test_real_data.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Code for pulling tweets and classifying the tweets\n",
    "# author - ADITYA SINGH\n",
    "# email -singhaditya08aug@gmail.com"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# import the dependencies\n",
    "import keras\n",
    "from keras import backend as k\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import re\n",
    "import nltk\n",
    "from nltk import PorterStemmer\n",
    "from nltk.tokenize import word_tokenize\n",
    "import time\n",
    "import pickle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# reading the data\n",
    "train_data = pd.read_csv('data/training.1600000.processed.noemoticon.csv', encoding = 'ISO-8859-1')"
   ]
  },
  {
   "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>0</th>\n",
       "      <th>1467810369</th>\n",
       "      <th>Mon Apr 06 22:19:45 PDT 2009</th>\n",
       "      <th>NO_QUERY</th>\n",
       "      <th>_TheSpecialOne_</th>\n",
       "      <th>@switchfoot http://twitpic.com/2y1zl - Awww, that's a bummer.  You shoulda got David Carr of Third Day to do it. ;D</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>1467810672</td>\n",
       "      <td>Mon Apr 06 22:19:49 PDT 2009</td>\n",
       "      <td>NO_QUERY</td>\n",
       "      <td>scotthamilton</td>\n",
       "      <td>is upset that he can't update his Facebook by ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>1467810917</td>\n",
       "      <td>Mon Apr 06 22:19:53 PDT 2009</td>\n",
       "      <td>NO_QUERY</td>\n",
       "      <td>mattycus</td>\n",
       "      <td>@Kenichan I dived many times for the ball. Man...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>1467811184</td>\n",
       "      <td>Mon Apr 06 22:19:57 PDT 2009</td>\n",
       "      <td>NO_QUERY</td>\n",
       "      <td>ElleCTF</td>\n",
       "      <td>my whole body feels itchy and like its on fire</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>1467811193</td>\n",
       "      <td>Mon Apr 06 22:19:57 PDT 2009</td>\n",
       "      <td>NO_QUERY</td>\n",
       "      <td>Karoli</td>\n",
       "      <td>@nationwideclass no, it's not behaving at all....</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>1467811372</td>\n",
       "      <td>Mon Apr 06 22:20:00 PDT 2009</td>\n",
       "      <td>NO_QUERY</td>\n",
       "      <td>joy_wolf</td>\n",
       "      <td>@Kwesidei not the whole crew</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0</td>\n",
       "      <td>1467811592</td>\n",
       "      <td>Mon Apr 06 22:20:03 PDT 2009</td>\n",
       "      <td>NO_QUERY</td>\n",
       "      <td>mybirch</td>\n",
       "      <td>Need a hug</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>0</td>\n",
       "      <td>1467811594</td>\n",
       "      <td>Mon Apr 06 22:20:03 PDT 2009</td>\n",
       "      <td>NO_QUERY</td>\n",
       "      <td>coZZ</td>\n",
       "      <td>@LOLTrish hey  long time no see! Yes.. Rains a...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>0</td>\n",
       "      <td>1467811795</td>\n",
       "      <td>Mon Apr 06 22:20:05 PDT 2009</td>\n",
       "      <td>NO_QUERY</td>\n",
       "      <td>2Hood4Hollywood</td>\n",
       "      <td>@Tatiana_K nope they didn't have it</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>0</td>\n",
       "      <td>1467812025</td>\n",
       "      <td>Mon Apr 06 22:20:09 PDT 2009</td>\n",
       "      <td>NO_QUERY</td>\n",
       "      <td>mimismo</td>\n",
       "      <td>@twittera que me muera ?</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>0</td>\n",
       "      <td>1467812416</td>\n",
       "      <td>Mon Apr 06 22:20:16 PDT 2009</td>\n",
       "      <td>NO_QUERY</td>\n",
       "      <td>erinx3leannexo</td>\n",
       "      <td>spring break in plain city... it's snowing</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   0  1467810369  Mon Apr 06 22:19:45 PDT 2009  NO_QUERY  _TheSpecialOne_  \\\n",
       "0  0  1467810672  Mon Apr 06 22:19:49 PDT 2009  NO_QUERY    scotthamilton   \n",
       "1  0  1467810917  Mon Apr 06 22:19:53 PDT 2009  NO_QUERY         mattycus   \n",
       "2  0  1467811184  Mon Apr 06 22:19:57 PDT 2009  NO_QUERY          ElleCTF   \n",
       "3  0  1467811193  Mon Apr 06 22:19:57 PDT 2009  NO_QUERY           Karoli   \n",
       "4  0  1467811372  Mon Apr 06 22:20:00 PDT 2009  NO_QUERY         joy_wolf   \n",
       "5  0  1467811592  Mon Apr 06 22:20:03 PDT 2009  NO_QUERY          mybirch   \n",
       "6  0  1467811594  Mon Apr 06 22:20:03 PDT 2009  NO_QUERY             coZZ   \n",
       "7  0  1467811795  Mon Apr 06 22:20:05 PDT 2009  NO_QUERY  2Hood4Hollywood   \n",
       "8  0  1467812025  Mon Apr 06 22:20:09 PDT 2009  NO_QUERY          mimismo   \n",
       "9  0  1467812416  Mon Apr 06 22:20:16 PDT 2009  NO_QUERY   erinx3leannexo   \n",
       "\n",
       "  @switchfoot http://twitpic.com/2y1zl - Awww, that's a bummer.  You shoulda got David Carr of Third Day to do it. ;D  \n",
       "0  is upset that he can't update his Facebook by ...                                                                   \n",
       "1  @Kenichan I dived many times for the ball. Man...                                                                   \n",
       "2    my whole body feels itchy and like its on fire                                                                    \n",
       "3  @nationwideclass no, it's not behaving at all....                                                                   \n",
       "4                      @Kwesidei not the whole crew                                                                    \n",
       "5                                        Need a hug                                                                    \n",
       "6  @LOLTrish hey  long time no see! Yes.. Rains a...                                                                   \n",
       "7               @Tatiana_K nope they didn't have it                                                                    \n",
       "8                          @twittera que me muera ?                                                                    \n",
       "9        spring break in plain city... it's snowing                                                                    "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# showing the structure of the data\n",
    "train_data.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "no of rows in the data set :  1599999\n"
     ]
    }
   ],
   "source": [
    "# no of rows in the dataset\n",
    "rows = train_data.count()\n",
    "print('no of rows in the data set : ', rows[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "# preprocessing the input text\n",
    "# note that we swithched off the emoji because sentiment140 dataset does not have any emoji\n",
    "SentimentText = []\n",
    "def preprocess(train_data):\n",
    "    i = 0\n",
    "    while i <rows[0]:\n",
    "        text =train_data.ix[:,5][i]\n",
    "        text =text.lower()\n",
    "        text = re.sub(r'@[A-Za-z0-9]+','',text)\n",
    "        text = re.sub(\"[^a-zA-Z]\", \" \", text)\n",
    "        #emoji is not present in the dataset\n",
    "        #text =emoji_pattern.sub(r'', text)\n",
    "        text = re.sub(r'[^\\w\\s]','',text)\n",
    "        text = re.sub(r'(.)\\1+', r'\\1\\1', text)\n",
    "        tokens = word_tokenize(text)\n",
    "        new_text =\"\"\n",
    "        for token in tokens:\n",
    "            new_text += PorterStemmer().stem(token)\n",
    "            new_text +=' ' \n",
    "        SentimentText.append(new_text)\n",
    "        \n",
    "        if i%100000 == 0:\n",
    "            print('processed :' , i , 'tweets')\n",
    "        i = i + 1\n",
    "    print('Done PreProcessing the text' )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "processed : 0 tweets\n",
      "processed : 100000 tweets\n",
      "processed : 200000 tweets\n",
      "processed : 300000 tweets\n",
      "processed : 400000 tweets\n",
      "processed : 500000 tweets\n",
      "processed : 600000 tweets\n",
      "processed : 700000 tweets\n",
      "processed : 800000 tweets\n",
      "processed : 900000 tweets\n",
      "processed : 1000000 tweets\n",
      "processed : 1100000 tweets\n",
      "processed : 1200000 tweets\n",
      "processed : 1300000 tweets\n",
      "processed : 1400000 tweets\n",
      "processed : 1500000 tweets\n",
      "Done PreProcessing the text\n",
      "time taken :  26.7050218184789  mins\n"
     ]
    }
   ],
   "source": [
    "# finially calling the preprocessing function\n",
    "curr_time = time.time()\n",
    "preprocess(train_data)\n",
    "new_time = time.time()\n",
    "\n",
    "time_taken = (new_time - curr_time)/60\n",
    "print('time taken : ',time_taken, ' mins')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pickle\n",
    "with open(\"training.1600000.processed.noemoticon.txt\", \"wb\") as fp:\n",
    "    pickle.dump(SentimentText, fp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Finding the unigram representation\n",
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "vectorizer=CountVectorizer()\n",
    "\n",
    "# #Finding the bigram representation \n",
    "bigram_vectorizer=CountVectorizer(ngram_range=(1,2))\n",
    "\n",
    "# finding the trigram representation\n",
    "trigram_vectorizer =CountVectorizer(ngram_range=(1,3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('sentiments/training.1600000.processed.noemoticon.txt', 'rb') as f:\n",
    "    SentimentText = pickle.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# our features that we will be using\n",
    "X_unigrams = vectorizer.fit_transform(SentimentText)\n",
    "X_bigrams = bigram_vectorizer.fit_transform(SentimentText)\n",
    "X_trigrams = trigram_vectorizer.fit_transform(SentimentText)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "from scipy.sparse import hstack \n",
    "X_combined = hstack([X_unigrams , X_bigrams, X_trigrams])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1599999, 15503914)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# combined uni_bi+trigrams that showed best performance for ME classifier\n",
    "X_combined.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\ADITYA\\Anaconda3\\envs\\ml\\lib\\site-packages\\ipykernel_launcher.py:1: DeprecationWarning: \n",
      ".ix is deprecated. Please use\n",
      ".loc for label based indexing or\n",
      ".iloc for positional indexing\n",
      "\n",
      "See the documentation here:\n",
      "http://pandas.pydata.org/pandas-docs/stable/indexing.html#ix-indexer-is-deprecated\n",
      "  \"\"\"Entry point for launching an IPython kernel.\n"
     ]
    }
   ],
   "source": [
    "# target values\n",
    "y =train_data.ix[:,0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn import linear_model\n",
    "model =  linear_model.LogisticRegression(C=1,solver ='saga',penalty = 'l2',max_iter = 50000,n_jobs=4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LogisticRegression(C=1, class_weight=None, dual=False, fit_intercept=True,\n",
       "          intercept_scaling=1, max_iter=50000, multi_class='ovr', n_jobs=4,\n",
       "          penalty='l2', random_state=None, solver='saga', tol=0.0001,\n",
       "          verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# fitting the model on the training data set\n",
    "model.fit(X_combined,y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "it took me about 30hrs to train this so i didnt have any option to try different model to increase the accuracy so i used ME classifier with uni grams , bi grams and tri grams features that showed best results for previous cases but try for other approached that can increase the accuracy "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "#we are saving the model for further use\n",
    "pkl_filename = \"model.pkl\"  \n",
    "with open(pkl_filename, 'wb') as file:  \n",
    "    pickle.dump(model, file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "#now we will run the included pull_tweets.py file to pull the tweets from a paticular hashtag"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "no of rows in the data set :  470\n"
     ]
    }
   ],
   "source": [
    "# reading the tweets file\n",
    "tweets =pd.read_csv('tweets.csv')\n",
    "# no of rows in the file\n",
    "rows = tweets.count()\n",
    "print('no of rows in the data set : ', rows[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\ADITYA\\Anaconda3\\envs\\ml\\lib\\site-packages\\ipykernel_launcher.py:11: DeprecationWarning: \n",
      ".ix is deprecated. Please use\n",
      ".loc for label based indexing or\n",
      ".iloc for positional indexing\n",
      "\n",
      "See the documentation here:\n",
      "http://pandas.pydata.org/pandas-docs/stable/indexing.html#ix-indexer-is-deprecated\n",
      "  # This is added back by InteractiveShellApp.init_path()\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "processed : 0 tweets\n",
      "processed : 100 tweets\n",
      "processed : 200 tweets\n",
      "processed : 300 tweets\n",
      "processed : 400 tweets\n",
      "Done PreProcessing the text\n"
     ]
    }
   ],
   "source": [
    "# preprocessing the newely aquired tweets\n",
    "i = 0\n",
    "emoji_pattern = re.compile(\"[\"\n",
    "        u\"\\U0001F600-\\U0001F64F\"  # emoticons\n",
    "        u\"\\U0001F300-\\U0001F5FF\"  # symbols & pictographs\n",
    "        u\"\\U0001F680-\\U0001F6FF\"  # transport & map symbols\n",
    "        u\"\\U0001F1E0-\\U0001F1FF\"  # flags (iOS)\n",
    "                           \"]+\", flags=re.UNICODE)\n",
    "SentimentText = []\n",
    "while i <rows[0]:\n",
    "    text =tweets.ix[:,1][i]\n",
    "    text =text.lower()\n",
    "    text = re.sub(r'@[A-Za-z0-9]+','',text)\n",
    "    text = re.sub(\"[^a-zA-Z]\", \" \", text)\n",
    "    #emoji is not present in the dataset\n",
    "    text =emoji_pattern.sub(r'', text)\n",
    "    text = re.sub(r'[^\\w\\s]','',text)\n",
    "    text = re.sub(r'(.)\\1+', r'\\1\\1', text)\n",
    "    tokens = word_tokenize(text)\n",
    "    new_text =\"\"\n",
    "    for token in tokens:\n",
    "        new_text += PorterStemmer().stem(token)\n",
    "        new_text +=' ' \n",
    "    SentimentText.append(new_text)\n",
    "\n",
    "    if i%100 == 0:\n",
    "        print('processed :' , i , 'tweets')\n",
    "    i = i + 1\n",
    "print('Done PreProcessing the text' )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [],
   "source": [
    "# now we will extract features from the cleaned text\n",
    "X_test_unigrams = vectorizer.transform(SentimentText)\n",
    "X_test_bigrams = bigram_vectorizer.transform(SentimentText)\n",
    "X_test_trigrams = trigram_vectorizer.transform(SentimentText)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [],
   "source": [
    "from scipy.sparse import hstack\n",
    "X_test_combined = hstack([X_test_unigrams , X_test_bigrams,X_test_trigrams])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(470, 15503914)"
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# checking the shape of the input\n",
    "X_test_combined.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [],
   "source": [
    "# finally predicting the values\n",
    "predicted = model.predict(X_test_combined)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [],
   "source": [
    "#counting the no of 0s 2s and 4s\n",
    "i = 0\n",
    "count_0 = 0\n",
    "count_2 = 0\n",
    "count_4 = 0\n",
    "while i<rows[0]:\n",
    "    if predicted[i] == 0:\n",
    "        count_0  = count_0 +1\n",
    "    elif predicted[i] == 2:\n",
    "        count_2 = count_2 +1\n",
    "    else:\n",
    "        count_4 = count_4 +1\n",
    "    i = i +1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "68\n",
      "0\n",
      "402\n"
     ]
    }
   ],
   "source": [
    "# printing the results for ramdaan\n",
    "print(count_0)\n",
    "print(count_2)\n",
    "print(count_4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2379c545f98>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# now we will construct a pie char to visualize the collected data\n",
    "import matplotlib.pyplot as plt\n",
    " \n",
    "labels = ['negative', 'neutral', 'positive']\n",
    "sizes = [count_0, count_2, count_4]\n",
    "colors = ['yellowgreen', 'gold', 'lightskyblue']\n",
    "patches, texts = plt.pie(sizes, colors=colors, shadow=True, startangle=90)\n",
    "plt.legend(patches, labels, loc=\"best\")\n",
    "plt.axis('equal')\n",
    "plt.tight_layout()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "no of rows in the data set :  3139\n"
     ]
    }
   ],
   "source": [
    "# reading the tweets file\n",
    "tweets =pd.read_csv('tweets_SaveDemocracy.csv')\n",
    "# no of rows in the file\n",
    "rows = tweets.count()\n",
    "print('no of rows in the data set : ', rows[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\ADITYA\\Anaconda3\\envs\\ml\\lib\\site-packages\\ipykernel_launcher.py:11: DeprecationWarning: \n",
      ".ix is deprecated. Please use\n",
      ".loc for label based indexing or\n",
      ".iloc for positional indexing\n",
      "\n",
      "See the documentation here:\n",
      "http://pandas.pydata.org/pandas-docs/stable/indexing.html#ix-indexer-is-deprecated\n",
      "  # This is added back by InteractiveShellApp.init_path()\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "processed : 0 tweets\n",
      "processed : 1000 tweets\n",
      "processed : 2000 tweets\n",
      "processed : 3000 tweets\n",
      "Done PreProcessing the text\n"
     ]
    }
   ],
   "source": [
    "# preprocessing the newely aquired tweets\n",
    "i = 0\n",
    "emoji_pattern = re.compile(\"[\"\n",
    "        u\"\\U0001F600-\\U0001F64F\"  # emoticons\n",
    "        u\"\\U0001F300-\\U0001F5FF\"  # symbols & pictographs\n",
    "        u\"\\U0001F680-\\U0001F6FF\"  # transport & map symbols\n",
    "        u\"\\U0001F1E0-\\U0001F1FF\"  # flags (iOS)\n",
    "                           \"]+\", flags=re.UNICODE)\n",
    "SentimentText = []\n",
    "while i <rows[0]:\n",
    "    text =tweets.ix[:,1][i]\n",
    "    text =text.lower()\n",
    "    text = re.sub(r'@[A-Za-z0-9]+','',text)\n",
    "    text = re.sub(\"[^a-zA-Z]\", \" \", text)\n",
    "    #emoji is not present in the dataset\n",
    "    text =emoji_pattern.sub(r'', text)\n",
    "    text = re.sub(r'[^\\w\\s]','',text)\n",
    "    text = re.sub(r'(.)\\1+', r'\\1\\1', text)\n",
    "    tokens = word_tokenize(text)\n",
    "    new_text =\"\"\n",
    "    for token in tokens:\n",
    "        new_text += PorterStemmer().stem(token)\n",
    "        new_text +=' ' \n",
    "    SentimentText.append(new_text)\n",
    "\n",
    "    if i%1000 == 0:\n",
    "        print('processed :' , i , 'tweets')\n",
    "    i = i + 1\n",
    "print('Done PreProcessing the text' )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [],
   "source": [
    "# now we will extract features from the cleaned text\n",
    "X_test_unigrams = vectorizer.transform(SentimentText)\n",
    "X_test_bigrams = bigram_vectorizer.transform(SentimentText)\n",
    "X_test_trigrams = trigram_vectorizer.transform(SentimentText)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [],
   "source": [
    "from scipy.sparse import hstack\n",
    "X_test_combined = hstack([X_test_unigrams , X_test_bigrams,X_test_trigrams])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(3139, 15503914)"
      ]
     },
     "execution_count": 94,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# checking the shape of the input\n",
    "X_test_combined.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [],
   "source": [
    "# finally predicting the values\n",
    "predicted = model.predict(X_test_combined)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [],
   "source": [
    "#counting the no of 0s 2s and 4s\n",
    "i = 0\n",
    "count_0 = 0\n",
    "count_2 = 0\n",
    "count_4 = 0\n",
    "while i<rows[0]:\n",
    "    if predicted[i] == 0:\n",
    "        count_0  = count_0 +1\n",
    "    elif predicted[i] == 2:\n",
    "        count_2 = count_2 +1\n",
    "    else:\n",
    "        count_4 = count_4 +1\n",
    "    i = i +1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "650\n",
      "0\n",
      "2489\n"
     ]
    }
   ],
   "source": [
    "# printing the results for net neturality\n",
    "print(count_0)\n",
    "print(count_2)\n",
    "print(count_4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2378f2ffe80>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# now we will construct a pie char to visualize the collected data\n",
    "import matplotlib.pyplot as plt\n",
    " \n",
    "labels = ['negative', 'neutral', 'positive']\n",
    "sizes = [count_0, count_2, count_4]\n",
    "colors = ['yellowgreen', 'gold', 'lightskyblue']\n",
    "patches, texts = plt.pie(sizes, colors=colors, shadow=True, startangle=90)\n",
    "plt.legend(patches, labels, loc=\"best\")\n",
    "plt.axis('equal')\n",
    "plt.tight_layout()"
   ]
  }
 ],
 "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": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}