Tweet_Sentiment_Analyzer_using_CountVectorizer_And_SVMClassifier.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# IMPORTS"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import re\n",
    "import string\n",
    "import emoji\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from textblob import TextBlob\n",
    "from nltk.corpus import stopwords\n",
    "from nltk.tokenize import word_tokenize\n",
    "from nltk.stem import WordNetLemmatizer\n",
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "from sklearn import svm\n",
    "from sklearn.svm import LinearSVC\n",
    "from sklearn.multiclass import OneVsRestClassifier\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.metrics import confusion_matrix \n",
    "from sklearn.metrics import precision_score\n",
    "from sklearn.metrics import recall_score\n",
    "from sklearn.metrics import f1_score\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Settings for better readabilty and shuting down tensorflow warnings\n",
    "     \n",
    "    Run cell below only once."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "pd.set_option('display.max_colwidth', -1) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Load And analyse the dataset we created by extracting tweets from twitter using Tweepy API"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "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>Date- Time</th>\n",
       "      <th>User_Name</th>\n",
       "      <th>Tweet</th>\n",
       "      <th>Sentiment</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1286896213657346050</td>\n",
       "      <td>2020-07-25 05:29:01</td>\n",
       "      <td>BALLl</td>\n",
       "      <td>i need to make my own SapceX, some people dont deserve to live here</td>\n",
       "      <td>positive</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1286905185273610241</td>\n",
       "      <td>2020-07-25 06:04:40</td>\n",
       "      <td>T:HE:JAMES. HMRH</td>\n",
       "      <td>Thre was mention of this earlier, the infraction on the views of the cosmos from the SapceX array and how astronomy would compensate. https://t.co/I5omeFq4SM</td>\n",
       "      <td>neutral</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1287391022218006531</td>\n",
       "      <td>2020-07-26 14:15:12</td>\n",
       "      <td>Anshi Pathak</td>\n",
       "      <td>@bhab005 @KanganaTeam @kunalpande8 Sachine Tendulkar handle his account and but here kangana team handles her. That's the difference here. And if she is doing that what's funny. She has earned that right. Also Jo log khud nahi bolte h wo dusron se bulwate Hain. Wo khud bol rahi h to galat. Kya h.</td>\n",
       "      <td>positive</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1287495418180501505</td>\n",
       "      <td>2020-07-26 21:10:02</td>\n",
       "      <td>Goodauld69th</td>\n",
       "      <td>Mainly stoltenberg scandanavian cowardly attacks oil pipe lines then running like weird inbred fish savages to underground bunkers i wont dive into it but that world is over and like nazis there being spitefull and abusive and fucking everything up because there dumb inbreds</td>\n",
       "      <td>negative</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1287731665452769280</td>\n",
       "      <td>2020-07-27 12:48:48</td>\n",
       "      <td>Sardar Khan</td>\n",
       "      <td>@gareebscientist @GauravG91 @Chandel1402 @isro @SpaceX He I mean... Vaibhav....spacex is doing great work.. But its unfair to compare two.. No offense to sapcex but I will chose ISRO over spacex AnyDay....</td>\n",
       "      <td>negative</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>1287756489428353025</td>\n",
       "      <td>2020-07-27 14:27:27</td>\n",
       "      <td>guapdaddy 3000</td>\n",
       "      <td>someone get me the panda pegion nike sb</td>\n",
       "      <td>neutral</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>1287756655413911554</td>\n",
       "      <td>2020-07-27 14:28:06</td>\n",
       "      <td>who?????me</td>\n",
       "      <td>Piers Morgan, A Scum Media \"Star\" Showing His Spitefull Side. What a Pathetic Little Man. https://t.co/fKIRzd7aGE</td>\n",
       "      <td>negative</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>1287773079675056128</td>\n",
       "      <td>2020-07-27 15:33:22</td>\n",
       "      <td>The Great Gay Trainwreck</td>\n",
       "      <td>Old PC game skyboxes emit such elegiac auras\\njust look at like\\nThe skybox from the Harry potter 2 qudditch area</td>\n",
       "      <td>negative</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>1287775000041271296</td>\n",
       "      <td>2020-07-27 15:41:00</td>\n",
       "      <td>Dawood Javed</td>\n",
       "      <td>\"Fantastic Beasts: The Crimes of Grindelwald, Gellert Grindelwald plans to raise an army of wizards to rule over non-magical beings. In response, Newt Scamander's former professor, Albus Dumbledore, seeks his help to stop him.\"\\n#FantasticBeastsTheCrimesOfGrindelwald https://t.co/KmVVV9z7FB</td>\n",
       "      <td>positive</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>1287808107817689090</td>\n",
       "      <td>2020-07-27 17:52:33</td>\n",
       "      <td>Xavier Dolanigga</td>\n",
       "      <td>@Selinakyle981 @filmsbyzoe also Kit Harrigton is good in this movie. I don't care what anybody says.</td>\n",
       "      <td>positive</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                    ID           Date- Time                 User_Name  \\\n",
       "0  1286896213657346050  2020-07-25 05:29:01  BALLl                      \n",
       "1  1286905185273610241  2020-07-25 06:04:40  T:HE:JAMES. HMRH           \n",
       "2  1287391022218006531  2020-07-26 14:15:12  Anshi Pathak               \n",
       "3  1287495418180501505  2020-07-26 21:10:02  Goodauld69th               \n",
       "4  1287731665452769280  2020-07-27 12:48:48  Sardar Khan                \n",
       "5  1287756489428353025  2020-07-27 14:27:27  guapdaddy 3000             \n",
       "6  1287756655413911554  2020-07-27 14:28:06  who?????me                 \n",
       "7  1287773079675056128  2020-07-27 15:33:22  The Great Gay Trainwreck   \n",
       "8  1287775000041271296  2020-07-27 15:41:00  Dawood Javed               \n",
       "9  1287808107817689090  2020-07-27 17:52:33  Xavier Dolanigga           \n",
       "\n",
       "                                                                                                                                                                                                                                                                                                       Tweet  \\\n",
       "0  i need to make my own SapceX, some people dont deserve to live here                                                                                                                                                                                                                                         \n",
       "1  Thre was mention of this earlier, the infraction on the views of the cosmos from the SapceX array and how astronomy would compensate. https://t.co/I5omeFq4SM                                                                                                                                               \n",
       "2  @bhab005 @KanganaTeam @kunalpande8 Sachine Tendulkar handle his account and but here kangana team handles her. That's the difference here. And if she is doing that what's funny. She has earned that right. Also Jo log khud nahi bolte h wo dusron se bulwate Hain. Wo khud bol rahi h to galat. Kya h.   \n",
       "3  Mainly stoltenberg scandanavian cowardly attacks oil pipe lines then running like weird inbred fish savages to underground bunkers i wont dive into it but that world is over and like nazis there being spitefull and abusive and fucking everything up because there dumb inbreds                         \n",
       "4  @gareebscientist @GauravG91 @Chandel1402 @isro @SpaceX He I mean... Vaibhav....spacex is doing great work.. But its unfair to compare two.. No offense to sapcex but I will chose ISRO over spacex AnyDay....                                                                                               \n",
       "5  someone get me the panda pegion nike sb                                                                                                                                                                                                                                                                     \n",
       "6  Piers Morgan, A Scum Media \"Star\" Showing His Spitefull Side. What a Pathetic Little Man. https://t.co/fKIRzd7aGE                                                                                                                                                                                           \n",
       "7  Old PC game skyboxes emit such elegiac auras\\njust look at like\\nThe skybox from the Harry potter 2 qudditch area                                                                                                                                                                                           \n",
       "8  \"Fantastic Beasts: The Crimes of Grindelwald, Gellert Grindelwald plans to raise an army of wizards to rule over non-magical beings. In response, Newt Scamander's former professor, Albus Dumbledore, seeks his help to stop him.\"\\n#FantasticBeastsTheCrimesOfGrindelwald https://t.co/KmVVV9z7FB         \n",
       "9  @Selinakyle981 @filmsbyzoe also Kit Harrigton is good in this movie. I don't care what anybody says.                                                                                                                                                                                                        \n",
       "\n",
       "  Sentiment  \n",
       "0  positive  \n",
       "1  neutral   \n",
       "2  positive  \n",
       "3  negative  \n",
       "4  negative  \n",
       "5  neutral   \n",
       "6  negative  \n",
       "7  negative  \n",
       "8  positive  \n",
       "9  positive  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Tweets = pd.read_csv(\"TweetlabelledFinal.csv\")\n",
    "Tweets.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 29348 entries, 0 to 29347\n",
      "Data columns (total 5 columns):\n",
      "ID            29348 non-null int64\n",
      "Date- Time    29348 non-null object\n",
      "User_Name     29345 non-null object\n",
      "Tweet         29348 non-null object\n",
      "Sentiment     29348 non-null object\n",
      "dtypes: int64(1), object(4)\n",
      "memory usage: 1.1+ MB\n"
     ]
    }
   ],
   "source": [
    "Tweets.info()"
   ]
  },
  {
   "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>ID</th>\n",
       "      <th>Date- Time</th>\n",
       "      <th>User_Name</th>\n",
       "      <th>Tweet</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Sentiment</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>negative</th>\n",
       "      <td>9468</td>\n",
       "      <td>9468</td>\n",
       "      <td>9468</td>\n",
       "      <td>9468</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>neutral</th>\n",
       "      <td>9587</td>\n",
       "      <td>9587</td>\n",
       "      <td>9584</td>\n",
       "      <td>9587</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>positive</th>\n",
       "      <td>10293</td>\n",
       "      <td>10293</td>\n",
       "      <td>10293</td>\n",
       "      <td>10293</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              ID  Date- Time  User_Name  Tweet\n",
       "Sentiment                                     \n",
       "negative   9468   9468        9468       9468 \n",
       "neutral    9587   9587        9584       9587 \n",
       "positive   10293  10293       10293      10293"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Tweets.groupby('Sentiment').count()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "    The dataset contains attributes : Tweet , tweet id , data time , username & sentiment , out of these we only need\n",
    "    tweet and sentiment for our supervised machine learning classification algorithm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "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>Tweet</th>\n",
       "      <th>Sentiment</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>i need to make my own SapceX, some people dont deserve to live here</td>\n",
       "      <td>positive</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Thre was mention of this earlier, the infraction on the views of the cosmos from the SapceX array and how astronomy would compensate. https://t.co/I5omeFq4SM</td>\n",
       "      <td>neutral</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>@bhab005 @KanganaTeam @kunalpande8 Sachine Tendulkar handle his account and but here kangana team handles her. That's the difference here. And if she is doing that what's funny. She has earned that right. Also Jo log khud nahi bolte h wo dusron se bulwate Hain. Wo khud bol rahi h to galat. Kya h.</td>\n",
       "      <td>positive</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Mainly stoltenberg scandanavian cowardly attacks oil pipe lines then running like weird inbred fish savages to underground bunkers i wont dive into it but that world is over and like nazis there being spitefull and abusive and fucking everything up because there dumb inbreds</td>\n",
       "      <td>negative</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>@gareebscientist @GauravG91 @Chandel1402 @isro @SpaceX He I mean... Vaibhav....spacex is doing great work.. But its unfair to compare two.. No offense to sapcex but I will chose ISRO over spacex AnyDay....</td>\n",
       "      <td>negative</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>someone get me the panda pegion nike sb</td>\n",
       "      <td>neutral</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>Piers Morgan, A Scum Media \"Star\" Showing His Spitefull Side. What a Pathetic Little Man. https://t.co/fKIRzd7aGE</td>\n",
       "      <td>negative</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Old PC game skyboxes emit such elegiac auras\\njust look at like\\nThe skybox from the Harry potter 2 qudditch area</td>\n",
       "      <td>negative</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>\"Fantastic Beasts: The Crimes of Grindelwald, Gellert Grindelwald plans to raise an army of wizards to rule over non-magical beings. In response, Newt Scamander's former professor, Albus Dumbledore, seeks his help to stop him.\"\\n#FantasticBeastsTheCrimesOfGrindelwald https://t.co/KmVVV9z7FB</td>\n",
       "      <td>positive</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>@Selinakyle981 @filmsbyzoe also Kit Harrigton is good in this movie. I don't care what anybody says.</td>\n",
       "      <td>positive</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                                                                                                                                                                                                                                                       Tweet  \\\n",
       "0  i need to make my own SapceX, some people dont deserve to live here                                                                                                                                                                                                                                         \n",
       "1  Thre was mention of this earlier, the infraction on the views of the cosmos from the SapceX array and how astronomy would compensate. https://t.co/I5omeFq4SM                                                                                                                                               \n",
       "2  @bhab005 @KanganaTeam @kunalpande8 Sachine Tendulkar handle his account and but here kangana team handles her. That's the difference here. And if she is doing that what's funny. She has earned that right. Also Jo log khud nahi bolte h wo dusron se bulwate Hain. Wo khud bol rahi h to galat. Kya h.   \n",
       "3  Mainly stoltenberg scandanavian cowardly attacks oil pipe lines then running like weird inbred fish savages to underground bunkers i wont dive into it but that world is over and like nazis there being spitefull and abusive and fucking everything up because there dumb inbreds                         \n",
       "4  @gareebscientist @GauravG91 @Chandel1402 @isro @SpaceX He I mean... Vaibhav....spacex is doing great work.. But its unfair to compare two.. No offense to sapcex but I will chose ISRO over spacex AnyDay....                                                                                               \n",
       "5  someone get me the panda pegion nike sb                                                                                                                                                                                                                                                                     \n",
       "6  Piers Morgan, A Scum Media \"Star\" Showing His Spitefull Side. What a Pathetic Little Man. https://t.co/fKIRzd7aGE                                                                                                                                                                                           \n",
       "7  Old PC game skyboxes emit such elegiac auras\\njust look at like\\nThe skybox from the Harry potter 2 qudditch area                                                                                                                                                                                           \n",
       "8  \"Fantastic Beasts: The Crimes of Grindelwald, Gellert Grindelwald plans to raise an army of wizards to rule over non-magical beings. In response, Newt Scamander's former professor, Albus Dumbledore, seeks his help to stop him.\"\\n#FantasticBeastsTheCrimesOfGrindelwald https://t.co/KmVVV9z7FB         \n",
       "9  @Selinakyle981 @filmsbyzoe also Kit Harrigton is good in this movie. I don't care what anybody says.                                                                                                                                                                                                        \n",
       "\n",
       "  Sentiment  \n",
       "0  positive  \n",
       "1  neutral   \n",
       "2  positive  \n",
       "3  negative  \n",
       "4  negative  \n",
       "5  neutral   \n",
       "6  negative  \n",
       "7  negative  \n",
       "8  positive  \n",
       "9  positive  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Tweets = Tweets[['Tweet','Sentiment']]\n",
    "Tweets.head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "    As the sentiment is given in string format we need to convert them into numerical values which will serve as the\n",
    "    class labels for the supervised machine learning algorithm."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "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>Tweet</th>\n",
       "      <th>Sentiment</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>i need to make my own SapceX, some people dont deserve to live here</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Thre was mention of this earlier, the infraction on the views of the cosmos from the SapceX array and how astronomy would compensate. https://t.co/I5omeFq4SM</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>@bhab005 @KanganaTeam @kunalpande8 Sachine Tendulkar handle his account and but here kangana team handles her. That's the difference here. And if she is doing that what's funny. She has earned that right. Also Jo log khud nahi bolte h wo dusron se bulwate Hain. Wo khud bol rahi h to galat. Kya h.</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Mainly stoltenberg scandanavian cowardly attacks oil pipe lines then running like weird inbred fish savages to underground bunkers i wont dive into it but that world is over and like nazis there being spitefull and abusive and fucking everything up because there dumb inbreds</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>@gareebscientist @GauravG91 @Chandel1402 @isro @SpaceX He I mean... Vaibhav....spacex is doing great work.. But its unfair to compare two.. No offense to sapcex but I will chose ISRO over spacex AnyDay....</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>someone get me the panda pegion nike sb</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>Piers Morgan, A Scum Media \"Star\" Showing His Spitefull Side. What a Pathetic Little Man. https://t.co/fKIRzd7aGE</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Old PC game skyboxes emit such elegiac auras\\njust look at like\\nThe skybox from the Harry potter 2 qudditch area</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>\"Fantastic Beasts: The Crimes of Grindelwald, Gellert Grindelwald plans to raise an army of wizards to rule over non-magical beings. In response, Newt Scamander's former professor, Albus Dumbledore, seeks his help to stop him.\"\\n#FantasticBeastsTheCrimesOfGrindelwald https://t.co/KmVVV9z7FB</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>@Selinakyle981 @filmsbyzoe also Kit Harrigton is good in this movie. I don't care what anybody says.</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                                                                                                                                                                                                                                                       Tweet  \\\n",
       "0  i need to make my own SapceX, some people dont deserve to live here                                                                                                                                                                                                                                         \n",
       "1  Thre was mention of this earlier, the infraction on the views of the cosmos from the SapceX array and how astronomy would compensate. https://t.co/I5omeFq4SM                                                                                                                                               \n",
       "2  @bhab005 @KanganaTeam @kunalpande8 Sachine Tendulkar handle his account and but here kangana team handles her. That's the difference here. And if she is doing that what's funny. She has earned that right. Also Jo log khud nahi bolte h wo dusron se bulwate Hain. Wo khud bol rahi h to galat. Kya h.   \n",
       "3  Mainly stoltenberg scandanavian cowardly attacks oil pipe lines then running like weird inbred fish savages to underground bunkers i wont dive into it but that world is over and like nazis there being spitefull and abusive and fucking everything up because there dumb inbreds                         \n",
       "4  @gareebscientist @GauravG91 @Chandel1402 @isro @SpaceX He I mean... Vaibhav....spacex is doing great work.. But its unfair to compare two.. No offense to sapcex but I will chose ISRO over spacex AnyDay....                                                                                               \n",
       "5  someone get me the panda pegion nike sb                                                                                                                                                                                                                                                                     \n",
       "6  Piers Morgan, A Scum Media \"Star\" Showing His Spitefull Side. What a Pathetic Little Man. https://t.co/fKIRzd7aGE                                                                                                                                                                                           \n",
       "7  Old PC game skyboxes emit such elegiac auras\\njust look at like\\nThe skybox from the Harry potter 2 qudditch area                                                                                                                                                                                           \n",
       "8  \"Fantastic Beasts: The Crimes of Grindelwald, Gellert Grindelwald plans to raise an army of wizards to rule over non-magical beings. In response, Newt Scamander's former professor, Albus Dumbledore, seeks his help to stop him.\"\\n#FantasticBeastsTheCrimesOfGrindelwald https://t.co/KmVVV9z7FB         \n",
       "9  @Selinakyle981 @filmsbyzoe also Kit Harrigton is good in this movie. I don't care what anybody says.                                                                                                                                                                                                        \n",
       "\n",
       "   Sentiment  \n",
       "0  1          \n",
       "1  2          \n",
       "2  1          \n",
       "3  0          \n",
       "4  0          \n",
       "5  2          \n",
       "6  0          \n",
       "7  0          \n",
       "8  1          \n",
       "9  1          "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Sentiment = {'negative': 0,'positive': 1,'neutral' : 2}\n",
    "Tweets.Sentiment = [Sentiment[item] for item in Tweets.Sentiment]\n",
    "Tweets.head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Cleaning Data\n",
    "\n",
    "        From the tweets we need to remove:\n",
    "        1. mentions or usernames : @abcd\n",
    "        2. Hashtags : #happy --> happy\n",
    "        3. convert all tweets in Lowercase :\n",
    "                this is done because upeercase and lower case charecters are treated as differnt as a result \n",
    "                while converting a word in vector , same words may give different vectors if the casing of the letters\n",
    "                in the words varies.\n",
    "        4. Remove extra white spaces\n",
    "        5. Remove numbers/digits\n",
    "        6. Remove Hyperlinks\n",
    "        7. Remove New Lines\n",
    "        \n",
    "         re.sub(r'@[A-za-z0-9]+','',text) :\n",
    "            r: tells python the text is raw text\n",
    "            @[A-za-z0-9]+ : any pattern starting with @ and followed by letterts or alphabet one or more.\n",
    "            \n",
    "        'http?s:\\/\\/\\S+' : remove http, '?s' means s may or may not be present folowwed by // till we  \n",
    "                               get one or more white space denoted by'S+'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Create a function to Clean the Tweets:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def cleanTxt(text):\n",
    "    text= re.sub(r'@[A-za-z0-9]+','',text) # remove mentions\n",
    "    text= re.sub(r'#','',text) # remove # from text\n",
    "    text= re.sub(r'http?s:\\/\\/\\S+','',text) # remove Hyperlink\n",
    "    text= re.sub(r'\\n',' ',text) # remove new line\n",
    "    text = re.sub(r'\\d+','', text) # remove numbers\n",
    "    \n",
    "    # remove punctuations and convert characters to lower case\n",
    "    text = \"\".join([char.lower() for char in text if char not in string.punctuation]) \n",
    "    # substitute multiple whitespace with single whitespace\n",
    "    # Also, removes leading and trailing whitespaces\n",
    "    text = re.sub('\\s+', ' ', text).strip()\n",
    "    \n",
    "    return text"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Appling CleanTxt to our Dataframe:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Tweet</th>\n",
       "      <th>Sentiment</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>i need to make my own sapcex some people dont deserve to live here</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>thre was mention of this earlier the infraction on the views of the cosmos from the sapcex array and how astronomy would compensate</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>sachine tendulkar handle his account and but here kangana team handles her thats the difference here and if she is doing that whats funny she has earned that right also jo log khud nahi bolte h wo dusron se bulwate hain wo khud bol rahi h to galat kya h</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>mainly stoltenberg scandanavian cowardly attacks oil pipe lines then running like weird inbred fish savages to underground bunkers i wont dive into it but that world is over and like nazis there being spitefull and abusive and fucking everything up because there dumb inbreds</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>he i mean vaibhavspacex is doing great work but its unfair to compare two no offense to sapcex but i will chose isro over spacex anyday</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>someone get me the panda pegion nike sb</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>piers morgan a scum media star showing his spitefull side what a pathetic little man</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>old pc game skyboxes emit such elegiac auras just look at like the skybox from the harry potter qudditch area</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>fantastic beasts the crimes of grindelwald gellert grindelwald plans to raise an army of wizards to rule over nonmagical beings in response newt scamanders former professor albus dumbledore seeks his help to stop him fantasticbeaststhecrimesofgrindelwald</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>also kit harrigton is good in this movie i dont care what anybody says</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>his excellency pegion a tragic loss we wont forget about</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>you think its not a problem damn what a pegion holed murrican</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>pegion</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>jk rowling reveals gellert grindelwald actually had aquamarine eyes short hair and a bossy personality</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>i was most famous for my defeat of gellert grindelwald the discovery of the twelve uses of dragons blood and my work on alchemy with nicolas flamel</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>cbicrucialforssr aditya thakrey why you are blocking the twitter account of everybody who is tweeting you beta why are you so scared as an youth come out from your pegion whole n demand cbi probe for your friend sushant</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>ive read qudditch legend renjun volleyball setter renjun and baseball captain renjun but never basketball player jun 💔</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>pegion</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>i had a white pegion but then he died</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>bro they wont understand 😂 you give them facts they will ignore it its like a game of chess against a pegion 🤡</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>hey aman i found an injured pegion in my parking area and thought i could help it survive can you help me out here</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>im still mad and sad about something very stupid and i will probably still be for a long time because im absolutely spitefull and i cant even tell the people involved because i think i am in the wrong by being childish and egoist smh fuck</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>pegion from phone camera 😅</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>her face says it all spitefull vindictive look that cant be hidden such a shame harry couldnt see this streak in her</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>greater good he leaned closer to her and whispered i am gellert grindelwald</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>pegion residency by universal located at tank hill rd end ahead of the stone quarry with a choice of bedroom bedroom and bedroom apartment starting at m our representatives are here to help julie dorah</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>please my people never compare with sake dancehall is all about partoa and wale speaks pegion not partoa 🧐🧐🧐</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>richard nixon is our gellert grindelwald comeymemo</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>i do not like you and because there all sad pathetic cowardly sneeky jealous spitefull bk stabbing they think your like them your just trying to protect your self from what</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>stupid bitch i wanna squish this and make a pegion sandwich</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>india has blamed pegion for spying for pakistan now they are blaming the infectious disease center for development anthrax like pathogen the article says that sinopharm of china is carrying vaccine trials for coronavirus on pakistan citizens while he ignored tests in uae</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>qudditch is cancelled af though but large amounts of beer is as it so happens intrinsic to the culture and history of morris dancing which is a sport on account of it has teams and sometimes apparently fights</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>democrats are shitty spitefull morons they are truely evil vile human beings after that donkey show today they should all be tossed out on their asses ps jerry nadler is a fat fucking slob</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>absolutely nasty vile wicked spitefull individual</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>whyidontliketrump yesterday agbarr besides the day trump sworn in was saddest ever raceist agbarr chief enabler to raceist trump ill informed nazi trump besides virus were also fighting corrupt hatefull spitefull dr death himself trump</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>we have tickets for aug th pegion forge tn not going to happen is it😢 they still have everything still up on their wed page</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36</th>\n",
       "      <td>yes it sure is one a himalayan green pegion checked my birds of india book before replying 😊</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37</th>\n",
       "      <td>this cruel thesis a spitefull angels litany will soon soar from the window of your destiny</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>38</th>\n",
       "      <td>india is a country who makes its people fool with fake surgical strikes propaganda videos fake pictures pegion and locusts in reality indianarmy always gets thrashed amp humiliated by pakistan and china with online available proves 😁😁😁😁</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39</th>\n",
       "      <td>even a pegion is introduced to a yrs child as pegion n not as kabootar cow as cow instead of gaay or gau instead of tata we r teaching bye instead of क्षमा we teach sorry instead of कृपया we teach please but we will wait for clear opinion till all points r cleared</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                                                                                                                                                                                                                                  Tweet  \\\n",
       "0   i need to make my own sapcex some people dont deserve to live here                                                                                                                                                                                                                    \n",
       "1   thre was mention of this earlier the infraction on the views of the cosmos from the sapcex array and how astronomy would compensate                                                                                                                                                   \n",
       "2   sachine tendulkar handle his account and but here kangana team handles her thats the difference here and if she is doing that whats funny she has earned that right also jo log khud nahi bolte h wo dusron se bulwate hain wo khud bol rahi h to galat kya h                         \n",
       "3   mainly stoltenberg scandanavian cowardly attacks oil pipe lines then running like weird inbred fish savages to underground bunkers i wont dive into it but that world is over and like nazis there being spitefull and abusive and fucking everything up because there dumb inbreds   \n",
       "4   he i mean vaibhavspacex is doing great work but its unfair to compare two no offense to sapcex but i will chose isro over spacex anyday                                                                                                                                               \n",
       "5   someone get me the panda pegion nike sb                                                                                                                                                                                                                                               \n",
       "6   piers morgan a scum media star showing his spitefull side what a pathetic little man                                                                                                                                                                                                  \n",
       "7   old pc game skyboxes emit such elegiac auras just look at like the skybox from the harry potter qudditch area                                                                                                                                                                         \n",
       "8   fantastic beasts the crimes of grindelwald gellert grindelwald plans to raise an army of wizards to rule over nonmagical beings in response newt scamanders former professor albus dumbledore seeks his help to stop him fantasticbeaststhecrimesofgrindelwald                        \n",
       "9   also kit harrigton is good in this movie i dont care what anybody says                                                                                                                                                                                                                \n",
       "10  his excellency pegion a tragic loss we wont forget about                                                                                                                                                                                                                              \n",
       "11  you think its not a problem damn what a pegion holed murrican                                                                                                                                                                                                                         \n",
       "12  pegion                                                                                                                                                                                                                                                                                \n",
       "13  jk rowling reveals gellert grindelwald actually had aquamarine eyes short hair and a bossy personality                                                                                                                                                                                \n",
       "14  i was most famous for my defeat of gellert grindelwald the discovery of the twelve uses of dragons blood and my work on alchemy with nicolas flamel                                                                                                                                   \n",
       "15  cbicrucialforssr aditya thakrey why you are blocking the twitter account of everybody who is tweeting you beta why are you so scared as an youth come out from your pegion whole n demand cbi probe for your friend sushant                                                           \n",
       "16  ive read qudditch legend renjun volleyball setter renjun and baseball captain renjun but never basketball player jun 💔                                                                                                                                                                \n",
       "17  pegion                                                                                                                                                                                                                                                                                \n",
       "18  i had a white pegion but then he died                                                                                                                                                                                                                                                 \n",
       "19  bro they wont understand 😂 you give them facts they will ignore it its like a game of chess against a pegion 🤡                                                                                                                                                                        \n",
       "20  hey aman i found an injured pegion in my parking area and thought i could help it survive can you help me out here                                                                                                                                                                    \n",
       "21  im still mad and sad about something very stupid and i will probably still be for a long time because im absolutely spitefull and i cant even tell the people involved because i think i am in the wrong by being childish and egoist smh fuck                                        \n",
       "22  pegion from phone camera 😅                                                                                                                                                                                                                                                            \n",
       "23  her face says it all spitefull vindictive look that cant be hidden such a shame harry couldnt see this streak in her                                                                                                                                                                  \n",
       "24  greater good he leaned closer to her and whispered i am gellert grindelwald                                                                                                                                                                                                           \n",
       "25  pegion residency by universal located at tank hill rd end ahead of the stone quarry with a choice of bedroom bedroom and bedroom apartment starting at m our representatives are here to help julie dorah                                                                             \n",
       "26  please my people never compare with sake dancehall is all about partoa and wale speaks pegion not partoa 🧐🧐🧐                                                                                                                                                                          \n",
       "27  richard nixon is our gellert grindelwald comeymemo                                                                                                                                                                                                                                    \n",
       "28  i do not like you and because there all sad pathetic cowardly sneeky jealous spitefull bk stabbing they think your like them your just trying to protect your self from what                                                                                                          \n",
       "29  stupid bitch i wanna squish this and make a pegion sandwich                                                                                                                                                                                                                           \n",
       "30  india has blamed pegion for spying for pakistan now they are blaming the infectious disease center for development anthrax like pathogen the article says that sinopharm of china is carrying vaccine trials for coronavirus on pakistan citizens while he ignored tests in uae       \n",
       "31  qudditch is cancelled af though but large amounts of beer is as it so happens intrinsic to the culture and history of morris dancing which is a sport on account of it has teams and sometimes apparently fights                                                                      \n",
       "32  democrats are shitty spitefull morons they are truely evil vile human beings after that donkey show today they should all be tossed out on their asses ps jerry nadler is a fat fucking slob                                                                                          \n",
       "33  absolutely nasty vile wicked spitefull individual                                                                                                                                                                                                                                     \n",
       "34  whyidontliketrump yesterday agbarr besides the day trump sworn in was saddest ever raceist agbarr chief enabler to raceist trump ill informed nazi trump besides virus were also fighting corrupt hatefull spitefull dr death himself trump                                           \n",
       "35  we have tickets for aug th pegion forge tn not going to happen is it😢 they still have everything still up on their wed page                                                                                                                                                           \n",
       "36  yes it sure is one a himalayan green pegion checked my birds of india book before replying 😊                                                                                                                                                                                          \n",
       "37  this cruel thesis a spitefull angels litany will soon soar from the window of your destiny                                                                                                                                                                                            \n",
       "38  india is a country who makes its people fool with fake surgical strikes propaganda videos fake pictures pegion and locusts in reality indianarmy always gets thrashed amp humiliated by pakistan and china with online available proves 😁😁😁😁                                          \n",
       "39  even a pegion is introduced to a yrs child as pegion n not as kabootar cow as cow instead of gaay or gau instead of tata we r teaching bye instead of क्षमा we teach sorry instead of कृपया we teach please but we will wait for clear opinion till all points r cleared              \n",
       "\n",
       "    Sentiment  \n",
       "0   1          \n",
       "1   2          \n",
       "2   1          \n",
       "3   0          \n",
       "4   0          \n",
       "5   2          \n",
       "6   0          \n",
       "7   0          \n",
       "8   1          \n",
       "9   1          \n",
       "10  0          \n",
       "11  2          \n",
       "12  2          \n",
       "13  2          \n",
       "14  1          \n",
       "15  1          \n",
       "16  2          \n",
       "17  2          \n",
       "18  2          \n",
       "19  0          \n",
       "20  2          \n",
       "21  0          \n",
       "22  2          \n",
       "23  0          \n",
       "24  1          \n",
       "25  2          \n",
       "26  2          \n",
       "27  2          \n",
       "28  0          \n",
       "29  0          \n",
       "30  0          \n",
       "31  1          \n",
       "32  0          \n",
       "33  0          \n",
       "34  0          \n",
       "35  2          \n",
       "36  1          \n",
       "37  0          \n",
       "38  0          \n",
       "39  0          "
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Tweets['Tweet'] = Tweets['Tweet'].apply(cleanTxt)\n",
    "Tweets.head(40)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# The Tweets Still contains Emoji 😉  \n",
    "\n",
    "    we can either remove them or replace them with equivalent text as sometimes these emoji do carry valuable\n",
    "    sentiments/information \n",
    "    example:\n",
    "    The food was 😍 : the emoji suggest the food was good (positive sentiment)\n",
    "    The food was 😑 : the emoji sugeest the food was bad or average (negative sentiment)\n",
    "    without emoji the text \"The food was\" seems to be Neutral\n",
    "    \n",
    "    Therefore its better to retain the emoji by converting them to text in order to retain the actual sentiment \n",
    "    \n",
    "    Note:\n",
    "    while replcaing emoji , emoji that are represented by multiple words ex: Happy_face or red-heart\n",
    "    so we also need to replace the \"_\", \"-\" with blank space.\n",
    "    \n",
    "    delimiters=(\"\", \"\") : by default delimeters are : , : i.e the text for emoji will place between two colons\n",
    "                          so replace the colon with single white space string."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def convertEmoji(text):\n",
    "    text = emoji.demojize(text, delimiters=(\" \", \" \"))\n",
    "    text = text = re.sub(r'_',' ', text) # replace underscore with single white space\n",
    "    text = text = re.sub(r'-',' ', text) # replace '-' with single white space\n",
    "    text = re.sub('\\s+',' ',text)\n",
    "    #replacing emoji places single white spaces before and after which may lead to multiple white spaces,\n",
    "    #hence replaced multiples white space with a single white space\n",
    "    return text"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Appling emoji conversion to our dataframe:"
   ]
  },
  {
   "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>Tweet</th>\n",
       "      <th>Sentiment</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>i need to make my own sapcex some people dont deserve to live here</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>thre was mention of this earlier the infraction on the views of the cosmos from the sapcex array and how astronomy would compensate</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>sachine tendulkar handle his account and but here kangana team handles her thats the difference here and if she is doing that whats funny she has earned that right also jo log khud nahi bolte h wo dusron se bulwate hain wo khud bol rahi h to galat kya h</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>mainly stoltenberg scandanavian cowardly attacks oil pipe lines then running like weird inbred fish savages to underground bunkers i wont dive into it but that world is over and like nazis there being spitefull and abusive and fucking everything up because there dumb inbreds</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>he i mean vaibhavspacex is doing great work but its unfair to compare two no offense to sapcex but i will chose isro over spacex anyday</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>someone get me the panda pegion nike sb</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>piers morgan a scum media star showing his spitefull side what a pathetic little man</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>old pc game skyboxes emit such elegiac auras just look at like the skybox from the harry potter qudditch area</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>fantastic beasts the crimes of grindelwald gellert grindelwald plans to raise an army of wizards to rule over nonmagical beings in response newt scamanders former professor albus dumbledore seeks his help to stop him fantasticbeaststhecrimesofgrindelwald</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>also kit harrigton is good in this movie i dont care what anybody says</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>his excellency pegion a tragic loss we wont forget about</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>you think its not a problem damn what a pegion holed murrican</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>pegion</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>jk rowling reveals gellert grindelwald actually had aquamarine eyes short hair and a bossy personality</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>i was most famous for my defeat of gellert grindelwald the discovery of the twelve uses of dragons blood and my work on alchemy with nicolas flamel</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>cbicrucialforssr aditya thakrey why you are blocking the twitter account of everybody who is tweeting you beta why are you so scared as an youth come out from your pegion whole n demand cbi probe for your friend sushant</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>ive read qudditch legend renjun volleyball setter renjun and baseball captain renjun but never basketball player jun broken heart</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>pegion</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>i had a white pegion but then he died</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>bro they wont understand face with tears of joy you give them facts they will ignore it its like a game of chess against a pegion clown face</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>hey aman i found an injured pegion in my parking area and thought i could help it survive can you help me out here</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>im still mad and sad about something very stupid and i will probably still be for a long time because im absolutely spitefull and i cant even tell the people involved because i think i am in the wrong by being childish and egoist smh fuck</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>pegion from phone camera grinning face with sweat</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>her face says it all spitefull vindictive look that cant be hidden such a shame harry couldnt see this streak in her</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>greater good he leaned closer to her and whispered i am gellert grindelwald</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>pegion residency by universal located at tank hill rd end ahead of the stone quarry with a choice of bedroom bedroom and bedroom apartment starting at m our representatives are here to help julie dorah</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>please my people never compare with sake dancehall is all about partoa and wale speaks pegion not partoa face with monocle face with monocle face with monocle</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>richard nixon is our gellert grindelwald comeymemo</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>i do not like you and because there all sad pathetic cowardly sneeky jealous spitefull bk stabbing they think your like them your just trying to protect your self from what</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>stupid bitch i wanna squish this and make a pegion sandwich</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>india has blamed pegion for spying for pakistan now they are blaming the infectious disease center for development anthrax like pathogen the article says that sinopharm of china is carrying vaccine trials for coronavirus on pakistan citizens while he ignored tests in uae</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>qudditch is cancelled af though but large amounts of beer is as it so happens intrinsic to the culture and history of morris dancing which is a sport on account of it has teams and sometimes apparently fights</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>democrats are shitty spitefull morons they are truely evil vile human beings after that donkey show today they should all be tossed out on their asses ps jerry nadler is a fat fucking slob</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>absolutely nasty vile wicked spitefull individual</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>whyidontliketrump yesterday agbarr besides the day trump sworn in was saddest ever raceist agbarr chief enabler to raceist trump ill informed nazi trump besides virus were also fighting corrupt hatefull spitefull dr death himself trump</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>we have tickets for aug th pegion forge tn not going to happen is it crying face they still have everything still up on their wed page</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36</th>\n",
       "      <td>yes it sure is one a himalayan green pegion checked my birds of india book before replying smiling face with smiling eyes</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37</th>\n",
       "      <td>this cruel thesis a spitefull angels litany will soon soar from the window of your destiny</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>38</th>\n",
       "      <td>india is a country who makes its people fool with fake surgical strikes propaganda videos fake pictures pegion and locusts in reality indianarmy always gets thrashed amp humiliated by pakistan and china with online available proves beaming face with smiling eyes beaming face with smiling eyes beaming face with smiling eyes beaming face with smiling eyes</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39</th>\n",
       "      <td>even a pegion is introduced to a yrs child as pegion n not as kabootar cow as cow instead of gaay or gau instead of tata we r teaching bye instead of क्षमा we teach sorry instead of कृपया we teach please but we will wait for clear opinion till all points r cleared</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                                                                                                                                                                                                                                                                                                                   Tweet  \\\n",
       "0   i need to make my own sapcex some people dont deserve to live here                                                                                                                                                                                                                                                                                                     \n",
       "1   thre was mention of this earlier the infraction on the views of the cosmos from the sapcex array and how astronomy would compensate                                                                                                                                                                                                                                    \n",
       "2   sachine tendulkar handle his account and but here kangana team handles her thats the difference here and if she is doing that whats funny she has earned that right also jo log khud nahi bolte h wo dusron se bulwate hain wo khud bol rahi h to galat kya h                                                                                                          \n",
       "3   mainly stoltenberg scandanavian cowardly attacks oil pipe lines then running like weird inbred fish savages to underground bunkers i wont dive into it but that world is over and like nazis there being spitefull and abusive and fucking everything up because there dumb inbreds                                                                                    \n",
       "4   he i mean vaibhavspacex is doing great work but its unfair to compare two no offense to sapcex but i will chose isro over spacex anyday                                                                                                                                                                                                                                \n",
       "5   someone get me the panda pegion nike sb                                                                                                                                                                                                                                                                                                                                \n",
       "6   piers morgan a scum media star showing his spitefull side what a pathetic little man                                                                                                                                                                                                                                                                                   \n",
       "7   old pc game skyboxes emit such elegiac auras just look at like the skybox from the harry potter qudditch area                                                                                                                                                                                                                                                          \n",
       "8   fantastic beasts the crimes of grindelwald gellert grindelwald plans to raise an army of wizards to rule over nonmagical beings in response newt scamanders former professor albus dumbledore seeks his help to stop him fantasticbeaststhecrimesofgrindelwald                                                                                                         \n",
       "9   also kit harrigton is good in this movie i dont care what anybody says                                                                                                                                                                                                                                                                                                 \n",
       "10  his excellency pegion a tragic loss we wont forget about                                                                                                                                                                                                                                                                                                               \n",
       "11  you think its not a problem damn what a pegion holed murrican                                                                                                                                                                                                                                                                                                          \n",
       "12  pegion                                                                                                                                                                                                                                                                                                                                                                 \n",
       "13  jk rowling reveals gellert grindelwald actually had aquamarine eyes short hair and a bossy personality                                                                                                                                                                                                                                                                 \n",
       "14  i was most famous for my defeat of gellert grindelwald the discovery of the twelve uses of dragons blood and my work on alchemy with nicolas flamel                                                                                                                                                                                                                    \n",
       "15  cbicrucialforssr aditya thakrey why you are blocking the twitter account of everybody who is tweeting you beta why are you so scared as an youth come out from your pegion whole n demand cbi probe for your friend sushant                                                                                                                                            \n",
       "16  ive read qudditch legend renjun volleyball setter renjun and baseball captain renjun but never basketball player jun broken heart                                                                                                                                                                                                                                      \n",
       "17  pegion                                                                                                                                                                                                                                                                                                                                                                 \n",
       "18  i had a white pegion but then he died                                                                                                                                                                                                                                                                                                                                  \n",
       "19  bro they wont understand face with tears of joy you give them facts they will ignore it its like a game of chess against a pegion clown face                                                                                                                                                                                                                           \n",
       "20  hey aman i found an injured pegion in my parking area and thought i could help it survive can you help me out here                                                                                                                                                                                                                                                     \n",
       "21  im still mad and sad about something very stupid and i will probably still be for a long time because im absolutely spitefull and i cant even tell the people involved because i think i am in the wrong by being childish and egoist smh fuck                                                                                                                         \n",
       "22  pegion from phone camera grinning face with sweat                                                                                                                                                                                                                                                                                                                      \n",
       "23  her face says it all spitefull vindictive look that cant be hidden such a shame harry couldnt see this streak in her                                                                                                                                                                                                                                                   \n",
       "24  greater good he leaned closer to her and whispered i am gellert grindelwald                                                                                                                                                                                                                                                                                            \n",
       "25  pegion residency by universal located at tank hill rd end ahead of the stone quarry with a choice of bedroom bedroom and bedroom apartment starting at m our representatives are here to help julie dorah                                                                                                                                                              \n",
       "26  please my people never compare with sake dancehall is all about partoa and wale speaks pegion not partoa face with monocle face with monocle face with monocle                                                                                                                                                                                                         \n",
       "27  richard nixon is our gellert grindelwald comeymemo                                                                                                                                                                                                                                                                                                                     \n",
       "28  i do not like you and because there all sad pathetic cowardly sneeky jealous spitefull bk stabbing they think your like them your just trying to protect your self from what                                                                                                                                                                                           \n",
       "29  stupid bitch i wanna squish this and make a pegion sandwich                                                                                                                                                                                                                                                                                                            \n",
       "30  india has blamed pegion for spying for pakistan now they are blaming the infectious disease center for development anthrax like pathogen the article says that sinopharm of china is carrying vaccine trials for coronavirus on pakistan citizens while he ignored tests in uae                                                                                        \n",
       "31  qudditch is cancelled af though but large amounts of beer is as it so happens intrinsic to the culture and history of morris dancing which is a sport on account of it has teams and sometimes apparently fights                                                                                                                                                       \n",
       "32  democrats are shitty spitefull morons they are truely evil vile human beings after that donkey show today they should all be tossed out on their asses ps jerry nadler is a fat fucking slob                                                                                                                                                                           \n",
       "33  absolutely nasty vile wicked spitefull individual                                                                                                                                                                                                                                                                                                                      \n",
       "34  whyidontliketrump yesterday agbarr besides the day trump sworn in was saddest ever raceist agbarr chief enabler to raceist trump ill informed nazi trump besides virus were also fighting corrupt hatefull spitefull dr death himself trump                                                                                                                            \n",
       "35  we have tickets for aug th pegion forge tn not going to happen is it crying face they still have everything still up on their wed page                                                                                                                                                                                                                                 \n",
       "36  yes it sure is one a himalayan green pegion checked my birds of india book before replying smiling face with smiling eyes                                                                                                                                                                                                                                              \n",
       "37  this cruel thesis a spitefull angels litany will soon soar from the window of your destiny                                                                                                                                                                                                                                                                             \n",
       "38  india is a country who makes its people fool with fake surgical strikes propaganda videos fake pictures pegion and locusts in reality indianarmy always gets thrashed amp humiliated by pakistan and china with online available proves beaming face with smiling eyes beaming face with smiling eyes beaming face with smiling eyes beaming face with smiling eyes    \n",
       "39  even a pegion is introduced to a yrs child as pegion n not as kabootar cow as cow instead of gaay or gau instead of tata we r teaching bye instead of क्षमा we teach sorry instead of कृपया we teach please but we will wait for clear opinion till all points r cleared                                                                                               \n",
       "\n",
       "    Sentiment  \n",
       "0   1          \n",
       "1   2          \n",
       "2   1          \n",
       "3   0          \n",
       "4   0          \n",
       "5   2          \n",
       "6   0          \n",
       "7   0          \n",
       "8   1          \n",
       "9   1          \n",
       "10  0          \n",
       "11  2          \n",
       "12  2          \n",
       "13  2          \n",
       "14  1          \n",
       "15  1          \n",
       "16  2          \n",
       "17  2          \n",
       "18  2          \n",
       "19  0          \n",
       "20  2          \n",
       "21  0          \n",
       "22  2          \n",
       "23  0          \n",
       "24  1          \n",
       "25  2          \n",
       "26  2          \n",
       "27  2          \n",
       "28  0          \n",
       "29  0          \n",
       "30  0          \n",
       "31  1          \n",
       "32  0          \n",
       "33  0          \n",
       "34  0          \n",
       "35  2          \n",
       "36  1          \n",
       "37  0          \n",
       "38  0          \n",
       "39  0          "
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Tweets['Tweet'] = Tweets['Tweet'].apply(convertEmoji)\n",
    "Tweets.head(40)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Now the Text free from mentions/Usernames , # , Hyperlinks and emojis are replaced with their meanings\n",
    "\n",
    "      also as the emoji are now converted to words , the sentiment of the tweet might change hence lets update \n",
    "      the sentiment accordingly"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_tweet_sentiment(tweet): \n",
    "    \n",
    "    # create TextBlob object of passed tweet text \n",
    "    analysis = TextBlob(tweet)\n",
    "    # set sentiment \n",
    "    if analysis.sentiment.polarity > 0: \n",
    "        return 1\n",
    "    elif analysis.sentiment.polarity == 0: \n",
    "        return 2\n",
    "    else: \n",
    "        return 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "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>Tweet</th>\n",
       "      <th>Sentiment</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>i need to make my own sapcex some people dont deserve to live here</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>thre was mention of this earlier the infraction on the views of the cosmos from the sapcex array and how astronomy would compensate</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>sachine tendulkar handle his account and but here kangana team handles her thats the difference here and if she is doing that whats funny she has earned that right also jo log khud nahi bolte h wo dusron se bulwate hain wo khud bol rahi h to galat kya h</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>mainly stoltenberg scandanavian cowardly attacks oil pipe lines then running like weird inbred fish savages to underground bunkers i wont dive into it but that world is over and like nazis there being spitefull and abusive and fucking everything up because there dumb inbreds</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>he i mean vaibhavspacex is doing great work but its unfair to compare two no offense to sapcex but i will chose isro over spacex anyday</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>someone get me the panda pegion nike sb</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>piers morgan a scum media star showing his spitefull side what a pathetic little man</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>old pc game skyboxes emit such elegiac auras just look at like the skybox from the harry potter qudditch area</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>fantastic beasts the crimes of grindelwald gellert grindelwald plans to raise an army of wizards to rule over nonmagical beings in response newt scamanders former professor albus dumbledore seeks his help to stop him fantasticbeaststhecrimesofgrindelwald</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>also kit harrigton is good in this movie i dont care what anybody says</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                                                                                                                                                                                                                                 Tweet  \\\n",
       "0  i need to make my own sapcex some people dont deserve to live here                                                                                                                                                                                                                    \n",
       "1  thre was mention of this earlier the infraction on the views of the cosmos from the sapcex array and how astronomy would compensate                                                                                                                                                   \n",
       "2  sachine tendulkar handle his account and but here kangana team handles her thats the difference here and if she is doing that whats funny she has earned that right also jo log khud nahi bolte h wo dusron se bulwate hain wo khud bol rahi h to galat kya h                         \n",
       "3  mainly stoltenberg scandanavian cowardly attacks oil pipe lines then running like weird inbred fish savages to underground bunkers i wont dive into it but that world is over and like nazis there being spitefull and abusive and fucking everything up because there dumb inbreds   \n",
       "4  he i mean vaibhavspacex is doing great work but its unfair to compare two no offense to sapcex but i will chose isro over spacex anyday                                                                                                                                               \n",
       "5  someone get me the panda pegion nike sb                                                                                                                                                                                                                                               \n",
       "6  piers morgan a scum media star showing his spitefull side what a pathetic little man                                                                                                                                                                                                  \n",
       "7  old pc game skyboxes emit such elegiac auras just look at like the skybox from the harry potter qudditch area                                                                                                                                                                         \n",
       "8  fantastic beasts the crimes of grindelwald gellert grindelwald plans to raise an army of wizards to rule over nonmagical beings in response newt scamanders former professor albus dumbledore seeks his help to stop him fantasticbeaststhecrimesofgrindelwald                        \n",
       "9  also kit harrigton is good in this movie i dont care what anybody says                                                                                                                                                                                                                \n",
       "\n",
       "   Sentiment  \n",
       "0  1          \n",
       "1  2          \n",
       "2  1          \n",
       "3  0          \n",
       "4  0          \n",
       "5  2          \n",
       "6  0          \n",
       "7  0          \n",
       "8  1          \n",
       "9  1          "
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Tweets['Sentiment']= Tweets['Tweet'].apply(get_tweet_sentiment)\n",
    "Tweets.head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Now to Normailze the text , convert the text into tokens and perfrom stemming\n",
    "\n",
    "    Tokenizing: Turning the tweets into tokens. Tokens are words separated by spaces in a text.\n",
    "    \n",
    "    Normalization: Normalization generally refers to a series of related tasks meant to put all text on the same level.\n",
    "                   Converting text to lower case, removing special characters, and removing stopwords will remove basic\n",
    "                   inconsistencies. Normalization improves text matching.\n",
    "                   \n",
    "                   we have already done majority of normalization ,only stopwords removal remians.\n",
    "            \n",
    "     Stopword removal: Some words do not contribute much to the machine learning model, so it's good to remove them.\n",
    "                       \n",
    "     Stemming: Eliminating affixes (circumfixes, suffixes, prefixes, infixes) from a word in order to obtain a word stem.\n",
    "               Porter Stemmer is the most widely used technique because it is very fast.\n",
    "               Generally, stemming chops off end of the word, and mostly it works fine.\n",
    "               Example: Working -> Work\n",
    "               \n",
    "     Lemmatization: The goal is same as with stemming, but stemming a word sometimes loses the actual meaning of the word.\n",
    "                    Lemmatization usually refers to doing things properly using vocabulary and morphological analysis of\n",
    "                    words. It returns the base or dictionary form of a word, also known as the lemma .\n",
    "                Example: Better -> Good.\n",
    "                \n",
    "      Here i am going to use lemmatization\n",
    "      \n",
    "      syntax .lemmatize(word, pos='a'):\n",
    "              pos : part of speech tag , defualt : n (Noun)\n",
    "              pos = 'a' : adjective \n",
    "              example:\n",
    "              lemmatize('better',pos='a')==> good"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "def Tokenization(tweet):\n",
    "    \n",
    "    #convert to tokens and filter out stopwords\n",
    "    stop_words = set(stopwords.words('english'))\n",
    "    tweet_tokens = word_tokenize(tweet) \n",
    "    filtered_words = [w for w in tweet_tokens if not w in stop_words]\n",
    "    \n",
    "    \n",
    "    lemmatizer = WordNetLemmatizer()\n",
    "    lemma_words = [lemmatizer.lemmatize(w, pos='a') for w in filtered_words]\n",
    "    \n",
    "    return \" \".join(lemma_words)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "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>Tweet</th>\n",
       "      <th>Sentiment</th>\n",
       "      <th>Tweets_Token</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>i need to make my own sapcex some people dont deserve to live here</td>\n",
       "      <td>1</td>\n",
       "      <td>need make sapcex people dont deserve live</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>thre was mention of this earlier the infraction on the views of the cosmos from the sapcex array and how astronomy would compensate</td>\n",
       "      <td>2</td>\n",
       "      <td>thre mention early infraction views cosmos sapcex array astronomy would compensate</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>sachine tendulkar handle his account and but here kangana team handles her thats the difference here and if she is doing that whats funny she has earned that right also jo log khud nahi bolte h wo dusron se bulwate hain wo khud bol rahi h to galat kya h</td>\n",
       "      <td>1</td>\n",
       "      <td>sachine tendulkar handle account kangana team handles thats difference whats funny earned right also jo log khud nahi bolte h wo dusron se bulwate hain wo khud bol rahi h galat kya h</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>mainly stoltenberg scandanavian cowardly attacks oil pipe lines then running like weird inbred fish savages to underground bunkers i wont dive into it but that world is over and like nazis there being spitefull and abusive and fucking everything up because there dumb inbreds</td>\n",
       "      <td>0</td>\n",
       "      <td>mainly stoltenberg scandanavian cowardly attacks oil pipe lines running like weird inbred fish savages underground bunkers wont dive world like nazis spitefull abusive fucking everything dumb inbreds</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>he i mean vaibhavspacex is doing great work but its unfair to compare two no offense to sapcex but i will chose isro over spacex anyday</td>\n",
       "      <td>0</td>\n",
       "      <td>mean vaibhavspacex great work unfair compare two offense sapcex chose isro spacex anyday</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>someone get me the panda pegion nike sb</td>\n",
       "      <td>2</td>\n",
       "      <td>someone get panda pegion nike sb</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>piers morgan a scum media star showing his spitefull side what a pathetic little man</td>\n",
       "      <td>0</td>\n",
       "      <td>piers morgan scum media star showing spitefull side pathetic little man</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>old pc game skyboxes emit such elegiac auras just look at like the skybox from the harry potter qudditch area</td>\n",
       "      <td>0</td>\n",
       "      <td>old pc game skyboxes emit elegiac auras look like skybox harry potter qudditch area</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>fantastic beasts the crimes of grindelwald gellert grindelwald plans to raise an army of wizards to rule over nonmagical beings in response newt scamanders former professor albus dumbledore seeks his help to stop him fantasticbeaststhecrimesofgrindelwald</td>\n",
       "      <td>1</td>\n",
       "      <td>fantastic beasts crimes grindelwald gellert grindelwald plans raise army wizards rule nonmagical beings response newt scamanders former professor albus dumbledore seeks help stop fantasticbeaststhecrimesofgrindelwald</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>also kit harrigton is good in this movie i dont care what anybody says</td>\n",
       "      <td>1</td>\n",
       "      <td>also kit harrigton good movie dont care anybody says</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>his excellency pegion a tragic loss we wont forget about</td>\n",
       "      <td>0</td>\n",
       "      <td>excellency pegion tragic loss wont forget</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>you think its not a problem damn what a pegion holed murrican</td>\n",
       "      <td>2</td>\n",
       "      <td>think problem damn pegion holed murrican</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>pegion</td>\n",
       "      <td>2</td>\n",
       "      <td>pegion</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>jk rowling reveals gellert grindelwald actually had aquamarine eyes short hair and a bossy personality</td>\n",
       "      <td>2</td>\n",
       "      <td>jk rowling reveals gellert grindelwald actually aquamarine eyes short hair bossy personality</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>i was most famous for my defeat of gellert grindelwald the discovery of the twelve uses of dragons blood and my work on alchemy with nicolas flamel</td>\n",
       "      <td>1</td>\n",
       "      <td>famous defeat gellert grindelwald discovery twelve uses dragons blood work alchemy nicolas flamel</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>cbicrucialforssr aditya thakrey why you are blocking the twitter account of everybody who is tweeting you beta why are you so scared as an youth come out from your pegion whole n demand cbi probe for your friend sushant</td>\n",
       "      <td>1</td>\n",
       "      <td>cbicrucialforssr aditya thakrey blocking twitter account everybody tweeting beta scared youth come pegion whole n demand cbi probe friend sushant</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>ive read qudditch legend renjun volleyball setter renjun and baseball captain renjun but never basketball player jun broken heart</td>\n",
       "      <td>0</td>\n",
       "      <td>ive read qudditch legend renjun volleyball setter renjun baseball captain renjun never basketball player jun broken heart</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>pegion</td>\n",
       "      <td>2</td>\n",
       "      <td>pegion</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>i had a white pegion but then he died</td>\n",
       "      <td>2</td>\n",
       "      <td>white pegion died</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>bro they wont understand face with tears of joy you give them facts they will ignore it its like a game of chess against a pegion clown face</td>\n",
       "      <td>1</td>\n",
       "      <td>bro wont understand face tears joy give facts ignore like game chess pegion clown face</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                                                                                                                                                                                                                                  Tweet  \\\n",
       "0   i need to make my own sapcex some people dont deserve to live here                                                                                                                                                                                                                    \n",
       "1   thre was mention of this earlier the infraction on the views of the cosmos from the sapcex array and how astronomy would compensate                                                                                                                                                   \n",
       "2   sachine tendulkar handle his account and but here kangana team handles her thats the difference here and if she is doing that whats funny she has earned that right also jo log khud nahi bolte h wo dusron se bulwate hain wo khud bol rahi h to galat kya h                         \n",
       "3   mainly stoltenberg scandanavian cowardly attacks oil pipe lines then running like weird inbred fish savages to underground bunkers i wont dive into it but that world is over and like nazis there being spitefull and abusive and fucking everything up because there dumb inbreds   \n",
       "4   he i mean vaibhavspacex is doing great work but its unfair to compare two no offense to sapcex but i will chose isro over spacex anyday                                                                                                                                               \n",
       "5   someone get me the panda pegion nike sb                                                                                                                                                                                                                                               \n",
       "6   piers morgan a scum media star showing his spitefull side what a pathetic little man                                                                                                                                                                                                  \n",
       "7   old pc game skyboxes emit such elegiac auras just look at like the skybox from the harry potter qudditch area                                                                                                                                                                         \n",
       "8   fantastic beasts the crimes of grindelwald gellert grindelwald plans to raise an army of wizards to rule over nonmagical beings in response newt scamanders former professor albus dumbledore seeks his help to stop him fantasticbeaststhecrimesofgrindelwald                        \n",
       "9   also kit harrigton is good in this movie i dont care what anybody says                                                                                                                                                                                                                \n",
       "10  his excellency pegion a tragic loss we wont forget about                                                                                                                                                                                                                              \n",
       "11  you think its not a problem damn what a pegion holed murrican                                                                                                                                                                                                                         \n",
       "12  pegion                                                                                                                                                                                                                                                                                \n",
       "13  jk rowling reveals gellert grindelwald actually had aquamarine eyes short hair and a bossy personality                                                                                                                                                                                \n",
       "14  i was most famous for my defeat of gellert grindelwald the discovery of the twelve uses of dragons blood and my work on alchemy with nicolas flamel                                                                                                                                   \n",
       "15  cbicrucialforssr aditya thakrey why you are blocking the twitter account of everybody who is tweeting you beta why are you so scared as an youth come out from your pegion whole n demand cbi probe for your friend sushant                                                           \n",
       "16  ive read qudditch legend renjun volleyball setter renjun and baseball captain renjun but never basketball player jun broken heart                                                                                                                                                     \n",
       "17  pegion                                                                                                                                                                                                                                                                                \n",
       "18  i had a white pegion but then he died                                                                                                                                                                                                                                                 \n",
       "19  bro they wont understand face with tears of joy you give them facts they will ignore it its like a game of chess against a pegion clown face                                                                                                                                          \n",
       "\n",
       "    Sentiment  \\\n",
       "0   1           \n",
       "1   2           \n",
       "2   1           \n",
       "3   0           \n",
       "4   0           \n",
       "5   2           \n",
       "6   0           \n",
       "7   0           \n",
       "8   1           \n",
       "9   1           \n",
       "10  0           \n",
       "11  2           \n",
       "12  2           \n",
       "13  2           \n",
       "14  1           \n",
       "15  1           \n",
       "16  0           \n",
       "17  2           \n",
       "18  2           \n",
       "19  1           \n",
       "\n",
       "                                                                                                                                                                                                                Tweets_Token  \n",
       "0   need make sapcex people dont deserve live                                                                                                                                                                                 \n",
       "1   thre mention early infraction views cosmos sapcex array astronomy would compensate                                                                                                                                        \n",
       "2   sachine tendulkar handle account kangana team handles thats difference whats funny earned right also jo log khud nahi bolte h wo dusron se bulwate hain wo khud bol rahi h galat kya h                                    \n",
       "3   mainly stoltenberg scandanavian cowardly attacks oil pipe lines running like weird inbred fish savages underground bunkers wont dive world like nazis spitefull abusive fucking everything dumb inbreds                   \n",
       "4   mean vaibhavspacex great work unfair compare two offense sapcex chose isro spacex anyday                                                                                                                                  \n",
       "5   someone get panda pegion nike sb                                                                                                                                                                                          \n",
       "6   piers morgan scum media star showing spitefull side pathetic little man                                                                                                                                                   \n",
       "7   old pc game skyboxes emit elegiac auras look like skybox harry potter qudditch area                                                                                                                                       \n",
       "8   fantastic beasts crimes grindelwald gellert grindelwald plans raise army wizards rule nonmagical beings response newt scamanders former professor albus dumbledore seeks help stop fantasticbeaststhecrimesofgrindelwald  \n",
       "9   also kit harrigton good movie dont care anybody says                                                                                                                                                                      \n",
       "10  excellency pegion tragic loss wont forget                                                                                                                                                                                 \n",
       "11  think problem damn pegion holed murrican                                                                                                                                                                                  \n",
       "12  pegion                                                                                                                                                                                                                    \n",
       "13  jk rowling reveals gellert grindelwald actually aquamarine eyes short hair bossy personality                                                                                                                              \n",
       "14  famous defeat gellert grindelwald discovery twelve uses dragons blood work alchemy nicolas flamel                                                                                                                         \n",
       "15  cbicrucialforssr aditya thakrey blocking twitter account everybody tweeting beta scared youth come pegion whole n demand cbi probe friend sushant                                                                         \n",
       "16  ive read qudditch legend renjun volleyball setter renjun baseball captain renjun never basketball player jun broken heart                                                                                                 \n",
       "17  pegion                                                                                                                                                                                                                    \n",
       "18  white pegion died                                                                                                                                                                                                         \n",
       "19  bro wont understand face tears joy give facts ignore like game chess pegion clown face                                                                                                                                    "
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Tweets['Tweets_Token'] = Tweets['Tweet'].apply(Tokenization)\n",
    "Tweets.head(20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['need make sapcex people dont deserve live',\n",
       " 'thre mention early infraction views cosmos sapcex array astronomy would compensate',\n",
       " 'sachine tendulkar handle account kangana team handles thats difference whats funny earned right also jo log khud nahi bolte h wo dusron se bulwate hain wo khud bol rahi h galat kya h',\n",
       " 'mainly stoltenberg scandanavian cowardly attacks oil pipe lines running like weird inbred fish savages underground bunkers wont dive world like nazis spitefull abusive fucking everything dumb inbreds',\n",
       " 'mean vaibhavspacex great work unfair compare two offense sapcex chose isro spacex anyday',\n",
       " 'someone get panda pegion nike sb',\n",
       " 'piers morgan scum media star showing spitefull side pathetic little man',\n",
       " 'old pc game skyboxes emit elegiac auras look like skybox harry potter qudditch area',\n",
       " 'fantastic beasts crimes grindelwald gellert grindelwald plans raise army wizards rule nonmagical beings response newt scamanders former professor albus dumbledore seeks help stop fantasticbeaststhecrimesofgrindelwald',\n",
       " 'also kit harrigton good movie dont care anybody says']"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tweets_array = Tweets['Tweets_Token'].tolist()\n",
    "tweets_array[0:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[1, 2, 1, 0, 0, 2, 0, 0, 1, 1]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tweets_label =Tweets['Sentiment'].tolist()\n",
    "tweets_label[0:10]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# What is Count Vectorizer ?\n",
    "\n",
    "        ref: https://scikit-learn.org/stable/modules/generated/sklearn.feature_extraction.text.CountVectorizer.html\n",
    "             https://towardsdatascience.com/natural-language-processing-count-vectorization-with-scikit-learn-e7804269bb5e\n",
    "\n",
    "        One of the most basic ways we can numerically represent words is through the one-hot encoding method \n",
    "        also called count vectorizing.\n",
    "        The goal is to create a vector that has as many dimensions(column) as our vocabulary has unique words.\n",
    "        Each unique word has a unique dimension/column and will be represented by a 1 in that dimension with 0s\n",
    "        everywhere else.\n",
    "        As a result, really huge and sparse vectors are generated for each word in the vocabulary but these encodings\n",
    "        capture absolutely no relational information.\n",
    "        if we need that semantic relationship information we can use other encoding methods such as Word2Vec.\n",
    "\n",
    "![](countvectorizer.png)\n",
    "\n",
    "        The CountVectorizer provides a simple way to both tokenize a collection of text documents and build a vocabulary\n",
    "        of known words, but also to encode new documents using that vocabulary.\n",
    "    \n",
    "        ngrams range: default (1,1) : treat each word seperately\n",
    "                    : if we use (1,2) : treat each word as single (as min range = 1)\n",
    "                                        also comibine 2 (max range= 2) neighbouring words and treat them as one\n",
    "                                        usefull for names. example below\n",
    "        how to use:\n",
    "        Create an instance of the CountVectorizer class.\n",
    "        Call the fit() function in order to learn a vocabulary from one or more sentences.\n",
    "        call fit_transform()  to learn a vocabulary from one or more sentences and encode them at the same time.\n",
    "        Call the transform() function on one or more sentences as needed to encode each as a vector.\n",
    "        An encoded vector is returned with a length of the entire vocabulary and an integer count for the number of\n",
    "        times each word appeared in the Sentence.\n",
    "    \n",
    "        Because these vectors will contain a lot of zeros, they are called sparse.\n",
    "        The vectors returned from a call to transform() will be sparse vectors, and we can transform them\n",
    "        to numpy arrays to look and better understand what is going on by calling the np.toarray() function."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Example of Count Vectorizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['sonu nigam lucky ali']"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "text= tweets_array[111]\n",
    "text = [text]\n",
    "text"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'sonu': 5, 'nigam': 3, 'lucky': 1, 'ali': 0, 'sonu nigam': 6, 'nigam lucky': 4, 'lucky ali': 2}\n",
      "(1, 7)\n",
      "<class 'scipy.sparse.csr.csr_matrix'>\n",
      "[[1 1 1 0 0 1 0]]\n"
     ]
    }
   ],
   "source": [
    "vectorizer = CountVectorizer(ngram_range=(1,2))\n",
    "#tokenize and build vocab\n",
    "vectorizer.fit_transform(text)\n",
    "# summarize\n",
    "print(vectorizer.vocabulary_)\n",
    "# encode document\n",
    "text1 = [\"sonu is better than lucky ali\"]\n",
    "vector = vectorizer.transform(text1)\n",
    "# summarize encoded vector\n",
    "print(vector.shape)\n",
    "print(type(vector))\n",
    "print(vector.toarray())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Creating our count vectorizer model :\n",
    "     \n",
    "    after analysis using default n grams : (1,1) works better than (1,2) for this dataset and svm classifier\n",
    "    after creating the model we fit the model with our dataset to create our vocabulary."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "count_vectorizer = CountVectorizer()\n",
    "vectorized_Tweets = count_vectorizer.fit_transform(tweets_array)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Analyzing the vocabulary created by count vectorizer which is stored in vectorized_Tweets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "44523\n"
     ]
    }
   ],
   "source": [
    "print(len(count_vectorizer.vocabulary_)) # how many words we have in our vocabulary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<29348x44523 sparse matrix of type '<class 'numpy.int64'>'\n",
       "\twith 360351 stored elements in Compressed Sparse Row format>"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vectorized_Tweets"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "vectorized_Tweets contains 29348 rows i.e number of tweets and each rows has 44523 column , these represent each word\n",
    "in our vocabulary .\n",
    "360351 repesents the total number of words in our dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<1x44523 sparse matrix of type '<class 'numpy.int64'>'\n",
       "\twith 7 stored elements in Compressed Sparse Row format>"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vectorized_Tweets[0][0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "so 0,0 refers to the first tweet that is tokenized in our dataset  : 'need make sapcex people dont deserve live'\n",
    "it has 7 words so vectorized_Tweets[0][0] is a single row containing 44523 columns and the columns that refer to \n",
    "need,make,sapcex,people,dont,deserve,live are marked with 1 rest of the columns are 0"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Finally the Data is ready to be used to train a Classification Algorithm\n",
    "\n",
    "    \n",
    "# Splitting the dataset in training and test sets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "Tweet_train, Tweet_test,Sentiment_train, Sentiment_test = train_test_split(vectorized_Tweets, tweets_label, test_size=0.25, random_state=101)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Using various svm kernels and checking their performances :"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy for training set with linear kernal regularization parameter C = 0.01 is : 0.765\n",
      "Accuracy for test set with linear kernal regularization parameter C = 0.01 is : 0.725\n",
      "\n",
      "\n",
      "Accuracy for training set with rbf kernal regularization parameter C = 0.01 is : 0.365\n",
      "Accuracy for test set with rbf kernal regularization parameter C = 0.01 is : 0.360\n",
      "\n",
      "\n",
      "Accuracy for training set with sigmoid kernal regularization parameter C = 0.01 is : 0.365\n",
      "Accuracy for test set with sigmoid kernal regularization parameter C = 0.01 is : 0.360\n",
      "\n",
      "\n",
      "Accuracy for training set with polynomial kernal regularization parameter C = 0.01 is : 0.365\n",
      "Accuracy for test set with polynomial kernal regularization parameter C = 0.01 is : 0.360\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "clf = svm.SVC(kernel= 'linear',C=0.01,max_iter=20000)\n",
    "clf.fit(Tweet_train,Sentiment_train)\n",
    "print (\"Accuracy for training set with linear kernal regularization parameter C = 0.01 is : %.3f\" % ( accuracy_score(Sentiment_train, clf.predict(Tweet_train))))\n",
    "print (\"Accuracy for test set with linear kernal regularization parameter C = 0.01 is : %.3f\" % ( accuracy_score(Sentiment_test, clf.predict(Tweet_test))))\n",
    "print (\"\\n\")\n",
    "\n",
    "clf = svm.SVC(kernel= 'rbf', gamma='auto' ,C=0.01,max_iter=20000)\n",
    "clf.fit(Tweet_train,Sentiment_train)\n",
    "print (\"Accuracy for training set with rbf kernal regularization parameter C = 0.01 is : %.3f\" % ( accuracy_score(Sentiment_train, clf.predict(Tweet_train))))\n",
    "print (\"Accuracy for test set with rbf kernal regularization parameter C = 0.01 is : %.3f\" % ( accuracy_score(Sentiment_test, clf.predict(Tweet_test))))\n",
    "print (\"\\n\")\n",
    "\n",
    "clf = svm.SVC(kernel= 'sigmoid',gamma='auto' ,C=0.01,max_iter=20000)\n",
    "clf.fit(Tweet_train,Sentiment_train)\n",
    "print (\"Accuracy for training set with sigmoid kernal regularization parameter C = 0.01 is : %.3f\" % ( accuracy_score(Sentiment_train, clf.predict(Tweet_train))))\n",
    "print (\"Accuracy for test set with sigmoid kernal regularization parameter C = 0.01 is : %.3f\" % ( accuracy_score(Sentiment_test, clf.predict(Tweet_test))))\n",
    "print (\"\\n\")\n",
    "\n",
    "clf = svm.SVC(kernel= 'poly',degree=2, gamma='auto' ,C=0.01,max_iter=20000)\n",
    "clf.fit(Tweet_train,Sentiment_train)\n",
    "print (\"Accuracy for training set with polynomial kernal regularization parameter C = 0.01 is : %.3f\" % ( accuracy_score(Sentiment_train, clf.predict(Tweet_train))))\n",
    "print (\"Accuracy for test set with polynomial kernal regularization parameter C = 0.01 is : %.3f\" % ( accuracy_score(Sentiment_test, clf.predict(Tweet_test))))\n",
    "print (\"\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy for training set with regularization parameter C = 0.001 is : 0.778\n",
      "Accuracy for test set with regularization parameter C = 0.001 is : 0.736\n",
      "\n",
      "\n",
      "Accuracy for training set with regularization parameter C = 0.005 is : 0.866\n",
      "Accuracy for test set with regularization parameter C = 0.005 is : 0.804\n",
      "\n",
      "\n",
      "Accuracy for training set with regularization parameter C = 0.007 is : 0.883\n",
      "Accuracy for test set with regularization parameter C = 0.007 is : 0.815\n",
      "\n",
      "\n",
      "Accuracy for training set with regularization parameter C = 0.010 is : 0.901\n",
      "Accuracy for test set with regularization parameter C = 0.010 is : 0.823\n",
      "\n",
      "\n",
      "Accuracy for training set with regularization parameter C = 0.015 is : 0.918\n",
      "Accuracy for test set with regularization parameter C = 0.015 is : 0.837\n",
      "\n",
      "\n",
      "Accuracy for training set with regularization parameter C = 0.020 is : 0.930\n",
      "Accuracy for test set with regularization parameter C = 0.020 is : 0.844\n",
      "\n",
      "\n",
      "Accuracy for training set with regularization parameter C = 0.025 is : 0.938\n",
      "Accuracy for test set with regularization parameter C = 0.025 is : 0.851\n",
      "\n",
      "\n",
      "Accuracy for training set with regularization parameter C = 0.050 is : 0.961\n",
      "Accuracy for test set with regularization parameter C = 0.050 is : 0.862\n",
      "\n",
      "\n",
      "Accuracy for training set with regularization parameter C = 0.075 is : 0.972\n",
      "Accuracy for test set with regularization parameter C = 0.075 is : 0.865\n",
      "\n",
      "\n",
      "Accuracy for training set with regularization parameter C = 0.100 is : 0.978\n",
      "Accuracy for test set with regularization parameter C = 0.100 is : 0.867\n",
      "\n",
      "\n",
      "Accuracy for training set with regularization parameter C = 0.250 is : 0.990\n",
      "Accuracy for test set with regularization parameter C = 0.250 is : 0.872\n",
      "\n",
      "\n",
      "Accuracy for training set with regularization parameter C = 0.500 is : 0.995\n",
      "Accuracy for test set with regularization parameter C = 0.500 is : 0.869\n",
      "\n",
      "\n",
      "Accuracy for training set with regularization parameter C = 1.000 is : 0.997\n",
      "Accuracy for test set with regularization parameter C = 1.000 is : 0.865\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "training_accuracy = []\n",
    "test_accuracy = []\n",
    "\n",
    "C = [0.001,0.005,0.007,0.01,0.015,0.020,0.025,0.05,0.075,0.1,0.25,0.5,1.0]\n",
    "for c in C:\n",
    "    clf = OneVsRestClassifier(LinearSVC(C=c,max_iter=20000),n_jobs = 1)\n",
    "    clf.fit(Tweet_train,Sentiment_train)\n",
    "    train_acc =accuracy_score(Sentiment_train, clf.predict(Tweet_train))\n",
    "    test_acc = accuracy_score(Sentiment_test, clf.predict(Tweet_test))\n",
    "    test_accuracy.append( test_acc )\n",
    "    training_accuracy.append( train_acc )\n",
    "    print (\"Accuracy for training set with regularization parameter C = %.3f is : %.3f\" % (c, train_acc))\n",
    "    print (\"Accuracy for test set with regularization parameter C = %.3f is : %.3f\" % (c, test_acc))\n",
    "    print (\"\\n\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x400 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "my_dpi = 100 # dots per inch .. (resolution)\n",
    "plt.figure(figsize=(1000/my_dpi, 400/my_dpi), dpi = my_dpi) # to enlarge the plot (can be omitted)\n",
    "\n",
    "plt.plot(C, training_accuracy, label = 'Accuracy of the Training Set')\n",
    "plt.plot(C, test_accuracy, label = 'Accuracy of the Test Set')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.xticks(C)\n",
    "plt.xlabel('Regularization Parameter') \n",
    "plt.legend()\n",
    "plt.grid()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x400 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "my_dpi = 100 # dots per inch .. (resolution)\n",
    "plt.figure(figsize=(1000/my_dpi, 400/my_dpi), dpi = my_dpi) # to enlarge the plot (can be omitted)\n",
    "\n",
    "plt.plot(C[3:10], training_accuracy[3:10], label = 'Accuracy of the Training Set')\n",
    "plt.plot(C[3:10], test_accuracy[3:10], label = 'Accuracy of the Test Set')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.xticks(C[3:10])\n",
    "plt.xlabel('Regularization Parameter') \n",
    "plt.legend()\n",
    "plt.grid()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy for training set with regularization parameter C = 0.05 is : 0.961\n",
      "Accuracy for test set with regularization parameter C = 0.05 is : 0.862\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "clf = OneVsRestClassifier(LinearSVC(C=0.05,max_iter=10000),n_jobs = 1)\n",
    "clf.fit(Tweet_train,Sentiment_train)\n",
    "train_acc =accuracy_score(Sentiment_train, clf.predict(Tweet_train))\n",
    "test_acc = accuracy_score(Sentiment_test, clf.predict(Tweet_test))\n",
    "print (\"Accuracy for training set with regularization parameter C = 0.05 is : %.3f\" % (train_acc))\n",
    "print (\"Accuracy for test set with regularization parameter C = 0.05 is : %.3f\" % (test_acc))\n",
    "print (\"\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Making Predictions and Evaluating the Classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Actual_sentiments</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Actual_sentiments\n",
       "0  0                \n",
       "1  1                \n",
       "2  0                \n",
       "3  2                \n",
       "4  1                "
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sentimentdf =  pd.DataFrame( Sentiment_test , columns=['Actual_sentiments'])\n",
    "sentimentdf.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "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>Actual_sentiments</th>\n",
       "      <th>Predicted_sentiments</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Actual_sentiments  Predicted_sentiments\n",
       "0  0                  0                   \n",
       "1  1                  1                   \n",
       "2  0                  0                   \n",
       "3  2                  2                   \n",
       "4  1                  0                   \n",
       "5  1                  2                   \n",
       "6  0                  0                   \n",
       "7  2                  2                   \n",
       "8  0                  0                   \n",
       "9  0                  0                   "
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "predictions = clf.predict(Tweet_test)\n",
    "Id = sentimentdf.index\n",
    "predicted_sentiments = pd.DataFrame(predictions, Id , columns = ['Predicted_sentiments'])\n",
    "predictions_comparions=sentimentdf.merge(predicted_sentiments, how = 'outer', left_index = True, right_index = True)\n",
    "predictions_comparions.head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Function to Convert numeric labels to their String equivalent"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "def convert_sent(number):\n",
    "    if number == 2 :\n",
    "        return 'Neutral'\n",
    "    elif number == 1:\n",
    "        return 'positive'\n",
    "    else :\n",
    "        return 'negetive'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "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>Actual_sentiments</th>\n",
       "      <th>Predicted_sentiments</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>negetive</td>\n",
       "      <td>negetive</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>positive</td>\n",
       "      <td>positive</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>negetive</td>\n",
       "      <td>negetive</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Neutral</td>\n",
       "      <td>Neutral</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>positive</td>\n",
       "      <td>negetive</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>positive</td>\n",
       "      <td>Neutral</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>negetive</td>\n",
       "      <td>negetive</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Neutral</td>\n",
       "      <td>Neutral</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>negetive</td>\n",
       "      <td>negetive</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>negetive</td>\n",
       "      <td>negetive</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Actual_sentiments Predicted_sentiments\n",
       "0  negetive          negetive           \n",
       "1  positive          positive           \n",
       "2  negetive          negetive           \n",
       "3  Neutral           Neutral            \n",
       "4  positive          negetive           \n",
       "5  positive          Neutral            \n",
       "6  negetive          negetive           \n",
       "7  Neutral           Neutral            \n",
       "8  negetive          negetive           \n",
       "9  negetive          negetive           "
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "predictions_comparions['Actual_sentiments'] = predictions_comparions['Actual_sentiments'].apply(convert_sent)\n",
    "predictions_comparions['Predicted_sentiments'] = predictions_comparions['Predicted_sentiments'].apply(convert_sent)\n",
    "predictions_comparions.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Predicted_sentiments</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Actual_sentiments</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Neutral</th>\n",
       "      <td>2289</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>negetive</th>\n",
       "      <td>2406</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>positive</th>\n",
       "      <td>2642</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                   Predicted_sentiments\n",
       "Actual_sentiments                      \n",
       "Neutral            2289                \n",
       "negetive           2406                \n",
       "positive           2642                "
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "predictions_comparions.groupby(['Actual_sentiments']).count()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Confusion Matrix And calculating Precision Recall and F1 Score\n",
    "\n",
    "    In the Confusion Matrix : Negetive = 0 ; Positive = 1 ; Neutral = 2 (class Labels)\n",
    "                Row represents actual Sentiments while the columns represents the predicted sentiments\n",
    "                \n",
    "                1st row : 1st column :Actual Sentiment = Negetive ; Predicted Sentiment = Negetive\n",
    "                1st row : 2nd column :Actual Sentiment = Negetive ; Predicted Sentiment = Positive\n",
    "                1st row : 3rd column :Actual Sentiment = Negetive ; Predicted Sentiment = Neutral\n",
    "                \n",
    "                2nd row : 1st column :Actual Sentiment = Positive ; Predicted Sentiment = Negetive\n",
    "                2nd row : 2nd column :Actual Sentiment = Positive ; Predicted Sentiment = Positive\n",
    "                2nd row : 3rd column :Actual Sentiment = Positive ; Predicted Sentiment = Neutral\n",
    "                \n",
    "                3rd row : 1st column :Actual Sentiment = Neutral ; Predicted Sentiment = Negetive\n",
    "                3rd row : 2nd column :Actual Sentiment = Neutral ; Predicted Sentiment = Positive\n",
    "                3rd row : 3rd column :Actual Sentiment = Neutral ; Predicted Sentiment = Neutral"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[1973  196  237]\n",
      " [ 182 2191  269]\n",
      " [  75   52 2162]]\n"
     ]
    }
   ],
   "source": [
    "cm  = confusion_matrix(Sentiment_test,predictions )\n",
    "print(cm)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Precision Recall & F1 Score:\n",
    "\n",
    "    Precison = True Postive/(Total predicted Positive) || True Postive/(True Positive + False Positive)\n",
    "               Precision talks about how precise/accurate the model is, i.e. out of total predicted positive,\n",
    "               how many of them are actual positive.\n",
    "               \n",
    "    Recall = True Positive/(Total actual Positive) || True Postive/(True Positive + False Negetive)\n",
    "             Recall calculates how many of the Actual Positives the model captures by labeling it \n",
    "             as Positive (True Positive)\n",
    "             \n",
    "    F1 Score = (2 * Precision * Recall) / (Precision + Recall)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision of the Classifier Model : 0.864\n",
      "Recall of the Classifier Model : 0.865\n",
      "F1 Score of the Classifier Model : 0.862\n"
     ]
    }
   ],
   "source": [
    "precision = precision_score(Sentiment_test, predictions, average='macro')\n",
    "print('Precision of the Classifier Model : %.3f' % precision)\n",
    "recall = recall_score(Sentiment_test, predictions, average='macro')\n",
    "print('Recall of the Classifier Model : %.3f' %recall)\n",
    "f1 = f1_score(Sentiment_test, predictions, average='macro')\n",
    "print('F1 Score of the Classifier Model : %.3f' %f1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# PREDICT SENTIMENT FROM CUSTOM TWEETS\n",
    "    (Tweets not in the dataset , neither in training nor in test set)\n",
    "\n",
    "    Test case 1:\n",
    "    \"Donald trump , i have no idea how this man became the president of the country #STUPID #WORST_PRESIDENT\",\n",
    "     \"@iamrohit you were right ...Kolkata is rightfully called the City of Joy..\",\n",
    "     \"Hell yeah,, @Rogerfedrrer win his 20th Grand Slam #legend\",\n",
    "     \"The food could not have been more bad, One of the worst restaurant i have dined at , the food teatsed like shit\",\n",
    "     \"Machine learning is one of the most intersteing feilds to work on\",\n",
    "     \"The movie was a bit boring and not so intersting for me also it was slow, overall it was below average !!! #Chappak\",\n",
    "     \"its a sad night for @RealMadridCF we lost the game 2-1, but we will come back stronger next time #Hala_Madrid\",\n",
    "     \"@SDER You are a murderer, you deserverd to be punished\",\n",
    "     \"Tears of Joy !! Finally, @IndianCricket team are the champions of the world 😍 #WINNERS #WORLDCHAPMPS\",\n",
    "     \"THE FOOD WAS Average 😊 😊\",\n",
    "     \"THE FOOD WAS Average 😢 😢 \",\n",
    "     \"rest in peace !! #RIP 😢 😢\"\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "User_Tweets = [\"Donald trump , i have no idea how this man became the president of the country #STUPID #WORST_PRESIDENT\",\n",
    "             \"@iamrohit you were right ...Kolkata is rightfully called the City of Joy..\",\n",
    "             \"Hell yeah,, @Rogerfedrrer win his 20th Grand Slam #legend\",\n",
    "             \" The food could not have been more bad, One of the worst restaurant i have dined at , the food teatsed like shit\",\n",
    "             \"Machine learning is one of the most intersteing feilds to work on\",\n",
    "             \"The movie was a bit boring and not so intersting for me also it was slow, overall it was below average !!! #Chappak\",\n",
    "             \"its a sad night for @RealMadridCF we lost the game 2-1, but we will come back stronger next time #Hala_Madrid\",\n",
    "             \"@SDER You are a murderer, you deserverd to be punished\",\n",
    "             \"Tears of Joy !! Finally, @IndianCricket team are the champions of the world 😍 #WINNERS #WORLDCHAPMPS\",\n",
    "             \"THE FOOD WAS Average 😊 😊\",\n",
    "             \"THE FOOD WAS Average 😢 😢 \",\n",
    "             \"rest in peace !! #RIP 😢 😢\" ]\n",
    "\n",
    "IP_Tweets =  pd.DataFrame (User_Tweets,columns=['Input_Tweets'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 12 entries, 0 to 11\n",
      "Data columns (total 1 columns):\n",
      "Input_Tweets    12 non-null object\n",
      "dtypes: object(1)\n",
      "memory usage: 176.0+ bytes\n"
     ]
    }
   ],
   "source": [
    "IP_Tweets.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Input_Tweets</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Donald trump , i have no idea how this man became the president of the country #STUPID #WORST_PRESIDENT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>@iamrohit you were right ...Kolkata is rightfully called the City of Joy..</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Hell yeah,, @Rogerfedrrer win his 20th Grand Slam #legend</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>The food could not have been more bad, One of the worst restaurant i have dined at , the food teatsed like shit</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Machine learning is one of the most intersteing feilds to work on</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>The movie was a bit boring and not so intersting for me also it was slow, overall it was below average !!! #Chappak</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>its a sad night for @RealMadridCF we lost the game 2-1, but we will come back stronger next time #Hala_Madrid</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>@SDER You are a murderer, you deserverd to be punished</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Tears of Joy !! Finally, @IndianCricket team are the champions of the world 😍 #WINNERS #WORLDCHAPMPS</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>THE FOOD WAS Average 😊 😊</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>THE FOOD WAS Average 😢 😢</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>rest in peace !! #RIP 😢 😢</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                                                           Input_Tweets\n",
       "0   Donald trump , i have no idea how this man became the president of the country #STUPID #WORST_PRESIDENT            \n",
       "1   @iamrohit you were right ...Kolkata is rightfully called the City of Joy..                                         \n",
       "2   Hell yeah,, @Rogerfedrrer win his 20th Grand Slam #legend                                                          \n",
       "3    The food could not have been more bad, One of the worst restaurant i have dined at , the food teatsed like shit   \n",
       "4   Machine learning is one of the most intersteing feilds to work on                                                  \n",
       "5   The movie was a bit boring and not so intersting for me also it was slow, overall it was below average !!! #Chappak\n",
       "6   its a sad night for @RealMadridCF we lost the game 2-1, but we will come back stronger next time #Hala_Madrid      \n",
       "7   @SDER You are a murderer, you deserverd to be punished                                                             \n",
       "8   Tears of Joy !! Finally, @IndianCricket team are the champions of the world 😍 #WINNERS #WORLDCHAPMPS               \n",
       "9   THE FOOD WAS Average 😊 😊                                                                                           \n",
       "10  THE FOOD WAS Average 😢 😢                                                                                           \n",
       "11  rest in peace !! #RIP 😢 😢                                                                                          "
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "IP_Tweets"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Pre Process the tweets and convert them into vectors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Input_Tweets</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>donald trump i have no idea how this man became the president of the country stupid worstpresident</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>you were right kolkata is rightfully called the city of joy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>hell yeah win his th grand slam legend</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>the food could not have been more bad one of the worst restaurant i have dined at the food teatsed like shit</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>machine learning is one of the most intersteing feilds to work on</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>the movie was a bit boring and not so intersting for me also it was slow overall it was below average chappak</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>its a sad night for we lost the game but we will come back stronger next time halamadrid</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>you are a murderer you deserverd to be punished</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>tears of joy finally team are the champions of the world 😍 winners worldchapmps</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>the food was average 😊 😊</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>the food was average 😢 😢</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>rest in peace rip 😢 😢</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                                                     Input_Tweets\n",
       "0   donald trump i have no idea how this man became the president of the country stupid worstpresident           \n",
       "1   you were right kolkata is rightfully called the city of joy                                                  \n",
       "2   hell yeah win his th grand slam legend                                                                       \n",
       "3   the food could not have been more bad one of the worst restaurant i have dined at the food teatsed like shit \n",
       "4   machine learning is one of the most intersteing feilds to work on                                            \n",
       "5   the movie was a bit boring and not so intersting for me also it was slow overall it was below average chappak\n",
       "6   its a sad night for we lost the game but we will come back stronger next time halamadrid                     \n",
       "7   you are a murderer you deserverd to be punished                                                              \n",
       "8   tears of joy finally team are the champions of the world 😍 winners worldchapmps                              \n",
       "9   the food was average 😊 😊                                                                                     \n",
       "10  the food was average 😢 😢                                                                                     \n",
       "11  rest in peace rip 😢 😢                                                                                        "
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "IP_Tweets['Input_Tweets'] = IP_Tweets['Input_Tweets'].apply(cleanTxt)\n",
    "IP_Tweets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "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>Input_Tweets</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>donald trump i have no idea how this man became the president of the country stupid worstpresident</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>you were right kolkata is rightfully called the city of joy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>hell yeah win his th grand slam legend</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>the food could not have been more bad one of the worst restaurant i have dined at the food teatsed like shit</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>machine learning is one of the most intersteing feilds to work on</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>the movie was a bit boring and not so intersting for me also it was slow overall it was below average chappak</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>its a sad night for we lost the game but we will come back stronger next time halamadrid</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>you are a murderer you deserverd to be punished</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>tears of joy finally team are the champions of the world smiling face with heart eyes winners worldchapmps</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>the food was average smiling face with smiling eyes smiling face with smiling eyes</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>the food was average crying face crying face</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>rest in peace rip crying face crying face</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                                                     Input_Tweets\n",
       "0   donald trump i have no idea how this man became the president of the country stupid worstpresident           \n",
       "1   you were right kolkata is rightfully called the city of joy                                                  \n",
       "2   hell yeah win his th grand slam legend                                                                       \n",
       "3   the food could not have been more bad one of the worst restaurant i have dined at the food teatsed like shit \n",
       "4   machine learning is one of the most intersteing feilds to work on                                            \n",
       "5   the movie was a bit boring and not so intersting for me also it was slow overall it was below average chappak\n",
       "6   its a sad night for we lost the game but we will come back stronger next time halamadrid                     \n",
       "7   you are a murderer you deserverd to be punished                                                              \n",
       "8   tears of joy finally team are the champions of the world smiling face with heart eyes winners worldchapmps   \n",
       "9   the food was average smiling face with smiling eyes smiling face with smiling eyes                           \n",
       "10  the food was average crying face crying face                                                                 \n",
       "11  rest in peace rip crying face crying face                                                                    "
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "IP_Tweets['Input_Tweets'] = IP_Tweets['Input_Tweets'].apply(convertEmoji)\n",
    "IP_Tweets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Input_Tweets</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>donald trump idea man became president country stupid worstpresident</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>right kolkata rightfully called city joy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>hell yeah win th grand slam legend</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>food could bad one bad restaurant dined food teatsed like shit</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>machine learning one intersteing feilds work</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>movie bit boring intersting also slow overall average chappak</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>sad night lost game come back strong next time halamadrid</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>murderer deserverd punished</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>tears joy finally team champions world smiling face heart eyes winners worldchapmps</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>food average smiling face smiling eyes smiling face smiling eyes</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>food average crying face crying face</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>rest peace rip crying face crying face</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                           Input_Tweets\n",
       "0   donald trump idea man became president country stupid worstpresident               \n",
       "1   right kolkata rightfully called city joy                                           \n",
       "2   hell yeah win th grand slam legend                                                 \n",
       "3   food could bad one bad restaurant dined food teatsed like shit                     \n",
       "4   machine learning one intersteing feilds work                                       \n",
       "5   movie bit boring intersting also slow overall average chappak                      \n",
       "6   sad night lost game come back strong next time halamadrid                          \n",
       "7   murderer deserverd punished                                                        \n",
       "8   tears joy finally team champions world smiling face heart eyes winners worldchapmps\n",
       "9   food average smiling face smiling eyes smiling face smiling eyes                   \n",
       "10  food average crying face crying face                                               \n",
       "11  rest peace rip crying face crying face                                             "
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "IP_Tweets['Input_Tweets'] = IP_Tweets['Input_Tweets'].apply(Tokenization)\n",
    "IP_Tweets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['donald trump idea man became president country stupid worstpresident',\n",
       "       'right kolkata rightfully called city joy',\n",
       "       'hell yeah win th grand slam legend',\n",
       "       'food could bad one bad restaurant dined food teatsed like shit',\n",
       "       'machine learning one intersteing feilds work',\n",
       "       'movie bit boring intersting also slow overall average chappak',\n",
       "       'sad night lost game come back strong next time halamadrid',\n",
       "       'murderer deserverd punished',\n",
       "       'tears joy finally team champions world smiling face heart eyes winners worldchapmps',\n",
       "       'food average smiling face smiling eyes smiling face smiling eyes',\n",
       "       'food average crying face crying face',\n",
       "       'rest peace rip crying face crying face'], dtype=object)"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "IPtweets_array = IP_Tweets['Input_Tweets'].values\n",
    "IPtweets_array"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "Vectorized_IPTweets = count_vectorizer.transform(IPtweets_array)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# predict Sentiment of these tweets using their coressponding tweet vectors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred = clf.predict(Vectorized_IPTweets)\n",
    "pred = pred.tolist() # to covert 0,1,2 to pos neg and nuetral (printing output purpose)\n",
    "\n",
    "for i in range(len(pred)):\n",
    "    if(pred[i]==0):\n",
    "        pred[i]=('Negetive')\n",
    "    elif(pred[i]==1):\n",
    "        pred[i]='Positive'\n",
    "    else:\n",
    "        pred[i]='Neutral'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Check Predictions of each given Tweet"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The sentiment for review :\n",
      " Donald trump , i have no idea how this man became the president of the country #STUPID #WORST_PRESIDENT is :\n",
      " Negetive \n",
      "\n",
      "The sentiment for review :\n",
      " @iamrohit you were right ...Kolkata is rightfully called the City of Joy.. is :\n",
      " Positive \n",
      "\n",
      "The sentiment for review :\n",
      " Hell yeah,, @Rogerfedrrer win his 20th Grand Slam #legend is :\n",
      " Positive \n",
      "\n",
      "The sentiment for review :\n",
      "  The food could not have been more bad, One of the worst restaurant i have dined at , the food teatsed like shit is :\n",
      " Negetive \n",
      "\n",
      "The sentiment for review :\n",
      " Machine learning is one of the most intersteing feilds to work on is :\n",
      " Neutral \n",
      "\n",
      "The sentiment for review :\n",
      " The movie was a bit boring and not so intersting for me also it was slow, overall it was below average !!! #Chappak is :\n",
      " Negetive \n",
      "\n",
      "The sentiment for review :\n",
      " its a sad night for @RealMadridCF we lost the game 2-1, but we will come back stronger next time #Hala_Madrid is :\n",
      " Negetive \n",
      "\n",
      "The sentiment for review :\n",
      " @SDER You are a murderer, you deserverd to be punished is :\n",
      " Neutral \n",
      "\n",
      "The sentiment for review :\n",
      " Tears of Joy !! Finally, @IndianCricket team are the champions of the world 😍 #WINNERS #WORLDCHAPMPS is :\n",
      " Positive \n",
      "\n",
      "The sentiment for review :\n",
      " THE FOOD WAS Average 😊 😊 is :\n",
      " Neutral \n",
      "\n",
      "The sentiment for review :\n",
      " THE FOOD WAS Average 😢 😢  is :\n",
      " Negetive \n",
      "\n",
      "The sentiment for review :\n",
      " rest in peace !! #RIP 😢 😢 is :\n",
      " Negetive \n",
      "\n"
     ]
    }
   ],
   "source": [
    "for i in range (len(User_Tweets)):\n",
    "    print(\"The sentiment for review :\\n\",User_Tweets[i],\"is :\\n\",pred[i],\"\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}