus words as keys and  model[word] as values\n",
    "# To use this code-snippet, download \"GoogleNews-vectors-negative300.bin\" \n",
    "# from https://drive.google.com/file/d/0B7XkCwpI5KDYNlNUTTlSS21pQmM/edit\n",
    "# it's 1.9GB in size.\n",
    "\n",
    "\n",
    "# http://kavita-ganesan.com/gensim-word2vec-tutorial-starter-code/#.W17SRFAzZPY\n",
    "# you can comment this whole cell\n",
    "# or change these varible according to your need\n",
    "\n",
    "is_your_ram_gt_16g=False\n",
    "want_to_use_google_w2v = False\n",
    "want_to_train_w2v = True\n",
    "\n",
    "if want_to_train_w2v:\n",
    "    # min_count = 5 considers only words that occured atleast 5 times\n",
    "    w2v_model=Word2Vec(list_of_sentance,min_count=5,size=50, workers=4)\n",
    "    print(w2v_model.wv.most_similar('great'))\n",
    "    print('='*50)\n",
    "    print(w2v_model.wv.most_similar('worst'))\n",
    "    \n",
    "elif want_to_use_google_w2v and is_your_ram_gt_16g:\n",
    "    if os.path.isfile('GoogleNews-vectors-negative300.bin'):\n",
    "        w2v_model=KeyedVectors.load_word2vec_format('GoogleNews-vectors-negative300.bin', binary=True)\n",
    "        print(w2v_model.wv.most_similar('great'))\n",
    "        print(w2v_model.wv.most_similar('worst'))\n",
    "    else:\n",
    "        print(\"you don't have gogole's word2vec file, keep want_to_train_w2v = True, to train your own w2v \")\n",
    "# print(\"\\n\"+\"#\"*60)\n",
    "# print(len(X_train_w2v),len(y_train_w2v))\n",
    "# print(len(X_cv_w2v),len(y_cv_w2v))\n",
    "# print(len(X_test_w2v),len(y_test_w2v))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "number of words that occured minimum 5 times  10047\n",
      "sample words  ['thought', 'wonderful', 'purchased', 'share', 'family', 'hit', 'dad', 'kept', 'sneaking', 'dinner', 'eventually', 'entire', 'tin', 'disappeared', 'seen', 'every', 'awhile', 'mood', 'one', 'fudge', 'comes', 'bag', 'inside', 'piece', 'individually', 'wrapped', 'moment', 'opened', 'packaged', 'wondered', 'would', 'taste', 'strongly', 'alcohol', 'smelled', 'not', 'something', 'like', 'rather', 'surprised', 'hint', 'sweet', 'quite', 'enjoyable', 'highly', 'recommend', 'puppies', 'super', 'small', 'last']\n"
     ]
    }
   ],
   "source": [
    "w2v_words = list(w2v_model.wv.vocab)\n",
    "print(\"number of words that occured minimum 5 times \",len(w2v_words))\n",
    "print(\"sample words \", w2v_words[0:50])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## [4.4.1] Converting text into vectors using Avg W2V, TFIDF-W2V"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### [4.4.1.1] Avg W2v"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████| 27642/27642 [00:41<00:00, 658.26it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "27642\n",
      "50\n"
     ]
    }
   ],
   "source": [
    "# TRAINING W2V DATA VECTORS\n",
    "\n",
    "# average Word2Vec\n",
    "# compute average word2vec for each review.\n",
    "sent_vectors_train = []; # the avg-w2v for each sentence/review is stored in this list\n",
    "for sent in tqdm(list_of_sentance): # for each review/sentence\n",
    "    sent_vec = np.zeros(50) # as word vectors are of zero length 50, you might need to change this to 300 if you use google's w2v\n",
    "    cnt_words =0; # num of words with a valid vector in the sentence/review\n",
    "    for word in sent: # for each word in a review/sentence\n",
    "        if word in w2v_words:\n",
    "            vec = w2v_model.wv[word]\n",
    "            sent_vec += vec\n",
    "            cnt_words += 1\n",
    "    if cnt_words != 0:\n",
    "        sent_vec /= cnt_words\n",
    "    sent_vectors_train.append(sent_vec)\n",
    "print(len(sent_vectors_train))\n",
    "print(len(sent_vectors_train[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|█████████████████████████████████████████████████████████████████████████████| 9214/9214 [00:14<00:00, 647.94it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "9214 9214\n",
      "50\n"
     ]
    }
   ],
   "source": [
    "# CV W2V DATA VECTORS\n",
    "\n",
    "sent_vectors_cv = []; # the avg-w2v for each sentence/review is stored in this list\n",
    "for sent in tqdm(list_of_sentance_cv): # for each review/sentence\n",
    "    sent_vec = np.zeros(50) # as word vectors are of zero length 50, you might need to change this to 300 if you use google's w2v\n",
    "    cnt_words =0; # num of words with a valid vector in the sentence/review\n",
    "    for word in sent: # for each word in a review/sentence\n",
    "        if word in w2v_words:\n",
    "            vec = w2v_model.wv[word]\n",
    "            sent_vec += vec\n",
    "            cnt_words += 1\n",
    "    if cnt_words != 0:\n",
    "        sent_vec /= cnt_words\n",
    "    sent_vectors_cv.append(sent_vec)\n",
    "print(len(sent_vectors_cv), len(y_cv_w2v))\n",
    "print(len(sent_vectors_cv[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|█████████████████████████████████████████████████████████████████████████████| 9215/9215 [00:14<00:00, 619.10it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "9215\n",
      "50\n"
     ]
    }
   ],
   "source": [
    "# TEST W2V DATA VECTORS\n",
    "\n",
    "sent_vectors_test = []; # the avg-w2v for each sentence/review is stored in this list\n",
    "for sent in tqdm(list_of_sentance_test): # for each review/sentence\n",
    "    sent_vec = np.zeros(50) # as word vectors are of zero length 50, you might need to change this to 300 if you use google's w2v\n",
    "    cnt_words =0; # num of words with a valid vector in the sentence/review\n",
    "    for word in sent: # for each word in a review/sentence\n",
    "        if word in w2v_words:\n",
    "            vec = w2v_model.wv[word]\n",
    "            sent_vec += vec\n",
    "            cnt_words += 1\n",
    "    if cnt_words != 0:\n",
    "        sent_vec /= cnt_words\n",
    "    sent_vectors_test.append(sent_vec)\n",
    "print(len(sent_vectors_test))\n",
    "print(len(sent_vectors_test[0]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### [4.4.1.2] TFIDF weighted W2v"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "# S = [\"abc def pqr\", \"def def def abc\", \"pqr pqr def\"]\n",
    "model = TfidfVectorizer()\n",
    "tf_idf_matrix = model.fit_transform(X_train_w2v)\n",
    "# we are converting a dictionary with word as a key, and the idf as a value\n",
    "dictionary = dict(zip(model.get_feature_names(), list(model.idf_)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████| 27642/27642 [07:48<00:00, 59.02it/s]\n"
     ]
    }
   ],
   "source": [
    "# TF-IDF weighted Word2Vec\n",
    "tfidf_feat = model.get_feature_names() # tfidf words/col-names\n",
    "# final_tf_idf is the sparse matrix with row= sentence, col=word and cell_val = tfidf\n",
    "\n",
    "tfidf_sent_vectors_train = []; # the tfidf-w2v for each sentence/review is stored in this list\n",
    "row=0;\n",
    "for sent in tqdm(list_of_sentance): # for each review/sentence \n",
    "    sent_vec = np.zeros(50) # as word vectors are of zero length\n",
    "    weight_sum =0; # num of words with a valid vector in the sentence/review\n",
    "    for word in sent: # for each word in a review/sentence\n",
    "        if word in w2v_words and word in tfidf_feat:\n",
    "            vec = w2v_model.wv[word]\n",
    "#             tf_idf = tf_idf_matrix[row, tfidf_feat.index(word)]\n",
    "            # to reduce the computation we are \n",
    "            # dictionary[word] = idf value of word in whole courpus\n",
    "            # sent.count(word) = tf valeus of word in this review\n",
    "            tf_idf = dictionary[word]*(sent.count(word)/len(sent))\n",
    "            sent_vec += (vec * tf_idf)\n",
    "            weight_sum += tf_idf\n",
    "    if weight_sum != 0:\n",
    "        sent_vec /= weight_sum\n",
    "    tfidf_sent_vectors_train.append(sent_vec)\n",
    "    row += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████████████████████████████████████████████| 9214/9214 [02:14<00:00, 68.39it/s]\n"
     ]
    }
   ],
   "source": [
    "\n",
    "tfidf_sent_vectors_cv = []; # the tfidf-w2v for each sentence/review is stored in this list\n",
    "row=0;\n",
    "for sent in tqdm(list_of_sentance_cv): # for each review/sentence \n",
    "    sent_vec = np.zeros(50) # as word vectors are of zero length\n",
    "    weight_sum =0; # num of words with a valid vector in the sentence/review\n",
    "    for word in sent: # for each word in a review/sentence\n",
    "        if word in w2v_words and word in tfidf_feat:\n",
    "            vec = w2v_model.wv[word]\n",
    "#             tf_idf = tf_idf_matrix[row, tfidf_feat.index(word)]\n",
    "            # to reduce the computation we are \n",
    "            # dictionary[word] = idf value of word in whole courpus\n",
    "            # sent.count(word) = tf valeus of word in this review\n",
    "            tf_idf = dictionary[word]*(sent.count(word)/len(sent))\n",
    "            sent_vec += (vec * tf_idf)\n",
    "            weight_sum += tf_idf\n",
    "    if weight_sum != 0:\n",
    "        sent_vec /= weight_sum\n",
    "    tfidf_sent_vectors_cv.append(sent_vec)\n",
    "    row += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████████████████████████████████████████████| 9215/9215 [02:24<00:00, 63.61it/s]\n"
     ]
    }
   ],
   "source": [
    "\n",
    "tfidf_sent_vectors_test = []; # the tfidf-w2v for each sentence/review is stored in this list\n",
    "row=0;\n",
    "for sent in tqdm(list_of_sentance_test): # for each review/sentence \n",
    "    sent_vec = np.zeros(50) # as word vectors are of zero length\n",
    "    weight_sum =0; # num of words with a valid vector in the sentence/review\n",
    "    for word in sent: # for each word in a review/sentence\n",
    "        if word in w2v_words and word in tfidf_feat:\n",
    "            vec = w2v_model.wv[word]\n",
    "#             tf_idf = tf_idf_matrix[row, tfidf_feat.index(word)]\n",
    "            # to reduce the computation we are \n",
    "            # dictionary[word] = idf value of word in whole courpus\n",
    "            # sent.count(word) = tf valeus of word in this review\n",
    "            tf_idf = dictionary[word]*(sent.count(word)/len(sent))\n",
    "            sent_vec += (vec * tf_idf)\n",
    "            weight_sum += tf_idf\n",
    "    if weight_sum != 0:\n",
    "        sent_vec /= weight_sum\n",
    "    tfidf_sent_vectors_test.append(sent_vec)\n",
    "    row += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# [5] Assignment 3: KNN"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<ol>\n",
    "    <li><strong>Apply Knn(brute force version) on these feature sets</strong>\n",
    "        <ul>\n",
    "            <li><font color='red'>SET 1:</font>Review text, preprocessed one converted into vectors using (BOW)</li>\n",
    "            <li><font color='red'>SET 2:</font>Review text, preprocessed one converted into vectors using (TFIDF)</li>\n",
    "            <li><font color='red'>SET 3:</font>Review text, preprocessed one converted into vectors using (AVG W2v)</li>\n",
    "            <li><font color='red'>SET 4:</font>Review text, preprocessed one converted into vectors using (TFIDF W2v)</li>\n",
    "        </ul>\n",
    "    </li>\n",
    "    <br>\n",
    "    <li><strong>Apply Knn(kd tree version) on these feature sets</strong>\n",
    "        <br><font color='red'>NOTE: </font>sklearn implementation of kd-tree accepts only dense matrices, you need to convert the sparse matrices of CountVectorizer/TfidfVectorizer into dense matices. You can convert sparse matrices to dense using .toarray() attribute. For more information please visit this <a href='https://docs.scipy.org/doc/scipy-0.18.1/reference/generated/scipy.sparse.csr_matrix.toarray.html'>link</a>\n",
    "        <ul>\n",
    "            <li><font color='red'>SET 5:</font>Review text, preprocessed one converted into vectors using (BOW) but with restriction on maximum features generated.\n",
    "            <pre>\n",
    "            count_vect = CountVectorizer(min_df=10, max_features=500) \n",
    "            count_vect.fit(preprocessed_reviews)\n",
    "            </pre>\n",
    "            </li>\n",
    "            <li><font color='red'>SET 6:</font>Review text, preprocessed one converted into vectors using (TFIDF) but with restriction on maximum features generated.\n",
    "            <pre>\n",
    "                tf_idf_vect = TfidfVectorizer(min_df=10, max_features=500)\n",
    "                tf_idf_vect.fit(preprocessed_reviews)\n",
    "            </pre>\n",
    "            </li>\n",
    "            <li><font color='red'>SET 3:</font>Review text, preprocessed one converted into vectors using (AVG W2v)</li>\n",
    "            <li><font color='red'>SET 4:</font>Review text, preprocessed one converted into vectors using (TFIDF W2v)</li>\n",
    "        </ul>\n",
    "    </li>\n",
    "    <br>\n",
    "    <li><strong>The hyper paramter tuning(find best K)</strong>\n",
    "        <ul>\n",
    "    <li>Find the best hyper parameter which will give the maximum <a href='https://www.appliedaicourse.com/course/applied-ai-course-online/lessons/receiver-operating-characteristic-curve-roc-curve-and-auc-1/'>AUC</a> value</li>\n",
    "    <li>Find the best hyper paramter using k-fold cross validation or simple cross validation data</li>\n",
    "    <li>Use gridsearch cv or randomsearch cv or you can also write your own for loops to do this task of hyperparameter tuning</li>\n",
    "        </ul>\n",
    "    </li>\n",
    "    <br>\n",
    "    <li>\n",
    "    <strong>Representation of results</strong>\n",
    "        <ul>\n",
    "    <li>You need to plot the performance of model both on train data and cross validation data for each hyper parameter, like shown in the figure\n",
    "    <img src='train_cv_auc.JPG' width=300px></li>\n",
    "    <li>Once after you found the best hyper parameter, you need to train your model with it, and find the AUC on test data and plot the ROC curve on both train and test.\n",
    "    <img src='train_test_auc.JPG' width=300px></li>\n",
    "    <li>Along with plotting ROC curve, you need to print the <a href='https://www.appliedaicourse.com/course/applied-ai-course-online/lessons/confusion-matrix-tpr-fpr-fnr-tnr-1/'>confusion matrix</a> with predicted and original labels of test data points\n",
    "    <img src='confusion_matrix.png' width=300px></li>\n",
    "        </ul>\n",
    "    </li>\n",
    "    <br>\n",
    "    <li><strong>Conclusion</strong>\n",
    "        <ul>\n",
    "    <li>You need to summarize the results at the end of the notebook, summarize it in the table format. To print out a table please refer to this prettytable library<a href='http://zetcode.com/python/prettytable/'> link</a> \n",
    "        <img src='summary.JPG' width=400px>\n",
    "    </li>\n",
    "        </ul>\n",
    "</ol>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h4><font color='red'>Note: Data Leakage</font></h4>\n",
    "\n",
    "1. There will be an issue of data-leakage if you vectorize the entire data and then split it into train/cv/test.\n",
    "2. To avoid the issue of data-leakag, make sure to split your data first and then vectorize it. \n",
    "3. While vectorizing your data, apply the method fit_transform() on you train data, and apply the method transform() on cv/test data.\n",
    "4. For more details please go through this <a href='https://soundcloud.com/applied-ai-course/leakage-bow-and-tfidf'>link.</a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## [5.1] Applying KNN brute force"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### [5.1.1] Applying KNN brute force on BOW,<font color='red'> SET 1</font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Please write all the code with proper documentation\n",
    "X_train_bow, X_test_bow, y_train, y_test, X_cv_bow, y_cv = bow_counts_train, bow_counts_test, y_train, y_test, bow_counts_cv, y_cv\n",
    "\n",
    "\n",
    "auc_train=[]\n",
    "ks_train=[]\n",
    "auc_cv=[]\n",
    "ks_cv=[]\n",
    "\n",
    "############ HYPER-PARAMETER TUNING ##############\n",
    "for i in range(1,50,6):\n",
    "    knn = KNeighborsClassifier(n_neighbors=i, algorithm='brute', metric='cosine') #with eucledian distance metric auc is only 0.67\n",
    "    knn.fit(X_train_bow, y_train)\n",
    "    # y_pred = knn.predict(X_test_bow)\n",
    "    proba = knn.predict_proba(X_train_bow)[:,1]\n",
    "    auc_train.append(roc_auc_score(y_train, proba))\n",
    "    ks_train.append(i)\n",
    "\n",
    "    proba = knn.predict_proba(X_cv_bow)[:,1]\n",
    "    auc_cv.append(roc_auc_score(y_cv, proba))\n",
    "    ks_cv.append(i)\n",
    "\n",
    "# print(\"Pre-def accuracy: \", metrics.accuracy_score(y_test, y_pred))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "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"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best K for KNN Model from cv_data is:  49 for cooresponding AUC value of: 0.8388002018700679\n"
     ]
    }
   ],
   "source": [
    "######## KNN BRUTE-FORCE AUC PLOTTING ########\n",
    "ax=sns.lineplot(ks_train,auc_train)\n",
    "ax=sns.lineplot(ks_cv,auc_cv)\n",
    "ax.set(xlabel='Differnt K values',ylabel='roc_auc_score')\n",
    "ax.set_title('AUC Score vs Different K value Curve')\n",
    "plt.show()\n",
    "\n",
    "###############################################\n",
    "\n",
    "############## BEST HYPER-PARAMETER ###########\n",
    "best_k = ks_cv[auc_cv.index(max(auc_cv))]\n",
    "print(\"Best K for KNN Model from cv_data is: \",best_k, \"for cooresponding AUC value of:\",max(auc_cv))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUC score on test_data with best_k:  0.839391109460291\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "######### FRESH-MODEL FOR BEST_K ######################\n",
    "knn = KNeighborsClassifier(n_neighbors=best_k,algorithm='brute', metric='cosine')\n",
    "knn.fit(X_train_bow, y_train)\n",
    "\n",
    "y_pred = knn.predict(X_test_bow)\n",
    "proba = knn.predict_proba(X_test_bow)[:,1]\n",
    "fpr, tpr, threshold = roc_curve(y_test, proba)\n",
    "roc_auc = auc(fpr, tpr)\n",
    "print(\"AUC score on test_data with best_k: \",roc_auc)\n",
    "proba = knn.predict_proba(X_train_bow)[:,1]\n",
    "fpr_train, tpr_train, threshold_train = roc_curve(y_train, proba)\n",
    "roc_auc_train = auc(fpr_train, tpr_train)\n",
    "\n",
    "# storing values for future use\n",
    "k_1 = best_k\n",
    "set_1_auc = roc_auc\n",
    "\n",
    "############# ROC CURVE ######################\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.plot(fpr_train, tpr_train, 'g', label = 'AUC_train = %0.3f' % roc_auc_train)\n",
    "plt.plot(fpr, tpr, 'b', label = 'AUC_test = %0.3f' % roc_auc)\n",
    "plt.legend(loc = 'lower right')\n",
    "plt.plot([0, 1], [0, 1],'r--')\n",
    "plt.xlim([0, 1])\n",
    "plt.ylim([0, 1])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.title('ROC Curve of kNN')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "########### CONFUSION MATRIX ############\n",
    "\n",
    "y_true = y_test\n",
    "y_pred = y_pred\n",
    "\n",
    "array = confusion_matrix(y_true, y_pred)\n",
    "df_cm = pd.DataFrame(array, index = [i for i in \"01\"],columns = [i for i in \"01\"])\n",
    "\n",
    "plt.figure(figsize = (5,5))\n",
    "sns.heatmap(df_cm, annot=True,annot_kws={\"size\": 16}, linewidths=6.3, linecolor='pink', cbar=False, square=True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### [5.1.2] Applying KNN brute force on TFIDF,<font color='red'> SET 2</font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 247,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Please write all the code with proper documentation\n",
    "\n",
    "X_train_tf, X_test_tf, X_cv_tf = final_tf_idf_train, final_tf_idf_test,final_tf_idf_cv\n",
    "\n",
    "auc_train=[]\n",
    "ks_train=[]\n",
    "auc_cv=[]\n",
    "ks_cv=[]\n",
    "\n",
    "####### HYPER PARAMETER TUNING ########\n",
    "for i in range(1,50,6):\n",
    "    knn = KNeighborsClassifier(n_neighbors=i, algorithm='brute', metric='cosine')\n",
    "    knn.fit(X_train_tf, y_train_tf)\n",
    "    # y_pred = knn.predict(X_test)\n",
    "    proba = knn.predict_proba(X_train_tf)[:,1]\n",
    "    auc_train.append(roc_auc_score(y_train_tf, proba))\n",
    "    ks_train.append(i)\n",
    "    \n",
    "    proba = knn.predict_proba(X_cv_tf)[:,1]\n",
    "    auc_cv.append(roc_auc_score(y_cv_tf, proba))\n",
    "    ks_cv.append(i)\n",
    "\n",
    "# print(\"Pre-def accuracy: \", metrics.accuracy_score(y_test, y_pred))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 248,
   "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"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best K for KNN Model from cv_data is:  49 for cooresponding AUC value of: 0.8558222562956248\n"
     ]
    }
   ],
   "source": [
    "\n",
    "######## AUC PLOTTING ########\n",
    "ax=sns.lineplot(ks_train,auc_train,dashes=True)\n",
    "ax=sns.lineplot(ks_cv,auc_cv)\n",
    "ax.set(xlabel='Differnt K values',ylabel='roc_auc_score')\n",
    "ax.set_title('AUC Score vs Different K value Curve')\n",
    "plt.show()\n",
    "######## AUC PLOTTING ########\n",
    "\n",
    "\n",
    "######## best_k ##############\n",
    "best_k = ks_cv[auc_cv.index(max(auc_cv))]\n",
    "print(\"Best K for KNN Model from cv_data is: \",best_k, \"for cooresponding AUC value of:\",max(auc_cv))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 249,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUC score on test_data with best_k:  0.8565216902657031\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "######### FRESH-MODEL FOR BEST_K ######################\n",
    "knn = KNeighborsClassifier(n_neighbors=best_k,algorithm='brute', metric='cosine')\n",
    "knn.fit(X_train_tf, y_train_tf)\n",
    "y_pred = knn.predict(X_test_tf)\n",
    "\n",
    "proba = knn.predict_proba(X_test_tf)[:,1]\n",
    "fpr, tpr, threshold = roc_curve(y_test_tf, proba)\n",
    "roc_auc = auc(fpr, tpr)\n",
    "print(\"AUC score on test_data with best_k: \",roc_auc)\n",
    "#############################################\n",
    "proba = knn.predict_proba(X_train_tf)[:,1]\n",
    "fpr_train, tpr_train, threshold_train = roc_curve(y_train_tf, proba)\n",
    "roc_auc_train = auc(fpr_train, tpr_train)\n",
    "\n",
    "\n",
    "# storing values for future use\n",
    "k_2 = best_k\n",
    "set_2_auc = roc_auc\n",
    "\n",
    "############# ROC ######################\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.plot(fpr_train, tpr_train, 'g', label = 'AUC_train = %0.2f' % roc_auc_train)\n",
    "plt.plot(fpr, tpr, 'b', label = 'AUC_test = %0.2f' % roc_auc)\n",
    "plt.legend(loc = 'lower right')\n",
    "plt.plot([0, 1], [0, 1],'r--')\n",
    "plt.xlim([0, 1])\n",
    "plt.ylim([0, 1])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.title('ROC Curve of kNN')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 250,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_true = y_test_tf\n",
    "y_pred = y_pred\n",
    "\n",
    "############ CONFUSION MATRIX ##################\n",
    "array = confusion_matrix(y_true, y_pred)\n",
    "df_cm = pd.DataFrame(array, index = [i for i in \"01\"],\n",
    "                  columns = [i for i in \"01\"])\n",
    "\n",
    "plt.figure(figsize = (5,5))\n",
    "plt.title('Heatmap Confusion Matrix')\n",
    "sns.heatmap(df_cm, annot=True,annot_kws={\"size\": 16}, linewidths=8.3, linecolor='pink', cbar=False, square=True)\n",
    "plt.show()\n",
    "\n",
    "# y_test_tf\n",
    "# unique, counts = np.unique(y_test_tf, return_counts=True)\n",
    "# dict(zip(unique, counts))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### [5.1.3] Applying KNN brute force on AVG W2V,<font color='red'> SET 3</font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 194,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Please write all the code with proper documentation\n",
    "\n",
    "X_train_avgw2v, X_cv_avgw2v, X_test_avgw2v = sent_vectors_train, sent_vectors_cv, sent_vectors_test\n",
    "auc_train=[]\n",
    "ks_train=[]\n",
    "auc_cv=[]\n",
    "ks_cv=[]\n",
    "\n",
    "##### HYPER PARAMETER TUNING ###### b\n",
    "for i in range(1,50,6):\n",
    "    knn = KNeighborsClassifier(n_neighbors=i, algorithm='brute')\n",
    "    knn.fit(X_train_avgw2v, y_train_w2v)\n",
    "    # y_pred = knn.predict(X_test_avgw2v)\n",
    "    proba = knn.predict_proba(X_train_avgw2v)[:,1]\n",
    "    auc_train.append(roc_auc_score(y_train_w2v, proba))\n",
    "    ks_train.append(i)\n",
    "    \n",
    "    proba = knn.predict_proba(X_cv_avgw2v)[:,1]\n",
    "    auc_cv.append(roc_auc_score(y_cv_w2v, proba))\n",
    "    ks_cv.append(i)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 195,
   "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"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best K for KNN Model from cv_data is:  49 for cooresponding AUC value of: 0.8779936776255125\n"
     ]
    }
   ],
   "source": [
    "\n",
    "######## AUC PLOTTING ########\n",
    "ax=sns.lineplot(ks_train,auc_train)\n",
    "ax=sns.lineplot(ks_cv,auc_cv)\n",
    "ax.set(xlabel='Differnt K values',ylabel='roc_auc_score')\n",
    "ax.set_title('AUC Score vs Different K value Curve')\n",
    "plt.show()\n",
    "\n",
    "######## best_k ##############\n",
    "best_k = ks_cv[auc_cv.index(max(auc_cv))]\n",
    "print(\"Best K for KNN Model from cv_data is: \",best_k, \"for cooresponding AUC value of:\",max(auc_cv))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 196,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUC score on test_data with best_k:  0.8735792965658518\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "######### FRESH-MODEL FOR BEST_K ######################\n",
    "knn = KNeighborsClassifier(n_neighbors=best_k,algorithm='brute')\n",
    "knn.fit(X_train_avgw2v, y_train_w2v)\n",
    "y_pred = knn.predict(X_test_avgw2v)\n",
    "proba = knn.predict_proba(X_test_avgw2v)[:,1]\n",
    "\n",
    "\n",
    "fpr, tpr, threshold = roc_curve(y_test_w2v, proba)\n",
    "roc_auc = auc(fpr, tpr)\n",
    "print(\"AUC score on test_data with best_k: \",roc_auc)\n",
    "#############################################\n",
    "proba = knn.predict_proba(X_train_avgw2v)[:,1]\n",
    "fpr_train, tpr_train, threshold_train = roc_curve(y_train_w2v, proba)\n",
    "roc_auc_train = auc(fpr_train, tpr_train)\n",
    "\n",
    "k_3 = best_k\n",
    "set_3_auc = roc_auc\n",
    "############# ROC ######################\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.plot(fpr_train, tpr_train, 'g', label = 'AUC_train = %0.3f' % roc_auc_train)\n",
    "plt.plot(fpr, tpr, 'b', label = 'AUC_test = %0.3f' % roc_auc)\n",
    "plt.legend(loc = 'lower right')\n",
    "plt.plot([0, 1], [0, 1],'r--')\n",
    "plt.xlim([0, 1])\n",
    "plt.ylim([0, 1])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.title('ROC Curve of kNN')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 197,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "########## CONFUSION MATRIX ###########\n",
    "\n",
    "y_true = y_test_w2v\n",
    "y_pred = y_pred\n",
    "\n",
    "array = confusion_matrix(y_true, y_pred)\n",
    "\n",
    "df_cm = pd.DataFrame(array, index = [i for i in \"01\"],\n",
    "                  columns = [i for i in \"01\"])\n",
    "\n",
    "\n",
    "plt.figure(figsize = (5,5))\n",
    "sns.heatmap(df_cm, annot=True,annot_kws={\"size\": 16}, linewidths=6.3, linecolor='pink', cbar=False, square=True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### [5.1.4] Applying KNN brute force on TFIDF W2V,<font color='red'> SET 4</font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 198,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Please write all the code with proper documentation\n",
    "\n",
    "X_train_tfw2v, X_cv_w2v, X_test_tfw2v = tfidf_sent_vectors_train, tfidf_sent_vectors_cv, tfidf_sent_vectors_test\n",
    "\n",
    "auc_train=[]\n",
    "ks_train=[]\n",
    "auc_cv=[]\n",
    "ks_cv=[]\n",
    "\n",
    "for i in range(1,50,4):\n",
    "    knn = KNeighborsClassifier(n_neighbors=i, algorithm='brute') #try distance cosine too\n",
    "    knn.fit(X_train_tfw2v, y_train_w2v)\n",
    "    # y_pred = knn.predict(X_test_tfw2v)\n",
    "    proba = knn.predict_proba(X_train_tfw2v)[:,1]\n",
    "    auc_train.append(roc_auc_score(y_train_w2v, proba))\n",
    "    ks_train.append(i)\n",
    "    #############################################################\n",
    "    proba = knn.predict_proba(X_cv_w2v)[:,1]\n",
    "    auc_cv.append(roc_auc_score(y_cv_w2v, proba))\n",
    "    ks_cv.append(i)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 199,
   "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"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best K for KNN Model from cv_data is:  45 for cooresponding AUC value of: 0.8504502687876312\n"
     ]
    }
   ],
   "source": [
    "######## AUC PLOTTING ########\n",
    "ax=sns.lineplot(ks_train,auc_train)\n",
    "ax=sns.lineplot(ks_cv,auc_cv)\n",
    "ax.set(xlabel='Differnt K values',ylabel='roc_auc_score')\n",
    "ax.set_title('AUC Score vs Different K value Curve')\n",
    "plt.show()\n",
    "\n",
    "######## best_k ##############A\n",
    "best_k = ks_cv[auc_cv.index(max(auc_cv))]\n",
    "print(\"Best K for KNN Model from cv_data is: \",best_k, \"for cooresponding AUC value of:\",max(auc_cv))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 200,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUC score on test_data with best_k:  0.8463395892582682\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "\n",
    "######### FRESH-MODEL FOR BEST_K ######################\n",
    "knn = KNeighborsClassifier(n_neighbors=best_k,algorithm='brute')\n",
    "knn.fit(X_train_tfw2v, y_train_w2v)\n",
    "y_pred = knn.predict(X_test_tfw2v)\n",
    "proba = knn.predict_proba(X_test_tfw2v)[:,1]\n",
    "\n",
    "\n",
    "fpr, tpr, threshold = roc_curve(y_test_w2v, proba)\n",
    "roc_auc = auc(fpr, tpr)\n",
    "print(\"AUC score on test_data with best_k: \",roc_auc)\n",
    "#############################################\n",
    "proba = knn.predict_proba(X_train_tfw2v)[:,1]\n",
    "fpr_train, tpr_train, threshold_train = roc_curve(y_train_w2v, proba)\n",
    "roc_auc_train = auc(fpr_train, tpr_train)\n",
    "\n",
    "# storing values for future references\n",
    "k_4 = best_k\n",
    "set_4_auc = roc_auc\n",
    "############# ROC ######################\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.plot(fpr_train, tpr_train, 'g', label = 'AUC_train = %0.3f' % roc_auc_train)\n",
    "plt.plot(fpr, tpr, 'b', label = 'AUC_test = %0.3f' % roc_auc)\n",
    "plt.legend(loc = 'lower right')\n",
    "plt.plot([0, 1], [0, 1],'r--')\n",
    "plt.xlim([0, 1])\n",
    "plt.ylim([0, 1])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.title('ROC Curve of kNN')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 201,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "# Confusion Martrix\n",
    "y_true = y_test_w2v\n",
    "y_pred = y_pred\n",
    "\n",
    "from sklearn.metrics import confusion_matrix\n",
    "\n",
    "array = confusion_matrix(y_true, y_pred)\n",
    "\n",
    "df_cm = pd.DataFrame(array, index = [i for i in \"01\"],\n",
    "                  columns = [i for i in \"01\"])\n",
    "\n",
    "\n",
    "plt.figure(figsize = (5,5))\n",
    "sns.heatmap(df_cm, annot=True,annot_kws={\"size\": 16}, linewidths=6.3, linecolor='pink', cbar=False, square=True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## [5.2] Applying KNN kd-tree"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### [5.2.1] Applying KNN kd-tree on BOW,<font color='red'> SET 5</font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 202,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "some feature names  ['aa', 'aaaa', 'aaaand', 'aachen', 'aadp', 'aafco', 'aahs', 'ab', 'aback', 'abandon']\n",
      "==================================================\n",
      "the type of count vectorizer  <class 'scipy.sparse.csr.csr_matrix'>\n",
      "the shape of out text BOW vectorizer  (12000, 20901)\n",
      "the number of unique words  20901\n"
     ]
    }
   ],
   "source": [
    "# Please write all the code with proper documentation\n",
    "\n",
    "# Using only 20k data points:\n",
    "X_temp= X[:20000]\n",
    "y_temp= y[:20000]\n",
    "\n",
    "#####  SPLIT  #######\n",
    "X_train_bow, X_test_bow, y_train_bow, y_test_bow = train_test_split(X_temp, y_temp, test_size=0.2, random_state=4)\n",
    "X_train_bow, X_cv_bow, y_train_bow, y_cv_bow = train_test_split(X_train_bow, y_train_bow, test_size=0.25, random_state=4)\n",
    "\n",
    "##### BOW Vectorization ######\n",
    "count_vect = CountVectorizer() #in scikit-learn\n",
    "count_vect.fit(X_train_bow)\n",
    "print(\"some feature names \", count_vect.get_feature_names()[:10])\n",
    "print('='*50)\n",
    "\n",
    "bow_counts_train = count_vect.transform(X_train_bow)\n",
    "bow_counts_cv = count_vect.transform(X_cv_bow)\n",
    "bow_counts_test = count_vect.transform(X_test_bow)\n",
    "print(\"the type of count vectorizer \",type(bow_counts_train))\n",
    "print(\"the shape of out text BOW vectorizer \",bow_counts_train.get_shape())\n",
    "print(\"the number of unique words \", bow_counts_train.get_shape()[1])\n",
    "bow_counts_test=bow_counts_test.toarray()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 203,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Please write all the code with proper documentation\n",
    "# KNN on BOW using kd-tree\n",
    "X_train_bow, X_test_bow, y_train, y_test, X_cv_bow, y_cv = bow_counts_train, bow_counts_test, y_train_bow, y_test_bow, bow_counts_cv, y_cv_bow\n",
    "\n",
    "\n",
    "auc_train=[]\n",
    "ks_train=[]\n",
    "auc_cv=[]\n",
    "ks_cv=[]\n",
    "\n",
    "############ HYPER-PARAMETER TUNING ##############\n",
    "for i in range(1,50,4):\n",
    "    knn = KNeighborsClassifier(n_neighbors=i, algorithm='kd_tree')\n",
    "    knn.fit(X_train_bow, y_train)\n",
    "    # y_pred = knn.predict(X_test_bow)\n",
    "    proba = knn.predict_proba(X_train_bow)[:,1]\n",
    "    auc_train.append(roc_auc_score(y_train, proba))\n",
    "    ks_train.append(i)\n",
    "\n",
    "    proba = knn.predict_proba(X_cv_bow)[:,1]\n",
    "    auc_cv.append(roc_auc_score(y_cv, proba))\n",
    "    ks_cv.append(i)\n",
    "\n",
    "# print(\"Pre-def accuracy: \", metrics.accuracy_score(y_test, y_pred))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 204,
   "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"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best K for KNN Model from cv_data is:  49 for cooresponding AUC value of: 0.6789526309933722\n"
     ]
    }
   ],
   "source": [
    "######## KNN kd-tree AUC PLOTTING ########\n",
    "ax=sns.lineplot(ks_train,auc_train)\n",
    "ax=sns.lineplot(ks_cv,auc_cv)\n",
    "ax.set(xlabel='Differnt K values',ylabel='roc_auc_score')\n",
    "ax.set_title('AUC Score vs Different K value Curve')\n",
    "plt.show()\n",
    "\n",
    "############## BEST HYPER-PARAMETER ###########\n",
    "best_k = ks_cv[auc_cv.index(max(auc_cv))]\n",
    "print(\"Best K for KNN Model from cv_data is: \",best_k, \"for cooresponding AUC value of:\",max(auc_cv))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 205,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUC score on test_data with best_k:  0.6777641325639919\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "######### FRESH-MODEL FOR BEST_K ######################\n",
    "knn = KNeighborsClassifier(n_neighbors=best_k,algorithm='kd_tree')\n",
    "knn.fit(X_train_bow, y_train)\n",
    "\n",
    "y_pred = knn.predict(X_test_bow)\n",
    "proba = knn.predict_proba(X_test_bow)[:,1]\n",
    "fpr, tpr, threshold = roc_curve(y_test, proba)\n",
    "roc_auc = auc(fpr, tpr)\n",
    "print(\"AUC score on test_data with best_k: \",roc_auc)\n",
    "\n",
    "proba = knn.predict_proba(X_train_bow)[:,1]\n",
    "fpr_train, tpr_train, threshold_train = roc_curve(y_train, proba)\n",
    "roc_auc_train = auc(fpr_train, tpr_train)\n",
    "\n",
    "k_5 = best_k\n",
    "set_5_auc = roc_auc\n",
    "\n",
    "############# ROC CURVE ######################\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.plot(fpr_train, tpr_train, 'g', label = 'AUC_train = %0.3f' % roc_auc_train)\n",
    "plt.plot(fpr, tpr, 'b', label = 'AUC_test = %0.3f' % roc_auc)\n",
    "plt.legend(loc = 'lower right')\n",
    "plt.plot([0, 1], [0, 1],'r--')\n",
    "plt.xlim([0, 1])\n",
    "plt.ylim([0, 1])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.title('ROC Curve of kNN')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 206,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "########### CONFUSION MATRIX ############\n",
    "\n",
    "y_true = y_test\n",
    "y_pred = y_pred\n",
    "\n",
    "array = confusion_matrix(y_true, y_pred)\n",
    "df_cm = pd.DataFrame(array, index = [i for i in \"01\"],columns = [i for i in \"01\"])\n",
    "\n",
    "plt.figure(figsize = (5,5))\n",
    "sns.heatmap(df_cm, annot=True,annot_kws={\"size\": 16}, linewidths=6.3, linecolor='pink', cbar=False, square=True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### [5.2.2] Applying KNN kd-tree on TFIDF,<font color='red'> SET 6</font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 207,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "some sample features(unique words in the corpus) ['ability', 'able', 'able buy', 'able find', 'able get', 'able order', 'able purchase', 'absolute', 'absolute favorite', 'absolutely']\n",
      "==================================================\n",
      "the type of count vectorizer  <class 'scipy.sparse.csr.csr_matrix'>\n",
      "the shape of out text TFIDF vectorizer  (9800, 6002)\n",
      "the number of unique words including both unigrams and bigrams  6002\n"
     ]
    }
   ],
   "source": [
    "# Please write all the code with proper documentation\n",
    "\n",
    "# Split\n",
    "X_train_tf, X_test_tf, y_train_tf, y_test_tf = train_test_split(X_temp, y_temp, test_size=0.3, random_state=4)\n",
    "X_train_tf, X_cv_tf, y_train_tf, y_cv_tf = train_test_split(X_train_tf, y_train_tf, test_size=0.3, random_state=4)\n",
    "# print(len(X_train_tf),len(X_test_tf),len(y_train_tf),len(y_test_tf),len(y_cv_tf),len(X_cv_tf))\n",
    "\n",
    "# TFIDF Vectorisation\n",
    "tf_idf_vect = TfidfVectorizer(ngram_range=(1,2), min_df=10)\n",
    "final_tf_idf_train = tf_idf_vect.fit_transform(X_train_tf)\n",
    "print(\"some sample features(unique words in the corpus)\",tf_idf_vect.get_feature_names()[0:10])\n",
    "print('='*50)\n",
    "\n",
    "\n",
    "final_tf_idf_cv = tf_idf_vect.transform(X_cv_tf)\n",
    "final_tf_idf_test = tf_idf_vect.transform(X_test_tf)\n",
    "print(\"the type of count vectorizer \",type(final_tf_idf_train))\n",
    "print(\"the shape of out text TFIDF vectorizer \",final_tf_idf_train.get_shape())\n",
    "print(\"the number of unique words including both unigrams and bigrams \", final_tf_idf_train.get_shape()[1])\n",
    "\n",
    "final_tf_idf_train =final_tf_idf_train.toarray()\n",
    "final_tf_idf_test =final_tf_idf_test.toarray()\n",
    "final_tf_idf_cv =final_tf_idf_cv.toarray()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 208,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Please write all the code with proper documentation\n",
    "# KNN on tfidf using kd-tree\n",
    "X_train_tf, X_test_tf, X_cv_tf = final_tf_idf_train, final_tf_idf_test,final_tf_idf_cv\n",
    "\n",
    "auc_train=[]\n",
    "ks_train=[]\n",
    "auc_cv=[]\n",
    "ks_cv=[]\n",
    "\n",
    "for i in range(1,50,6):\n",
    "    knn = KNeighborsClassifier(n_neighbors=i, algorithm='kd_tree')\n",
    "    knn.fit(X_train_tf, y_train_tf)\n",
    "    # y_pred = knn.predict(X_test)\n",
    "    proba = knn.predict_proba(X_train_tf)[:,1]\n",
    "    auc_train.append(roc_auc_score(y_train_tf, proba))\n",
    "    ks_train.append(i)\n",
    "    \n",
    "    proba = knn.predict_proba(X_cv_tf)[:,1]\n",
    "    auc_cv.append(roc_auc_score(y_cv_tf, proba))\n",
    "    ks_cv.append(i)\n",
    "\n",
    "# print(\"Pre-def accuracy: \", metrics.accuracy_score(y_test, y_pred))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 209,
   "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"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best K for KNN Model from cv_data is:  49 for cooresponding AUC value of: 0.8225042096288817\n"
     ]
    }
   ],
   "source": [
    "\n",
    "######## AUC PLOTTING ########\n",
    "ax=sns.lineplot(ks_train,auc_train,dashes=True)\n",
    "ax=sns.lineplot(ks_cv,auc_cv)\n",
    "ax.set(xlabel='Differnt K values',ylabel='roc_auc_score')\n",
    "ax.set_title('AUC Score vs Different K value Curve')\n",
    "plt.show()\n",
    "######## AUC PLOTTING ########\n",
    "\n",
    "\n",
    "######## best_k ##############\n",
    "best_k = ks_cv[auc_cv.index(max(auc_cv))]\n",
    "print(\"Best K for KNN Model from cv_data is: \",best_k, \"for cooresponding AUC value of:\",max(auc_cv))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 210,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUC score on test_data with best_k:  0.8565216902657031\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "######### FRESH-MODEL FOR BEST_K ######################\n",
    "knn = KNeighborsClassifier(n_neighbors=best_k,algorithm='brute', metric='cosine')\n",
    "knn.fit(X_train_tf, y_train_tf)\n",
    "y_pred = knn.predict(X_test_tf)\n",
    "\n",
    "\n",
    "proba = knn.predict_proba(X_test_tf)[:,1]\n",
    "fpr, tpr, threshold = roc_curve(y_test_tf, proba)\n",
    "roc_auc = auc(fpr, tpr)\n",
    "print(\"AUC score on test_data with best_k: \",roc_auc)\n",
    "#############################################\n",
    "proba = knn.predict_proba(X_train_tf)[:,1]\n",
    "fpr_train, tpr_train, threshold_train = roc_curve(y_train_tf, proba)\n",
    "roc_auc_train = auc(fpr_train, tpr_train)\n",
    "\n",
    "k_6 = best_k\n",
    "set_6_auc = roc_auc\n",
    "############# ROC ######################\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.plot(fpr_train, tpr_train, 'g', label = 'AUC_train = %0.2f' % roc_auc_train)\n",
    "plt.plot(fpr, tpr, 'b', label = 'AUC_test = %0.2f' % roc_auc)\n",
    "plt.legend(loc = 'lower right')\n",
    "plt.plot([0, 1], [0, 1],'r--')\n",
    "plt.xlim([0, 1])\n",
    "plt.ylim([0, 1])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.title('ROC Curve of kNN')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 221,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_true = y_test_tf\n",
    "y_pred = y_pred\n",
    "\n",
    "############ CONFUSION MATRIX ##################\n",
    "array = confusion_matrix(y_true, y_pred)\n",
    "df_cm = pd.DataFrame(array, index = [i for i in \"01\"],\n",
    "                  columns = [i for i in \"01\"])\n",
    "\n",
    "\n",
    "plt.figure(figsize = (5,5))\n",
    "sns.heatmap(df_cm, annot=True,annot_kws={\"size\": 16}, linewidths=6.3, linecolor='pink', cbar=False, square=True)\n",
    "plt.title('Heatmap Confusion Matrix')\n",
    "plt.show()\n",
    "# y_test_tf\n",
    "# unique, counts = np.unique(y_test_tf, return_counts=True)\n",
    "# dict(zip(unique, counts))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### [5.2.3] Applying KNN kd-tree on AVG W2V,<font color='red'> SET 3</font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### [5.2.3.1] Word2Vec for 20k data-points\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 212,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Train your own Word2Vec model using your own text corpus\n",
    "\n",
    "\n",
    "#####  SPLIT  #######\n",
    "X_train_w2v, X_test_w2v, y_train_w2v, y_test_w2v = train_test_split(X_temp, y_temp, test_size=0.2, random_state=4)\n",
    "X_train_w2v, X_cv_w2v, y_train_w2v, y_cv_w2v = train_test_split(X_train_w2v, y_train_w2v, test_size=0.25, random_state=4)\n",
    "\n",
    "i=0\n",
    "list_of_sentance_test=[]\n",
    "for sentance in X_test_w2v:\n",
    "    list_of_sentance_test.append(sentance.split())\n",
    "\n",
    "i=0\n",
    "list_of_sentance_cv=[]\n",
    "for sentance in X_cv_w2v:\n",
    "    list_of_sentance_cv.append(sentance.split())\n",
    "\n",
    "i=0\n",
    "list_of_sentance=[]\n",
    "for sentance in X_train_w2v:\n",
    "    list_of_sentance.append(sentance.split())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 213,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[('excellent', 0.8336389064788818), ('good', 0.8065310716629028), ('wonderful', 0.7779756784439087), ('perfect', 0.7609016299247742), ('quick', 0.735637903213501), ('unbeatable', 0.7200546264648438), ('awesome', 0.714890718460083), ('makes', 0.7064382433891296), ('aveda', 0.6980257034301758), ('fraction', 0.6970323324203491)]\n",
      "==================================================\n",
      "[('hands', 0.9646154642105103), ('absolute', 0.9634360671043396), ('britt', 0.9538313150405884), ('varieties', 0.9519646763801575), ('world', 0.9514032602310181), ('bahlsen', 0.9454204440116882), ('baklava', 0.9452576041221619), ('experienced', 0.9410791397094727), ('hooked', 0.938496470451355), ('crystallized', 0.9363467693328857)]\n"
     ]
    }
   ],
   "source": [
    "# Using Google News Word2Vectors\n",
    "\n",
    "# in this project we are using a pretrained model by google\n",
    "# its 3.3G file, once you load this into your memory \n",
    "# it occupies ~9Gb, so please do this step only if you have >12G of ram\n",
    "# we will provide a pickle file wich contains a dict , \n",
    "# and it contains all our courpus words as keys and  model[word] as values\n",
    "# To use this code-snippet, download \"GoogleNews-vectors-negative300.bin\" \n",
    "# from https://drive.google.com/file/d/0B7XkCwpI5KDYNlNUTTlSS21pQmM/edit\n",
    "# it's 1.9GB in size.\n",
    "\n",
    "\n",
    "# http://kavita-ganesan.com/gensim-word2vec-tutorial-starter-code/#.W17SRFAzZPY\n",
    "# you can comment this whole cell\n",
    "# or change these varible according to your need\n",
    "\n",
    "is_your_ram_gt_16g=False\n",
    "want_to_use_google_w2v = False\n",
    "want_to_train_w2v = True\n",
    "\n",
    "if want_to_train_w2v:\n",
    "    # min_count = 5 considers only words that occured atleast 5 times\n",
    "    w2v_model=Word2Vec(list_of_sentance,min_count=5,size=50, workers=4)\n",
    "    print(w2v_model.wv.most_similar('great'))\n",
    "    print('='*50)\n",
    "    print(w2v_model.wv.most_similar('worst'))\n",
    "    \n",
    "elif want_to_use_google_w2v and is_your_ram_gt_16g:\n",
    "    if os.path.isfile('GoogleNews-vectors-negative300.bin'):\n",
    "        w2v_model=KeyedVectors.load_word2vec_format('GoogleNews-vectors-negative300.bin', binary=True)\n",
    "        print(w2v_model.wv.most_similar('great'))\n",
    "        print(w2v_model.wv.most_similar('worst'))\n",
    "    else:\n",
    "        print(\"you don't have gogole's word2vec file, keep want_to_train_w2v = True, to train your own w2v \")\n",
    "# print(\"\\n\"+\"#\"*60)\n",
    "# print(len(X_train_w2v),len(y_train_w2v))\n",
    "# print(len(X_cv_w2v),len(y_cv_w2v))\n",
    "# print(len(X_test_w2v),len(y_test_w2v))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 214,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "number of words that occured minimum 5 times  6666\n",
      "sample words  ['since', 'discovering', 'yogi', 'teas', 'years', 'ago', 'enjoyed', 'rich', 'aroma', 'tastes', 'flavors', 'well', 'fact', 'de', 'caf', 'love', 'peach', 'tea', 'general', 'definitely', 'favorite', 'flavor', 'herbal', 'companies', 'vanilla', 'almond', 'also', 'basket', 'drink', 'everyday', 'mix', 'good', 'cant', 'beat', 'warm', 'ice', 'cream', 'tasty', 'happy', 'see', 'could', 'order', 'product', 'online', 'shipped', 'home', 'difficult', 'find', 'stores', 'stuff']\n"
     ]
    }
   ],
   "source": [
    "w2v_words = list(w2v_model.wv.vocab)\n",
    "print(\"number of words that occured minimum 5 times \",len(w2v_words))\n",
    "print(\"sample words \", w2v_words[0:50])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### [5.2.3.2] Converting text into vectors using Avg W2V, TFIDF-W2V for 20k data-points"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####  Avg W2v\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 215,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████| 12000/12000 [00:12<00:00, 959.36it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "12000\n",
      "50\n"
     ]
    }
   ],
   "source": [
    "\n",
    "\n",
    "# TRAINING W2V DATA VECTORS\n",
    "# average Word2Vec\n",
    "# compute average word2vec for each review.\n",
    "sent_vectors_train = []; # the avg-w2v for each sentence/review is stored in this list\n",
    "for sent in tqdm(list_of_sentance): # for each review/sentence\n",
    "    sent_vec = np.zeros(50) # as word vectors are of zero length 50, you might need to change this to 300 if you use google's w2v\n",
    "    cnt_words =0; # num of words with a valid vector in the sentence/review\n",
    "    for word in sent: # for each word in a review/sentence\n",
    "        if word in w2v_words:\n",
    "            vec = w2v_model.wv[word]\n",
    "            sent_vec += vec\n",
    "            cnt_words += 1\n",
    "    if cnt_words != 0:\n",
    "        sent_vec /= cnt_words\n",
    "    sent_vectors_train.append(sent_vec)\n",
    "print(len(sent_vectors_train))\n",
    "print(len(sent_vectors_train[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 216,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|█████████████████████████████████████████████████████████████████████████████| 4000/4000 [00:04<00:00, 917.98it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4000 4000\n",
      "50\n"
     ]
    }
   ],
   "source": [
    "# CV W2V DATA VECTORS\n",
    "\n",
    "sent_vectors_cv = []; # the avg-w2v for each sentence/review is stored in this list\n",
    "for sent in tqdm(list_of_sentance_cv): # for each review/sentence\n",
    "    sent_vec = np.zeros(50) # as word vectors are of zero length 50, you might need to change this to 300 if you use google's w2v\n",
    "    cnt_words =0; # num of words with a valid vector in the sentence/review\n",
    "    for word in sent: # for each word in a review/sentence\n",
    "        if word in w2v_words:\n",
    "            vec = w2v_model.wv[word]\n",
    "            sent_vec += vec\n",
    "            cnt_words += 1\n",
    "    if cnt_words != 0:\n",
    "        sent_vec /= cnt_words\n",
    "    sent_vectors_cv.append(sent_vec)\n",
    "print(len(sent_vectors_cv), len(y_cv_w2v))\n",
    "print(len(sent_vectors_cv[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 217,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|█████████████████████████████████████████████████████████████████████████████| 4000/4000 [00:04<00:00, 924.60it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4000\n",
      "50\n"
     ]
    }
   ],
   "source": [
    "# TEST W2V DATA VECTORS\n",
    "\n",
    "sent_vectors_test = []; # the avg-w2v for each sentence/review is stored in this list\n",
    "for sent in tqdm(list_of_sentance_test): # for each review/sentence\n",
    "    sent_vec = np.zeros(50) # as word vectors are of zero length 50, you might need to change this to 300 if you use google's w2v\n",
    "    cnt_words =0; # num of words with a valid vector in the sentence/review\n",
    "    for word in sent: # for each word in a review/sentence\n",
    "        if word in w2v_words:\n",
    "            vec = w2v_model.wv[word]\n",
    "            sent_vec += vec\n",
    "            cnt_words += 1\n",
    "    if cnt_words != 0:\n",
    "        sent_vec /= cnt_words\n",
    "    sent_vectors_test.append(sent_vec)\n",
    "print(len(sent_vectors_test))\n",
    "print(len(sent_vectors_test[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 223,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████████████| 9/9 [02:15<00:00, 16.44s/it]\n"
     ]
    }
   ],
   "source": [
    "# Please write all the code with proper documentation\n",
    "# KNN \n",
    "X_train_avgw2v, X_cv_avgw2v, X_test_avgw2v = sent_vectors_train, sent_vectors_cv, sent_vectors_test\n",
    "auc_train=[]\n",
    "ks_train=[]\n",
    "auc_cv=[]\n",
    "ks_cv=[]\n",
    "for i in range(1,50,6):\n",
    "    knn = KNeighborsClassifier(n_neighbors=i, algorithm='kd_tree') #try distance cosine too\n",
    "    knn.fit(X_train_avgw2v, y_train_w2v)\n",
    "    # y_pred = knn.predict(X_test_avgw2v)\n",
    "    proba = knn.predict_proba(X_train_avgw2v)[:,1]\n",
    "    auc_train.append(roc_auc_score(y_train_w2v, proba))\n",
    "    ks_train.append(i)\n",
    "    #############################################################\n",
    "    proba = knn.predict_proba(X_cv_avgw2v)[:,1]\n",
    "    auc_cv.append(roc_auc_score(y_cv_w2v, proba))\n",
    "    ks_cv.append(i)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 224,
   "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"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best K for KNN Model from cv_data is:  49 for cooresponding AUC value of: 0.7961835653161907\n"
     ]
    }
   ],
   "source": [
    "\n",
    "######## AUC PLOTTING ########\n",
    "ax=sns.lineplot(ks_train,auc_train)\n",
    "ax=sns.lineplot(ks_cv,auc_cv)\n",
    "ax.set(xlabel='Differnt K values',ylabel='roc_auc_score')\n",
    "ax.set_title('AUC Score vs Different K value Curve')\n",
    "plt.show()\n",
    "\n",
    "######## best_k ##############\n",
    "best_k = ks_cv[auc_cv.index(max(auc_cv))]\n",
    "print(\"Best K for KNN Model from cv_data is: \",best_k, \"for cooresponding AUC value of:\",max(auc_cv))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 225,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUC score on test_data with best_k:  0.8151342884391001\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "######### FRESH-MODEL FOR BEST_K ######################\n",
    "knn = KNeighborsClassifier(n_neighbors=best_k,algorithm='brute')\n",
    "knn.fit(X_train_avgw2v, y_train_w2v)\n",
    "y_pred = knn.predict(X_test_avgw2v)\n",
    "proba = knn.predict_proba(X_test_avgw2v)[:,1]\n",
    "\n",
    "\n",
    "fpr, tpr, threshold = roc_curve(y_test_w2v, proba)\n",
    "roc_auc = auc(fpr, tpr)\n",
    "print(\"AUC score on test_data with best_k: \",roc_auc)\n",
    "#############################################\n",
    "proba = knn.predict_proba(X_train_avgw2v)[:,1]\n",
    "fpr_train, tpr_train, threshold_train = roc_curve(y_train_w2v, proba)\n",
    "roc_auc_train = auc(fpr_train, tpr_train)\n",
    "\n",
    "k_7 = best_k\n",
    "set_7_auc = roc_auc\n",
    "############# ROC ######################\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.plot(fpr_train, tpr_train, 'g', label = 'AUC_train = %0.3f' % roc_auc_train)\n",
    "plt.plot(fpr, tpr, 'b', label = 'AUC_test = %0.3f' % roc_auc)\n",
    "plt.legend(loc = 'lower right')\n",
    "plt.plot([0, 1], [0, 1],'r--')\n",
    "plt.xlim([0, 1])\n",
    "plt.ylim([0, 1])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.title('ROC Curve of kNN')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 226,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "########## CONFUSION MATRIX ###########\n",
    "\n",
    "y_true = y_test_w2v\n",
    "y_pred = y_pred\n",
    "\n",
    "array = confusion_matrix(y_true, y_pred)\n",
    "\n",
    "df_cm = pd.DataFrame(array, index = [i for i in \"01\"],\n",
    "                  columns = [i for i in \"01\"])\n",
    "\n",
    "\n",
    "plt.figure(figsize = (5,5))\n",
    "sns.heatmap(df_cm, annot=True,annot_kws={\"size\": 16}, linewidths=6.3, linecolor='pink', cbar=False, square=True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### [5.2.4] Applying KNN kd-tree on TFIDF W2V,<font color='red'> SET 4</font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####  TFIDF weighted W2v"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 227,
   "metadata": {},
   "outputs": [],
   "source": [
    "# S = [\"abc def pqr\", \"def def def abc\", \"pqr pqr def\"]\n",
    "model = TfidfVectorizer()\n",
    "tf_idf_matrix = model.fit_transform(X_train_w2v)\n",
    "# we are converting a dictionary with word as a key, and the idf as a value\n",
    "dictionary = dict(zip(model.get_feature_names(), list(model.idf_)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 228,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████| 12000/12000 [02:23<00:00, 83.46it/s]\n"
     ]
    }
   ],
   "source": [
    "# TF-IDF weighted Word2Vec\n",
    "tfidf_feat = model.get_feature_names() # tfidf words/col-names\n",
    "# final_tf_idf is the sparse matrix with row= sentence, col=word and cell_val = tfidf\n",
    "\n",
    "tfidf_sent_vectors_train = []; # the tfidf-w2v for each sentence/review is stored in this list\n",
    "row=0;\n",
    "for sent in tqdm(list_of_sentance): # for each review/sentence \n",
    "    sent_vec = np.zeros(50) # as word vectors are of zero length\n",
    "    weight_sum =0; # num of words with a valid vector in the sentence/review\n",
    "    for word in sent: # for each word in a review/sentence\n",
    "        if word in w2v_words and word in tfidf_feat:\n",
    "            vec = w2v_model.wv[word]\n",
    "#             tf_idf = tf_idf_matrix[row, tfidf_feat.index(word)]\n",
    "            # to reduce the computation we are \n",
    "            # dictionary[word] = idf value of word in whole courpus\n",
    "            # sent.count(word) = tf valeus of word in this review\n",
    "            tf_idf = dictionary[word]*(sent.count(word)/len(sent))\n",
    "            sent_vec += (vec * tf_idf)\n",
    "            weight_sum += tf_idf\n",
    "    if weight_sum != 0:\n",
    "        sent_vec /= weight_sum\n",
    "    tfidf_sent_vectors_train.append(sent_vec)\n",
    "    row += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 229,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████████████████████████████████████████████| 4000/4000 [00:48<00:00, 62.06it/s]\n"
     ]
    }
   ],
   "source": [
    "\n",
    "tfidf_sent_vectors_cv = []; # the tfidf-w2v for each sentence/review is stored in this list\n",
    "row=0;\n",
    "for sent in tqdm(list_of_sentance_cv): # for each review/sentence \n",
    "    sent_vec = np.zeros(50) # as word vectors are of zero length\n",
    "    weight_sum =0; # num of words with a valid vector in the sentence/review\n",
    "    for word in sent: # for each word in a review/sentence\n",
    "        if word in w2v_words and word in tfidf_feat:\n",
    "            vec = w2v_model.wv[word]\n",
    "#             tf_idf = tf_idf_matrix[row, tfidf_feat.index(word)]\n",
    "            # to reduce the computation we are \n",
    "            # dictionary[word] = idf value of word in whole courpus\n",
    "            # sent.count(word) = tf valeus of word in this review\n",
    "            tf_idf = dictionary[word]*(sent.count(word)/len(sent))\n",
    "            sent_vec += (vec * tf_idf)\n",
    "            weight_sum += tf_idf\n",
    "    if weight_sum != 0:\n",
    "        sent_vec /= weight_sum\n",
    "    tfidf_sent_vectors_cv.append(sent_vec)\n",
    "    row += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 230,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████████████████████████████████████████████| 4000/4000 [00:46<00:00, 85.37it/s]\n"
     ]
    }
   ],
   "source": [
    "\n",
    "tfidf_sent_vectors_test = []; # the tfidf-w2v for each sentence/review is stored in this list\n",
    "row=0;\n",
    "for sent in tqdm(list_of_sentance_test): # for each review/sentence \n",
    "    sent_vec = np.zeros(50) # as word vectors are of zero length\n",
    "    weight_sum =0; # num of words with a valid vector in the sentence/review\n",
    "    for word in sent: # for each word in a review/sentence\n",
    "        if word in w2v_words and word in tfidf_feat:\n",
    "            vec = w2v_model.wv[word]\n",
    "#             tf_idf = tf_idf_matrix[row, tfidf_feat.index(word)]\n",
    "            # to reduce the computation we are \n",
    "            # dictionary[word] = idf value of word in whole courpus\n",
    "            # sent.count(word) = tf valeus of word in this review\n",
    "            tf_idf = dictionary[word]*(sent.count(word)/len(sent))\n",
    "            sent_vec += (vec * tf_idf)\n",
    "            weight_sum += tf_idf\n",
    "    if weight_sum != 0:\n",
    "        sent_vec /= weight_sum\n",
    "    tfidf_sent_vectors_test.append(sent_vec)\n",
    "    row += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 231,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Please write all the code with proper documentation\n",
    "\n",
    "X_train_tfw2v, X_cv_w2v, X_test_tfw2v = tfidf_sent_vectors_train, tfidf_sent_vectors_cv, tfidf_sent_vectors_test\n",
    "\n",
    "auc_train=[]\n",
    "ks_train=[]\n",
    "auc_cv=[]\n",
    "ks_cv=[]\n",
    "\n",
    "for i in range(1,50,4):\n",
    "    knn = KNeighborsClassifier(n_neighbors=i, algorithm='kd_tree') #try distance cosine too\n",
    "    knn.fit(X_train_tfw2v, y_train_w2v)\n",
    "    # y_pred = knn.predict(X_test_tfw2v)\n",
    "    proba = knn.predict_proba(X_train_tfw2v)[:,1]\n",
    "    auc_train.append(roc_auc_score(y_train_w2v, proba))\n",
    "    ks_train.append(i)\n",
    "    #############################################################\n",
    "    proba = knn.predict_proba(X_cv_w2v)[:,1]\n",
    "    auc_cv.append(roc_auc_score(y_cv_w2v, proba))\n",
    "    ks_cv.append(i)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 232,
   "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"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best K for KNN Model from cv_data is:  49 for cooresponding AUC value of: 0.761159042515019\n"
     ]
    }
   ],
   "source": [
    "######## AUC PLOTTING ########\n",
    "ax=sns.lineplot(ks_train,auc_train)\n",
    "ax=sns.lineplot(ks_cv,auc_cv)\n",
    "ax.set(xlabel='Differnt K values',ylabel='roc_auc_score')\n",
    "ax.set_title('AUC Score vs Different K value Curve')\n",
    "plt.show()\n",
    "\n",
    "######## best_k ##############A\n",
    "best_k = ks_cv[auc_cv.index(max(auc_cv))]\n",
    "print(\"Best K for KNN Model from cv_data is: \",best_k, \"for cooresponding AUC value of:\",max(auc_cv))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 233,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUC score on test_data with best_k:  0.7685493509791699\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "\n",
    "######### FRESH-MODEL FOR BEST_K ######################\n",
    "knn = KNeighborsClassifier(n_neighbors=best_k,algorithm='kd_tree')\n",
    "knn.fit(X_train_tfw2v, y_train_w2v)\n",
    "y_pred = knn.predict(X_test_tfw2v)\n",
    "proba = knn.predict_proba(X_test_tfw2v)[:,1]\n",
    "\n",
    "\n",
    "fpr, tpr, threshold = roc_curve(y_test_w2v, proba)\n",
    "roc_auc = auc(fpr, tpr)\n",
    "print(\"AUC score on test_data with best_k: \",roc_auc)\n",
    "#############################################\n",
    "proba = knn.predict_proba(X_train_tfw2v)[:,1]\n",
    "fpr_train, tpr_train, threshold_train = roc_curve(y_train_w2v, proba)\n",
    "roc_auc_train = auc(fpr_train, tpr_train)\n",
    "\n",
    "k_8 = best_k\n",
    "set_8_auc = roc_auc\n",
    "############# ROC ######################\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.plot(fpr_train, tpr_train, 'g', label = 'AUC_train = %0.3f' % roc_auc_train)\n",
    "plt.plot(fpr, tpr, 'b', label = 'AUC_test = %0.3f' % roc_auc)\n",
    "plt.legend(loc = 'lower right')\n",
    "plt.plot([0, 1], [0, 1],'r--')\n",
    "plt.xlim([0, 1])\n",
    "plt.ylim([0, 1])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.title('ROC Curve of kNN')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 234,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "\n",
    "y_true = y_test_w2v\n",
    "y_pred = y_pred\n",
    "\n",
    "from sklearn.metrics import confusion_matrix\n",
    "\n",
    "array = confusion_matrix(y_true, y_pred)\n",
    "\n",
    "df_cm = pd.DataFrame(array, index = [i for i in \"01\"],\n",
    "                  columns = [i for i in \"01\"])\n",
    "\n",
    "\n",
    "plt.figure(figsize = (5,5))\n",
    "sns.heatmap(df_cm, annot=True,annot_kws={\"size\": 16}, linewidths=6.3, linecolor='pink', cbar=False, square=True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# [6] Conclusions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 245,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------------+---------+-----------------+------+\n",
      "| Vectoriser |  Model  | Hyper Parameter | AUC  |\n",
      "+------------+---------+-----------------+------+\n",
      "|    BOW     |  Brute  |        45       | 0.84 |\n",
      "|   TFIDF    |  Brute  |        49       | 0.87 |\n",
      "|  AVG-W2V   |  Brute  |        49       | 0.87 |\n",
      "| TFIDF W2V  |  Brute  |        45       | 0.85 |\n",
      "|    ----    |   ----  |       ----      | ---- |\n",
      "|    BOW     | KD-Tree |        49       | 0.68 |\n",
      "|   TFIDF    | KD-Tree |        49       | 0.86 |\n",
      "|  AVG-W2V   | KD-Tree |        49       | 0.82 |\n",
      "| TFIDF W2V  | KD-Tree |        49       | 0.77 |\n",
      "+------------+---------+-----------------+------+\n"
     ]
    }
   ],
   "source": [
    "# Please compare all your models using Prettytable library\n",
    "# Reference: http://zetcode.com/python/prettytable/\n",
    "\n",
    "from prettytable import PrettyTable\n",
    "x= PrettyTable()\n",
    "\n",
    "x.field_names = [\"Vectoriser\", \"Model\", \"Hyper Parameter\", \"AUC\"]\n",
    "\n",
    "x.add_row([\"BOW\", \"Brute\", k_1, \"{0:.2f}\".format(set_1_auc)])\n",
    "x.add_row([\"TFIDF\", \"Brute\", k_2, \"{0:.2f}\".format(set_2_auc)])\n",
    "x.add_row([\"AVG-W2V\", \"Brute\", k_3, \"{0:.2f}\".format(set_3_auc)])\n",
    "x.add_row([\"TFIDF W2V\", \"Brute\", k_4, \"{0:.2f}\".format(set_4_auc)])\n",
    "x.add_row([\"----\", \"----\", \"----\", \"----\"])\n",
    "x.add_row([\"BOW\", \"KD-Tree\", k_5, \"{0:.2f}\".format(set_5_auc)])\n",
    "x.add_row([\"TFIDF\", \"KD-Tree\", k_6, \"{0:.2f}\".format(set_6_auc)])\n",
    "x.add_row([\"AVG-W2V\", \"KD-Tree\", k_7, \"{0:.2f}\".format(set_7_auc)])\n",
    "x.add_row([\"TFIDF W2V\", \"KD-Tree\", k_8, \"{0:.2f}\".format(set_8_auc)])\n",
    "\n",
    "print(x)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* It is easy to depict from above table that for all four vectorisers Brute force method of KNN proved to be more accurate, providing maximum auc-values.\n",
    "\n",
    "\n",
    "* AUC curves are showing decent graphs for brute force but there is too much fluctutations for kd-tree algorithm specially for Avg-w2v vector auc graph is too ambiguous.\n",
    "\n",
    "\n",
    "* tf-idf and Avg-w2v brute force algorithm are holding the best accuracy metric: auc = 0.87\n",
    "\n",
    "\n",
    "* It is observed that hyper parameter for all algorithms are not differeing by large value."
   ]
  }
 ],
 "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": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}