Fake News.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Data Import"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "True_news = pd.read_csv('True.csv')\n",
    "Fake_news = pd.read_csv('Fake.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "True_news['label'] = 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "Fake_news['label'] = 1"
   ]
  },
  {
   "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>title</th>\n",
       "      <th>text</th>\n",
       "      <th>subject</th>\n",
       "      <th>date</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>As U.S. budget fight looms, Republicans flip t...</td>\n",
       "      <td>WASHINGTON (Reuters) - The head of a conservat...</td>\n",
       "      <td>politicsNews</td>\n",
       "      <td>December 31, 2017</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>U.S. military to accept transgender recruits o...</td>\n",
       "      <td>WASHINGTON (Reuters) - Transgender people will...</td>\n",
       "      <td>politicsNews</td>\n",
       "      <td>December 29, 2017</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Senior U.S. Republican senator: 'Let Mr. Muell...</td>\n",
       "      <td>WASHINGTON (Reuters) - The special counsel inv...</td>\n",
       "      <td>politicsNews</td>\n",
       "      <td>December 31, 2017</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>FBI Russia probe helped by Australian diplomat...</td>\n",
       "      <td>WASHINGTON (Reuters) - Trump campaign adviser ...</td>\n",
       "      <td>politicsNews</td>\n",
       "      <td>December 30, 2017</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Trump wants Postal Service to charge 'much mor...</td>\n",
       "      <td>SEATTLE/WASHINGTON (Reuters) - President Donal...</td>\n",
       "      <td>politicsNews</td>\n",
       "      <td>December 29, 2017</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                               title  \\\n",
       "0  As U.S. budget fight looms, Republicans flip t...   \n",
       "1  U.S. military to accept transgender recruits o...   \n",
       "2  Senior U.S. Republican senator: 'Let Mr. Muell...   \n",
       "3  FBI Russia probe helped by Australian diplomat...   \n",
       "4  Trump wants Postal Service to charge 'much mor...   \n",
       "\n",
       "                                                text       subject  \\\n",
       "0  WASHINGTON (Reuters) - The head of a conservat...  politicsNews   \n",
       "1  WASHINGTON (Reuters) - Transgender people will...  politicsNews   \n",
       "2  WASHINGTON (Reuters) - The special counsel inv...  politicsNews   \n",
       "3  WASHINGTON (Reuters) - Trump campaign adviser ...  politicsNews   \n",
       "4  SEATTLE/WASHINGTON (Reuters) - President Donal...  politicsNews   \n",
       "\n",
       "                 date  label  \n",
       "0  December 31, 2017       0  \n",
       "1  December 29, 2017       0  \n",
       "2  December 31, 2017       0  \n",
       "3  December 30, 2017       0  \n",
       "4  December 29, 2017       0  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "True_news.head()"
   ]
  },
  {
   "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>title</th>\n",
       "      <th>text</th>\n",
       "      <th>subject</th>\n",
       "      <th>date</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Donald Trump Sends Out Embarrassing New Year’...</td>\n",
       "      <td>Donald Trump just couldn t wish all Americans ...</td>\n",
       "      <td>News</td>\n",
       "      <td>December 31, 2017</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Drunk Bragging Trump Staffer Started Russian ...</td>\n",
       "      <td>House Intelligence Committee Chairman Devin Nu...</td>\n",
       "      <td>News</td>\n",
       "      <td>December 31, 2017</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Sheriff David Clarke Becomes An Internet Joke...</td>\n",
       "      <td>On Friday, it was revealed that former Milwauk...</td>\n",
       "      <td>News</td>\n",
       "      <td>December 30, 2017</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Trump Is So Obsessed He Even Has Obama’s Name...</td>\n",
       "      <td>On Christmas day, Donald Trump announced that ...</td>\n",
       "      <td>News</td>\n",
       "      <td>December 29, 2017</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Pope Francis Just Called Out Donald Trump Dur...</td>\n",
       "      <td>Pope Francis used his annual Christmas Day mes...</td>\n",
       "      <td>News</td>\n",
       "      <td>December 25, 2017</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                               title  \\\n",
       "0   Donald Trump Sends Out Embarrassing New Year’...   \n",
       "1   Drunk Bragging Trump Staffer Started Russian ...   \n",
       "2   Sheriff David Clarke Becomes An Internet Joke...   \n",
       "3   Trump Is So Obsessed He Even Has Obama’s Name...   \n",
       "4   Pope Francis Just Called Out Donald Trump Dur...   \n",
       "\n",
       "                                                text subject  \\\n",
       "0  Donald Trump just couldn t wish all Americans ...    News   \n",
       "1  House Intelligence Committee Chairman Devin Nu...    News   \n",
       "2  On Friday, it was revealed that former Milwauk...    News   \n",
       "3  On Christmas day, Donald Trump announced that ...    News   \n",
       "4  Pope Francis used his annual Christmas Day mes...    News   \n",
       "\n",
       "                date  label  \n",
       "0  December 31, 2017      1  \n",
       "1  December 31, 2017      1  \n",
       "2  December 30, 2017      1  \n",
       "3  December 29, 2017      1  \n",
       "4  December 25, 2017      1  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Fake_news.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset1 = True_news[['text','label']]\n",
    "dataset2 = Fake_news[['text','label']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = pd.concat([dataset1 , dataset2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(44898, 2)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataset.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Null values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "text     0\n",
       "label    0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataset.isnull().sum() # no null values"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Balanced or Unbalanced dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1    23481\n",
       "0    21417\n",
       "Name: label, dtype: int64"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataset['label'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(21417, 2)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataset1.shape # true news"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(23481, 2)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataset2.shape # fake news"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Shuffle or Resample"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = dataset.sample(frac = 1)"
   ]
  },
  {
   "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>text</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>22161</th>\n",
       "      <td>21st Century Wire says It was an awkward fit o...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17723</th>\n",
       "      <td>MONROVIA (Reuters) - One of Liberia s leading ...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13759</th>\n",
       "      <td>MOSCOW (Reuters) - Former Russian economy mini...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9387</th>\n",
       "      <td>(Reuters) - Officials from 11 U.S. states sued...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18704</th>\n",
       "      <td>COLBERT   LANGUAGE WARNING! This late night ho...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2476</th>\n",
       "      <td>Kellyanne Conway, who serves as a senior advis...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16571</th>\n",
       "      <td>I ll bet you re thinking this is a joke, right...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6836</th>\n",
       "      <td>WASHINGTON (Reuters) - U.S. President-elect Do...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13742</th>\n",
       "      <td>Don t buy into the media lie that every LEGAL ...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>468</th>\n",
       "      <td>Donald Trump has the maturity of a toddler, an...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13729</th>\n",
       "      <td>TEGUCIGALPA (Reuters) - With 70 percent of bal...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7098</th>\n",
       "      <td>Who needs experts, really, when one can rely o...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3084</th>\n",
       "      <td>There s a man who s been permanently banned fr...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5895</th>\n",
       "      <td>BERLIN (Reuters) - German Chancellor Angela Me...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4123</th>\n",
       "      <td>WASHINGTON (Reuters) - U.S. Treasury Secretary...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6989</th>\n",
       "      <td>NEW YORK (Reuters) - President-elect Donald Tr...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10640</th>\n",
       "      <td>Will Austria s new mandates help to save their...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7968</th>\n",
       "      <td>NEW YORK (Reuters) - A majority of Americans ...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2934</th>\n",
       "      <td>WASHINGTON (Reuters) - Twenty moderate Republi...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5332</th>\n",
       "      <td>NEW YORK (Reuters) - NATO is the “strongest al...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                    text  label\n",
       "22161  21st Century Wire says It was an awkward fit o...      1\n",
       "17723  MONROVIA (Reuters) - One of Liberia s leading ...      0\n",
       "13759  MOSCOW (Reuters) - Former Russian economy mini...      0\n",
       "9387   (Reuters) - Officials from 11 U.S. states sued...      0\n",
       "18704  COLBERT   LANGUAGE WARNING! This late night ho...      1\n",
       "2476   Kellyanne Conway, who serves as a senior advis...      1\n",
       "16571  I ll bet you re thinking this is a joke, right...      1\n",
       "6836   WASHINGTON (Reuters) - U.S. President-elect Do...      0\n",
       "13742  Don t buy into the media lie that every LEGAL ...      1\n",
       "468    Donald Trump has the maturity of a toddler, an...      1\n",
       "13729  TEGUCIGALPA (Reuters) - With 70 percent of bal...      0\n",
       "7098   Who needs experts, really, when one can rely o...      1\n",
       "3084   There s a man who s been permanently banned fr...      1\n",
       "5895   BERLIN (Reuters) - German Chancellor Angela Me...      0\n",
       "4123   WASHINGTON (Reuters) - U.S. Treasury Secretary...      0\n",
       "6989   NEW YORK (Reuters) - President-elect Donald Tr...      0\n",
       "10640  Will Austria s new mandates help to save their...      1\n",
       "7968    NEW YORK (Reuters) - A majority of Americans ...      0\n",
       "2934   WASHINGTON (Reuters) - Twenty moderate Republi...      0\n",
       "5332   NEW YORK (Reuters) - NATO is the “strongest al...      0"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataset.head(20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "import nltk"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "import re\n",
    "from nltk.corpus import stopwords\n",
    "from nltk.stem import WordNetLemmatizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "ps = WordNetLemmatizer()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "stopwords = stopwords.words('english')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[nltk_data] Downloading package wordnet to /root/nltk_data...\n",
      "[nltk_data]   Package wordnet is already up-to-date!\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nltk.download('wordnet')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "def cleaning_data(row):\n",
    "    \n",
    "    # convert text to into lower case\n",
    "    row = row.lower() \n",
    "    \n",
    "    # this line of code only take words from text and remove number and special character using RegX\n",
    "    row = re.sub('[^a-zA-Z]' , ' ' , row)\n",
    "    \n",
    "    # split the data and make token.\n",
    "    token = row.split() \n",
    "    \n",
    "    # lemmatize the word and remove stop words like a, an , the , is ,are ...\n",
    "    news = [ps.lemmatize(word) for word in token if not word in stopwords]  \n",
    "    \n",
    "    # finaly join all the token with space\n",
    "    cleanned_news = ' '.join(news) \n",
    "    \n",
    "    # return cleanned data\n",
    "    return cleanned_news "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset['text'] = dataset['text'].apply(lambda x : cleaning_data(x))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "text     0\n",
       "label    0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataset.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import TfidfVectorizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "vectorizer = TfidfVectorizer(max_features = 50000 , lowercase=False , ngram_range=(1,2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(44898, 2)"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataset.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "X = dataset.iloc[:35000,0]\n",
    "y = dataset.iloc[:35000,1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "22161    st century wire say awkward fit nerve donald m...\n",
       "17723    monrovia reuters one liberia leading political...\n",
       "13759    moscow reuters former russian economy minister...\n",
       "9387     reuters official u state sued obama administra...\n",
       "18704    colbert language warning late night host one a...\n",
       "Name: text, dtype: object"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "22161    1\n",
       "17723    0\n",
       "13759    0\n",
       "9387     0\n",
       "18704    1\n",
       "Name: label, dtype: int64"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "train_data , test_data , train_label , test_label = train_test_split(X , y , test_size = 0.2 ,random_state = 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "vec_train_data = vectorizer.fit_transform(train_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "vec_train_data = vec_train_data.toarray()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((28000,), (7000,))"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data.shape , test_data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "vec_test_data = vectorizer.transform(test_data).toarray()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((28000, 50000), (7000, 50000))"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vec_train_data.shape , vec_test_data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1    14615\n",
       "0    13385\n",
       "Name: label, dtype: int64"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_label.value_counts() # balanced partition"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1    3674\n",
       "0    3326\n",
       "Name: label, dtype: int64"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_label.value_counts() # balanced partition"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "training_data = pd.DataFrame(vec_train_data , columns=vectorizer.get_feature_names())\n",
    "testing_data = pd.DataFrame(vec_test_data , columns= vectorizer.get_feature_names())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.naive_bayes import MultinomialNB"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import accuracy_score,classification_report"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "clf = MultinomialNB()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "clf.fit(training_data, train_label)\n",
    "y_pred  = clf.predict(testing_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### MultinomialNB"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1    3734\n",
       "0    3266\n",
       "dtype: int64"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.Series(y_pred).value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1    3674\n",
       "0    3326\n",
       "Name: label, dtype: int64"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_label.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.96      0.94      0.95      3326\n",
      "           1       0.95      0.96      0.96      3674\n",
      "\n",
      "    accuracy                           0.95      7000\n",
      "   macro avg       0.95      0.95      0.95      7000\n",
      "weighted avg       0.95      0.95      0.95      7000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(test_label , y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now predict on both train set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.96      0.95      0.96     13385\n",
      "           1       0.96      0.96      0.96     14615\n",
      "\n",
      "    accuracy                           0.96     28000\n",
      "   macro avg       0.96      0.96      0.96     28000\n",
      "weighted avg       0.96      0.96      0.96     28000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "y_pred_train = clf.predict(training_data)\n",
    "print(classification_report(train_label , y_pred_train))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9584642857142858"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_score(train_label , y_pred_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9531428571428572"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_score(test_label , y_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "news = cleaning_data(str(\"Imposters posing as army personnel on the social media have been called out by the Indian Army as false news and disinformation.\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1])"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "single_prediction = clf.predict(vectorizer.transform([news]).toarray())\n",
    "single_prediction"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Save the Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import joblib "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['model.pkl']"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "joblib.dump(clf , 'model.pkl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = joblib.load('model.pkl')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}