_EXPORT_POS.ipynb copy

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data & Lib"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn.metrics import f1_score, precision_score, recall_score\n",
    "from transformers import AdamW, get_linear_schedule_with_warmup\n",
    "\n",
    "import emoji, re\n",
    "from fetchData import fetchdata, cv_events\n",
    "import __MLP\n",
    "# from __MLP import getSamplers, convert_df_to_unsqueezed_tensor, train_sequential, clf_report\n",
    "import random\n",
    "\n",
    "import os\n",
    "import re\n",
    "from tqdm import tqdm\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "from sklearn.pipeline import make_pipeline\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.svm import LinearSVC\n",
    "\n",
    "\n",
    "from sklearn.model_selection import cross_val_score\n",
    "from sklearn.metrics import accuracy_score, recall_score, precision_score, f1_score\n",
    "from sklearn.metrics import classification_report\n",
    "\n",
    "from sklearn.naive_bayes import GaussianNB\n",
    "\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.ensemble import AdaBoostClassifier\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.model_selection import RandomizedSearchCV, GridSearchCV\n",
    "from sklearn.ensemble import GradientBoostingClassifier\n",
    "\n",
    "pd.set_option('display.max_columns', None)\n",
    "pd.set_option('display.max_rows', 150)\n",
    "pd.set_option('display.max_colwidth', 150)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {},
   "outputs": [],
   "source": [
    "# ext_sparse_processed = pd.read_csv('./data/tagged/_PHEMEext_textonly_token.csv')\n",
    "pheme_textonly= pd.read_csv('./data/tagged/_PHEME_textonly.csv',sep='\\t')\n",
    "pheme_text= pd.read_csv('./data/_PHEME_text.csv')\n",
    "pheme_extonly= pd.read_csv('./data/_PHEMEext_textonly.csv',sep='\\t')\n",
    "pheme_pos= pd.read_csv('./data/tagged/_PHEME_textonly_token.csv',sep='\\t')\n",
    "ext_pos= pd.read_csv('./data/tagged/_PHEMEext_textonly_token2.csv',sep='\\t')\n",
    "pheme_sparse= pd.read_csv('./data/_PHEME_sparse.csv')\n",
    "ext_sparse= pd.read_csv('./data/_PHEMEext_sparse.csv')\n",
    " \n",
    "pheme_pos_final= pd.read_csv('./data/tagged/_PHEME_pos_final.csv',sep='\\t')\n",
    "ext_pos_final= pd.read_csv('./data/tagged/_PHEMEext_pos_final.csv',sep='\\t')\n",
    "\n",
    "pheme_y= fetchdata('pheme','target').target\n",
    "ext_y= fetchdata('ext','target').target\n",
    "\n",
    "pheme_text = fetchdata('pheme','text')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 214,
   "metadata": {},
   "outputs": [],
   "source": [
    "phemeall =  pd.read_csv('./data/all/_PHEMEall_text.csv').text\n",
    "phemeall.to_csv('./data/all/_PHEMEall_text.csv',index=False)\n",
    "extall =  pd.read_csv('./data/all/_PHEMEextall_text.csv').text\n",
    "extall.to_csv('./data/all/_PHEMEextall_text.csv',index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "pheme_sparse_pos = pheme_sparse[['Noun', 'Verb', 'Adjective', 'Pronoun',  'Adverb', 'Numeral',\n",
    "                                 'Conjunction_inj', 'Particle', 'Determiner', 'Modal', 'Whs']]\n",
    "ext_sparse_pos = ext_sparse[['Noun', 'Verb', 'Adjective', 'Pronoun',  'Adverb', 'Numeral',\n",
    "                             'Conjunction_inj', 'Particle', 'Determiner', 'Modal', 'Whs']]\n",
    "pheme_sparse_pos_pronouns = pheme_sparse[['Noun', 'Verb', 'Adjective', 'Pronoun',  'Adverb', 'Numeral',\n",
    "                                          'Conjunction_inj', 'Particle', 'Determiner', 'Modal', 'Whs', 'FirstPersonPronoun',\n",
    "                                          'SecondPersonPronoun', 'ThirdPersonPronoun']]\n",
    "ext_sparse_pos_pronouns = ext_sparse[['Noun', 'Verb', 'Adjective', 'Pronoun',  'Adverb', 'Numeral',\n",
    "                                      'Conjunction_inj', 'Particle', 'Determiner', 'Modal', 'Whs', 'FirstPersonPronoun',\n",
    "                                      'SecondPersonPronoun', 'ThirdPersonPronoun']]\n",
    "pheme_sparse_pronouns = pheme_sparse[['FirstPersonPronoun',\n",
    "                                      'SecondPersonPronoun', 'ThirdPersonPronoun']]\n",
    "ext_sparse_pronouns = ext_sparse[['FirstPersonPronoun',\n",
    "                                  'SecondPersonPronoun', 'ThirdPersonPronoun']]\n",
    "pheme_sparse_nopos = pheme_sparse.drop(['Noun', 'Verb', 'Adjective', 'Pronoun',  'Adverb', 'Numeral',\n",
    "                             'Conjunction_inj', 'Particle', 'Determiner', 'Modal', 'Whs'], axis=1)\n",
    "ext_sparse_nopos = ext_sparse.drop(['Noun', 'Verb', 'Adjective', 'Pronoun',  'Adverb', 'Numeral',\n",
    "                             'Conjunction_inj', 'Particle', 'Determiner', 'Modal', 'Whs'],axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## PHEME ALL"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 236,
   "metadata": {},
   "outputs": [],
   "source": [
    "extall_POS =  pd.read_csv('/Users/june/Documents/Code/FYP/TweetNLP/temp_result.csv',sep='\\n')\n",
    "extall_POS2 =  pd.read_csv('/Users/june/Documents/Code/FYP/TweetNLP/temp_result.txt',sep='\\t')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 237,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1963,)\n",
      "(2140, 1)\n"
     ]
    },
    {
     "data": {
      "text/html": "<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>text\\tN\\t0.7899\\ttext</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>@Mourinholic 😕😕 http://t.co/sFoV1v8uDo\\t@ G U\\t0.9992 0.2120 0.9967\\t@Mourinholic 😕😕 http://t.co/sFoV1v8uDo</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>“ @Mourinholic : Micheal Essien denying the Ebola rumours like https://t.co/8Yo8iLgISS ”\\t, @ ~ ^ ^ V D ^ N P U ,\\t0.9421 0.9968 0.9910 0.9994 0.9...</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>@Mourinholic Hmmm .\\t@ ! ,\\t0.9990 0.9982 0.9969\\t@Mourinholic Hmmm.</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>@Mourinholic Even though it was against us , it was a bloody amazing goal . \\t, @ R P O V P O , O V D A A N , ,\\t0.9203 0.9983 0.9970 0.9575 0.99...</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>@CdtChoco1er thanks bro .\\t@ N N ,\\t0.9985 0.9199 0.9924 0.9988\\t@CdtChoco1er thanks bro.</td>\n    </tr>\n  </tbody>\n</table>\n</div>",
      "text/plain": "                                                                                                                                   text\\tN\\t0.7899\\ttext\n0                                            @Mourinholic 😕😕 http://t.co/sFoV1v8uDo\\t@ G U\\t0.9992 0.2120 0.9967\\t@Mourinholic 😕😕 http://t.co/sFoV1v8uDo\n1  “ @Mourinholic : Micheal Essien denying the Ebola rumours like https://t.co/8Yo8iLgISS ”\\t, @ ~ ^ ^ V D ^ N P U ,\\t0.9421 0.9968 0.9910 0.9994 0.9...\n2                                                                                   @Mourinholic Hmmm .\\t@ ! ,\\t0.9990 0.9982 0.9969\\t@Mourinholic Hmmm.\n3   @Mourinholic Even though it was against us , it was a bloody amazing goal . \\t, @ R P O V P O , O V D A A N , ,\\t0.9203 0.9983 0.9970 0.9575 0.99...\n4                                                              @CdtChoco1er thanks bro .\\t@ N N ,\\t0.9985 0.9199 0.9924 0.9988\\t@CdtChoco1er thanks bro."
     },
     "execution_count": 237,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(extall.shape)\n",
    "print(extall_POS.shape)\n",
    "extall_POS.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 235,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1963,)\n",
      "(2184, 4)\n"
     ]
    },
    {
     "data": {
      "text/html": "<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>text</th>\n      <th>N</th>\n      <th>0.7899</th>\n      <th>text.1</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>@Mourinholic 😕😕 http://t.co/sFoV1v8uDo</td>\n      <td>@ G U</td>\n      <td>0.9992 0.2120 0.9967</td>\n      <td>@Mourinholic 😕😕 http://t.co/sFoV1v8uDo</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>“ @Mourinholic : Micheal Essien denying the Ebola rumours like https://t.co/8Yo8iLgISS ”</td>\n      <td>, @ ~ ^ ^ V D ^ N P U ,</td>\n      <td>0.9421 0.9968 0.9910 0.9994 0.9937 0.9997 0.9993 0.8274 0.8219 0.9100 0.9868 0.9449</td>\n      <td>“@Mourinholic: Micheal Essien denying the Ebola rumours like https://t.co/8Yo8iLgISS”</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>@Mourinholic Hmmm .</td>\n      <td>@ ! ,</td>\n      <td>0.9990 0.9982 0.9969</td>\n      <td>@Mourinholic Hmmm.</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>@Mourinholic Even though it was against us , it was a bloody amazing goal .</td>\n      <td>, @ R P O V P O , O V D A A N , ,</td>\n      <td>0.9203 0.9983 0.9970 0.9575 0.9987 0.9995 0.9853 0.9344 0.9977 0.9971 0.9997 0.9988 0.8880 0.9957 0.9986 0.9989 0.9709</td>\n      <td>@Mourinholic Even though it was against us, it was a bloody amazing goal.</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>@CdtChoco1er thanks bro .</td>\n      <td>@ N N ,</td>\n      <td>0.9985 0.9199 0.9924 0.9988</td>\n      <td>@CdtChoco1er thanks bro.</td>\n    </tr>\n  </tbody>\n</table>\n</div>",
      "text/plain": "                                                                                       text  \\\n0                                                    @Mourinholic 😕😕 http://t.co/sFoV1v8uDo   \n1  “ @Mourinholic : Micheal Essien denying the Ebola rumours like https://t.co/8Yo8iLgISS ”   \n2                                                                       @Mourinholic Hmmm .   \n3              @Mourinholic Even though it was against us , it was a bloody amazing goal .    \n4                                                                 @CdtChoco1er thanks bro .   \n\n                                   N  \\\n0                              @ G U   \n1            , @ ~ ^ ^ V D ^ N P U ,   \n2                              @ ! ,   \n3  , @ R P O V P O , O V D A A N , ,   \n4                            @ N N ,   \n\n                                                                                                                   0.7899  \\\n0                                                                                                    0.9992 0.2120 0.9967   \n1                                     0.9421 0.9968 0.9910 0.9994 0.9937 0.9997 0.9993 0.8274 0.8219 0.9100 0.9868 0.9449   \n2                                                                                                    0.9990 0.9982 0.9969   \n3  0.9203 0.9983 0.9970 0.9575 0.9987 0.9995 0.9853 0.9344 0.9977 0.9971 0.9997 0.9988 0.8880 0.9957 0.9986 0.9989 0.9709   \n4                                                                                             0.9985 0.9199 0.9924 0.9988   \n\n                                                                                  text.1  \n0                                                 @Mourinholic 😕😕 http://t.co/sFoV1v8uDo  \n1  “@Mourinholic: Micheal Essien denying the Ebola rumours like https://t.co/8Yo8iLgISS”  \n2                                                                     @Mourinholic Hmmm.  \n3              @Mourinholic Even though it was against us, it was a bloody amazing goal.  \n4                                                               @CdtChoco1er thanks bro.  "
     },
     "execution_count": 235,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(extall.shape)\n",
    "print(extall_POS2.shape)\n",
    "extall_POS2.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data Check"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 145,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(5802, 1)\n",
      "(485, 1)\n"
     ]
    }
   ],
   "source": [
    "print(pheme_textonly.shape); print(pheme_extonly.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 146,
   "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>processed_text</th>\n      <th>POStags</th>\n      <th>proba</th>\n      <th>original_text</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>BREAKING : Armed man takes hostage in kosher grocery east of Paris http://t.co/PBs3sMwhLt</td>\n      <td>N , A N V N P A N N P ^ U</td>\n      <td>0.6086 0.9832 0.4822 0.9973 0.9819 0.9780 0.9852 0.9810 0.9235 0.5818 0.9989 0.9989 0.9981</td>\n      <td>BREAKING: Armed man takes hostage in kosher grocery east of Paris http://t.co/PBs3sMwhLt                                                          ...</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>#CharlieHebdo killers dead , confirmed by gendarmerie .</td>\n      <td># N A , V P ^ ,</td>\n      <td>0.7572 0.5413 0.7980 0.9988 0.9810 0.9893 0.7627 0.9939</td>\n      <td>#CharlieHebdo killers dead, confirmed by gendarmerie.                                                                                             ...</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>Top French cartoonists Charb , Cabu , Wolinski , Tignous confirmed among dead in #Paris #CharlieHebdo attack . Editor is critically wounded .</td>\n      <td>A ^ N V , ^ , ^ , ^ V P A P ^ ^ N , N V R A ,</td>\n      <td>0.9429 0.5236 0.9900 0.3360 0.9967 0.8644 0.9948 0.9043 0.9948 0.6443 0.9759 0.9830 0.8947 0.9914 0.6474 0.8114 0.5487 0.9977 0.9616 0.9952 0.9975...</td>\n      <td>Top French cartoonists Charb, Cabu, Wolinski, Tignous confirmed among dead in #Paris #CharlieHebdo attack. Editor is critically wounded.          ...</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>Police have surrounded the area where the #CharlieHebdo attack suspects are believed to be : http://t.co/3tGXEIX4F2 https://t.co/aBSezf2QWS</td>\n      <td>^ V V D N R D ^ V V V V P V , U U</td>\n      <td>0.5557 0.9941 0.9846 0.9993 0.8847 0.9220 0.9868 0.9265 0.8443 0.8949 0.9825 0.9309 0.9817 0.9960 0.9067 0.9947 0.9977</td>\n      <td>Police have surrounded the area where the #CharlieHebdo attack suspects are believed to be: http://t.co/3tGXEIX4F2 https://t.co/aBSezf2QWS        ...</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>PHOTO : Armed gunmen face police officers near #CharlieHebdo HQ in Paris http://t.co/3Jsosc7yl3 http://t.co/iOpVNO6Iq0</td>\n      <td>N , A N N N N P ^ ^ P ^ U U</td>\n      <td>0.9901 0.9859 0.6329 0.9879 0.9015 0.6475 0.7908 0.9858 0.8528 0.9006 0.9963 0.9977 0.9955 0.9994</td>\n      <td>PHOTO: Armed gunmen face police officers near #CharlieHebdo HQ in Paris http://t.co/3Jsosc7yl3 http://t.co/iOpVNO6Iq0                             ...</td>\n    </tr>\n  </tbody>\n</table>\n</div>",
      "text/plain": "                                                                                                                                  processed_text  \\\n0                                                      BREAKING : Armed man takes hostage in kosher grocery east of Paris http://t.co/PBs3sMwhLt   \n1                                                                                        #CharlieHebdo killers dead , confirmed by gendarmerie .   \n2  Top French cartoonists Charb , Cabu , Wolinski , Tignous confirmed among dead in #Paris #CharlieHebdo attack . Editor is critically wounded .   \n3    Police have surrounded the area where the #CharlieHebdo attack suspects are believed to be : http://t.co/3tGXEIX4F2 https://t.co/aBSezf2QWS   \n4                         PHOTO : Armed gunmen face police officers near #CharlieHebdo HQ in Paris http://t.co/3Jsosc7yl3 http://t.co/iOpVNO6Iq0   \n\n                                         POStags  \\\n0                      N , A N V N P A N N P ^ U   \n1                                # N A , V P ^ ,   \n2  A ^ N V , ^ , ^ , ^ V P A P ^ ^ N , N V R A ,   \n3              ^ V V D N R D ^ V V V V P V , U U   \n4                    N , A N N N N P ^ ^ P ^ U U   \n\n                                                                                                                                                   proba  \\\n0                                                             0.6086 0.9832 0.4822 0.9973 0.9819 0.9780 0.9852 0.9810 0.9235 0.5818 0.9989 0.9989 0.9981   \n1                                                                                                0.7572 0.5413 0.7980 0.9988 0.9810 0.9893 0.7627 0.9939   \n2  0.9429 0.5236 0.9900 0.3360 0.9967 0.8644 0.9948 0.9043 0.9948 0.6443 0.9759 0.9830 0.8947 0.9914 0.6474 0.8114 0.5487 0.9977 0.9616 0.9952 0.9975...   \n3                                 0.5557 0.9941 0.9846 0.9993 0.8847 0.9220 0.9868 0.9265 0.8443 0.8949 0.9825 0.9309 0.9817 0.9960 0.9067 0.9947 0.9977   \n4                                                      0.9901 0.9859 0.6329 0.9879 0.9015 0.6475 0.7908 0.9858 0.8528 0.9006 0.9963 0.9977 0.9955 0.9994   \n\n                                                                                                                                           original_text  \n0  BREAKING: Armed man takes hostage in kosher grocery east of Paris http://t.co/PBs3sMwhLt                                                          ...  \n1  #CharlieHebdo killers dead, confirmed by gendarmerie.                                                                                             ...  \n2  Top French cartoonists Charb, Cabu, Wolinski, Tignous confirmed among dead in #Paris #CharlieHebdo attack. Editor is critically wounded.          ...  \n3  Police have surrounded the area where the #CharlieHebdo attack suspects are believed to be: http://t.co/3tGXEIX4F2 https://t.co/aBSezf2QWS        ...  \n4  PHOTO: Armed gunmen face police officers near #CharlieHebdo HQ in Paris http://t.co/3Jsosc7yl3 http://t.co/iOpVNO6Iq0                             ...  "
     },
     "execution_count": 146,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pheme_pos_final.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 147,
   "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>processed_text</th>\n      <th>POStags</th>\n      <th>proba</th>\n      <th>original_text</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>Micheal Essien denying the Ebola rumours like https://t.co/H2E1TAzeha</td>\n      <td>^ ^ V D ^ N P U</td>\n      <td>0.9987 0.9934 0.9997 0.9994 0.8274 0.8091 0.9177 0.9935</td>\n      <td>Micheal Essien denying the Ebola rumours like https://t.co/H2E1TAzeha</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>No truth in internet rumours that I have contracted Ebola . i m very well &amp; I'm doing very gud &amp; will be training as usual tomorrow . #falsenews</td>\n      <td>D N P N N P O V V ^ , O V R R &amp; L V R A &amp; V V N P A N , #</td>\n      <td>0.8772 0.9968 0.9936 0.9153 0.9477 0.9752 0.9995 0.9998 0.9695 0.7780 0.9965 0.9333 0.9256 0.9917 0.5153 0.9895 0.9959 0.9939 0.9895 0.9981 0.9942...</td>\n      <td>No truth in internet rumours that I have contracted Ebola.i m very well &amp;amp; I'm doing very gud &amp;amp; will be training as usual tomorrow.#falsenews</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>Essien and his lawyers are considering to file a lawsuit against the Nigerian media that reported the fake Ebola story .</td>\n      <td>^ &amp; D N V V P V D N P D ^ N P V D A ^ N ,</td>\n      <td>0.9064 0.9984 0.9978 0.9972 0.9944 0.9660 0.9899 0.9900 0.9966 0.9953 0.9998 0.9985 0.8974 0.8566 0.8431 0.9956 0.9996 0.9763 0.6529 0.9933 0.9978</td>\n      <td>Essien and his lawyers are considering to file a lawsuit against the Nigerian media that reported the fake Ebola story.</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>Good news : The rumours that Michael Essien has contracted the Ebola virus are false . http://t.co/5d7hCL46mR http://t.co/VtGuLnjWBD</td>\n      <td>A N , D N P ^ ^ V V D ^ N V A , U U</td>\n      <td>0.9648 0.9985 0.7902 0.9930 0.9866 0.8974 0.9996 0.9949 0.9966 0.9981 0.9993 0.9321 0.9860 0.9878 0.9896 0.9979 0.9954 0.9986</td>\n      <td>Good news: The rumours that Michael Essien has contracted the Ebola virus are false. http://t.co/5d7hCL46mR http://t.co/VtGuLnjWBD</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>Milan have stated that the reports about Essien having Ebola are completely false . http://t.co/Sb9v9ulfTX @MichaelEssien</td>\n      <td>^ V V P D N P ^ V ^ V R A , U @</td>\n      <td>0.9883 0.9935 0.9839 0.8956 0.9968 0.9879 0.9860 0.9582 0.9981 0.7131 0.9523 0.9994 0.9815 0.9984 0.9959 0.9964</td>\n      <td>Milan have stated that the reports about Essien having Ebola are completely false. http://t.co/Sb9v9ulfTX @MichaelEssien</td>\n    </tr>\n  </tbody>\n</table>\n</div>",
      "text/plain": "                                                                                                                                     processed_text  \\\n0                                                                             Micheal Essien denying the Ebola rumours like https://t.co/H2E1TAzeha   \n1  No truth in internet rumours that I have contracted Ebola . i m very well & I'm doing very gud & will be training as usual tomorrow . #falsenews   \n2                          Essien and his lawyers are considering to file a lawsuit against the Nigerian media that reported the fake Ebola story .   \n3              Good news : The rumours that Michael Essien has contracted the Ebola virus are false . http://t.co/5d7hCL46mR http://t.co/VtGuLnjWBD   \n4                         Milan have stated that the reports about Essien having Ebola are completely false . http://t.co/Sb9v9ulfTX @MichaelEssien   \n\n                                                     POStags  \\\n0                                            ^ ^ V D ^ N P U   \n1  D N P N N P O V V ^ , O V R R & L V R A & V V N P A N , #   \n2                  ^ & D N V V P V D N P D ^ N P V D A ^ N ,   \n3                        A N , D N P ^ ^ V V D ^ N V A , U U   \n4                            ^ V V P D N P ^ V ^ V R A , U @   \n\n                                                                                                                                                   proba  \\\n0                                                                                                0.9987 0.9934 0.9997 0.9994 0.8274 0.8091 0.9177 0.9935   \n1  0.8772 0.9968 0.9936 0.9153 0.9477 0.9752 0.9995 0.9998 0.9695 0.7780 0.9965 0.9333 0.9256 0.9917 0.5153 0.9895 0.9959 0.9939 0.9895 0.9981 0.9942...   \n2     0.9064 0.9984 0.9978 0.9972 0.9944 0.9660 0.9899 0.9900 0.9966 0.9953 0.9998 0.9985 0.8974 0.8566 0.8431 0.9956 0.9996 0.9763 0.6529 0.9933 0.9978   \n3                          0.9648 0.9985 0.7902 0.9930 0.9866 0.8974 0.9996 0.9949 0.9966 0.9981 0.9993 0.9321 0.9860 0.9878 0.9896 0.9979 0.9954 0.9986   \n4                                        0.9883 0.9935 0.9839 0.8956 0.9968 0.9879 0.9860 0.9582 0.9981 0.7131 0.9523 0.9994 0.9815 0.9984 0.9959 0.9964   \n\n                                                                                                                                          original_text  \n0  Micheal Essien denying the Ebola rumours like https://t.co/H2E1TAzeha                                                                                 \n1  No truth in internet rumours that I have contracted Ebola.i m very well &amp; I'm doing very gud &amp; will be training as usual tomorrow.#falsenews  \n2  Essien and his lawyers are considering to file a lawsuit against the Nigerian media that reported the fake Ebola story.                               \n3  Good news: The rumours that Michael Essien has contracted the Ebola virus are false. http://t.co/5d7hCL46mR http://t.co/VtGuLnjWBD                    \n4  Milan have stated that the reports about Essien having Ebola are completely false. http://t.co/Sb9v9ulfTX @MichaelEssien                              "
     },
     "execution_count": 147,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ext_pos_final.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [],
   "source": [
    "def text_preprocessing_simple(text, lemma=False, twttknzr=True): # Create a function to tokenize a set of texts\n",
    "\n",
    "    \"\"\"\n",
    "    - Remove entity mentions (eg. '@united')\n",
    "    - Correct errors (eg. '&amp;' to '&')\n",
    "    @param    text (str): a string to be processed.\n",
    "    @return   text (Str): the processed string.\n",
    "    \"\"\"\n",
    "\n",
    "    # text = re.sub(r\"http\\S+\", \"*\", text)  # http link -> '*'\n",
    "    # sent = re.sub(r'([^\\s\\w@#\\*]|_)+', '', sent) # Erasing Special Characters\n",
    "\n",
    "    text = emoji.demojize(text)\n",
    "    text = re.sub(r':[^:\\s]*:', r' \\g<0>', text)  # http link -> '*'\n",
    "\n",
    "    # text = re.sub(r':[^:\\s]*(?:::[^:\\s]*)*:', r' \\g<0> ', text)  # http link -> '*'\n",
    "\n",
    "    # text = re.sub(r\"\\n\", \" \", text)   # mention -> '@'\n",
    "    text = re.sub(r\"http\\S+\", \"HTTPURL\", text)  # http link -> '*'\n",
    "    # text = re.sub(r\"@\\S+\", \"@USER\", text)   # mention -> '@'\n",
    "    \n",
    "    text = re.sub(r\"@[A-Za-z0-9]+\", \"@USER\", text)   # mention -> '@'\n",
    "\n",
    "    # Remove trailing whitespace\n",
    "    text = re.sub(r'\\s+', ' ', text).strip()\n",
    "    text = re.sub(r'&amp;', '&', text)\n",
    "    # text = tweetTokenizer.tokenize(text)\n",
    "    # text = [emoji.demojize(token) for token in text]\n",
    "\n",
    "    return text\n",
    "\n",
    "def text_preprocessing_pos(text, lemma=False, twttknzr=False): # Create a function to tokenize a set of texts\n",
    "\n",
    "    \"\"\"\n",
    "    - Remove entity mentions (eg. '@united')\n",
    "    - Correct errors (eg. '&amp;' to '&')\n",
    "    @param    text (str): a string to be processed.\n",
    "    @return   text (Str): the processed string.\n",
    "    \"\"\"\n",
    "\n",
    "    # text = re.sub(r\"http\\S+\", \"*\", text)  # http link -> '*'\n",
    "    # sent = re.sub(r'([^\\s\\w@#\\*]|_)+', '', sent) # Erasing Special Characters\n",
    "\n",
    "    text = emoji.demojize(text)\n",
    "    text = re.sub(r':[^:\\s]*:', r' \\g<0>', text)  # http link -> '*'\n",
    "\n",
    "    # text = re.sub(r':[^:\\s]*(?:::[^:\\s]*)*:', r' \\g<0> ', text)  # http link -> '*'\n",
    "\n",
    "    # text = re.sub(r\"\\n\", \" \", text)   # mention -> '@'\n",
    "    text = re.sub(r\"http\\S+\", \"HTTPURL\", text)  # http link -> '*'\n",
    "    # text = re.sub(r\"@\\S+\", \"@USER\", text)   # mention -> '@'\n",
    "    \n",
    "    text = re.sub(r\"@[A-Za-z0-9]+\", \"@USER\", text)   # mention -> '@'\n",
    "\n",
    "    # Remove trailing whitespace\n",
    "    text = re.sub(r'\\s+', ' ', text).strip()\n",
    "    text = re.sub(r'&amp;', '&', text)\n",
    "    # text = tweetTokenizer.tokenize(text)\n",
    "    # text = [emoji.demojize(token) for token in text]\n",
    "\n",
    "    return text"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "result = [text_preprocessing_pos(sent) for sent in pheme_textonly.text]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 144,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['I@@', \"'m\", 'very', 'fit', 'and', 'very', 'health@@', 'y@@', ',@@', 'No', 'truth', 'in', 'the', 'internet', 'rumours', 'that', 'I', 'have', 'contracted', 'E@@', 'bol@@', 'a.@@', 'im', 'well', '&@@', '…', 'HTTPURL']\n"
     ]
    }
   ],
   "source": [
    "print(tokenizer.tokenize(text_preprocessing_simple(pheme_extonly.iloc[5][0])))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 142,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "I'm very fit and very healthy,No truth in the internet rumours that I have contracted Ebola.im well &amp;… http://t.co/TGidyI5JVG\n",
      "I'm very fit and very healthy , No truth in the internet rumours that I have contracted Ebola.im well & … http://t.co/TGidyI5JVG\n",
      "L R V & R A , D N P D N N P O V V U R & , U\n",
      "I'm very fit and very healthy,No truth in the internet rumours that I have contracted Ebola.im well &… HTTPURL\n"
     ]
    }
   ],
   "source": [
    "print(pheme_extonly.iloc[5][0])\n",
    "print(ext_pos_final.iloc[5][0])\n",
    "print(ext_pos_final.iloc[5][1])\n",
    "print(text_preprocessing_simple(pheme_extonly.iloc[5][0]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Processing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "from sklearn.feature_extraction.text import TfidfVectorizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "pheme_pos_final= pd.read_csv('./data/tagged/_PHEME_pos_final.csv',sep='\\t')\n",
    "\n",
    "# pheme_pos_final.iloc[1304].processed_text=\"' We are all Charlie ” : European newspapers show solidarity with Charlie Hebdo after attack http://t.co/xq1OQhDJVF http://t.co/jt7D6fuIBH\"\n",
    "# pheme_pos_final.iloc[1304].POStags=\", O V D ^ , , A N V N P ^ ^ P N U U\"\n",
    "# pheme_pos_final.iloc[1304].proba=\"0.9905 0.9951 0.9998 0.8048 0.9974 0.9494 0.9120 0.9475 0.9980 0.5189 0.9962 0.9990 0.9996 0.9758 0.8693 0.7830 0.9949 0.9991\"\n",
    "# pheme_pos_final.iloc[1304].original_text=\"We are all Charlie”: European newspapers show solidarity with Charlie Hebdo after attack http://t.co/xq1OQhDJVF http://t.co/jt7D6fuIBH\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Empty DataFrame\n",
      "Columns: [processed_text, POStags, proba, original_text]\n",
      "Index: []\n"
     ]
    }
   ],
   "source": [
    "print(pheme_pos_final.loc[pheme_pos_final.POStags.isna() == True])\n",
    "# print(pheme_pos_final.iloc[1304])\n",
    "# print(pheme_pos_final.iloc[1305].processed_text)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 191,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'n': 13, ' ': 0, ',': 5, 'a': 8, 'v': 20, 'p': 15, '^': 7, 'u': 19, '#': 2, 'r': 16, 'd': 9, '$': 3, 's': 17, 'o': 14, '&': 4, 'l': 12, '@': 6, 't': 18, '~': 23, '!': 1, 'g': 11, 'z': 22, 'x': 21, 'e': 10}\n"
     ]
    }
   ],
   "source": [
    "pheme_POS = [sent for sent in pheme_pos_final['POStags'].values]\n",
    "ext_POS = [sent for sent in ext_pos_final['POStags'].values]\n",
    "\n",
    "# vector = CountVectorizer(analyzer='char_wb')\n",
    "vector = CountVectorizer(analyzer='char')\n",
    "train = vector.fit_transform(pheme_POS)\n",
    "test = vector.transform(ext_POS)\n",
    "print(vector.vocabulary_)  # 각 단어의 인덱스가 어떻게 부여되었는지를 보여준다.\n",
    "# print(vector.transform(pheme_POS).toarray())  # 코퍼스로부터 각 단어의 빈도 수를 기록한다. \n",
    "# pd.DataFrame(vector.fit_transform(POS).toarray())\n",
    "countvect_train = pd.DataFrame(train.todense(), columns=vector.get_feature_names())\n",
    "countvect_test = pd.DataFrame(test.todense(), columns=vector.get_feature_names())\n",
    "countvect_train = countvect_train.iloc[:,1:].drop(['e','$'],axis=1)\n",
    "countvect_test = countvect_test.iloc[:,1:].drop(['e','$'],axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Export"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 192,
   "metadata": {},
   "outputs": [],
   "source": [
    "countvect_train.to_csv(\"./data/_PHEME_postags\", index = False)\n",
    "countvect_test.to_csv(\"./data/_PHEMEext_postags\",index = False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 데이터 비교"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 194,
   "metadata": {},
   "outputs": [],
   "source": [
    "train = pd.concat([countvect_train,pheme_sparse_pos,pheme_text, pheme_sparse_nopos],axis=1)\n",
    "test = pd.concat([countvect_test,ext_sparse_pos,ext_sparse_nopos],axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 201,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": "Index(['!', '#', '&', ',', '@', '^', 'a', 'd', 'g', 'l', 'n', 'o', 'p', 'r',\n       's', 't', 'u', 'v', 'x', 'z', '~', 'Noun', 'Verb', 'Adjective',\n       'Pronoun', 'Adverb', 'Numeral', 'Conjunction_inj', 'Particle',\n       'Determiner', 'Modal', 'Whs', 'text', 'Event', 'target', 'emoji_count',\n       'URLcount', 'has_media', 'Skepticism', 'MentionCount',\n       'FirstPersonPronoun', 'SecondPersonPronoun', 'ThirdPersonPronoun',\n       'char_count', 'word_count', 'HashTag', 'has_question', 'has_exclaim',\n       'has_period', 'capital_ratio', 'retweet_count', 'isRT', 'tweet_count',\n       'listed_count', 'friends_count', 'follower_count', 'followers/friend',\n       'favourites_count', 'account_age_days', 'verified'],\n      dtype='object')"
     },
     "execution_count": 201,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.columns\n",
    "\n",
    "# ndex(['!', '#', '$', '&', ',', '@', '^', 'a', 'd', 'e', 'g', 'l', 'n', 'o',\n",
    "#        'p', 'r', 's', 't', 'u', 'v', 'x', 'z', '~', 'Noun', 'Verb',\n",
    "#        'Adjective', 'Pronoun', 'Adverb', 'Numeral', 'Conjunction_inj',\n",
    "#        'Particle', 'Determiner', 'Modal', 'Whs', 'text', 'Event', 'target',\n",
    "#        'emoji_count', 'URLcount', 'has_media', 'Skepticism', 'MentionCount',\n",
    "#        'FirstPersonPronoun', 'SecondPersonPronoun', 'ThirdPersonPronoun',\n",
    "#        'char_count', 'word_count', 'HashTag', 'has_question', 'has_exclaim',\n",
    "#        'has_period', 'capital_ratio', 'retweet_count', 'isRT', 'tweet_count',\n",
    "#        'listed_count', 'friends_count', 'follower_count', 'followers/friend',\n",
    "#        'favourites_count', 'account_age_days', 'verified'],\n",
    "#       dtype='object')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 238,
   "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>^</th>\n      <th>v</th>\n      <th>Verb</th>\n      <th>n</th>\n      <th>Noun</th>\n      <th>o</th>\n      <th>Pronoun</th>\n      <th>a</th>\n      <th>Adjective</th>\n      <th>Adverb</th>\n      <th>r</th>\n      <th>#</th>\n      <th>HashTag</th>\n      <th>u</th>\n      <th>URLcount</th>\n      <th>@</th>\n      <th>MentionCount</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>BREAKING: Armed man takes hostage in kosher grocery east of Paris http://t.co/PBs3sMwhLt</td>\n      <td>1</td>\n      <td>1</td>\n      <td>3</td>\n      <td>5</td>\n      <td>6</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>#CharlieHebdo killers dead, confirmed by gendarmerie.</td>\n      <td>1</td>\n      <td>1</td>\n      <td>2</td>\n      <td>1</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>Top French cartoonists Charb, Cabu, Wolinski, Tignous confirmed among dead in #Paris #CharlieHebdo attack. Editor is critically wounded.</td>\n      <td>6</td>\n      <td>3</td>\n      <td>4</td>\n      <td>3</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>6</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>Police have surrounded the area where the #CharlieHebdo attack suspects are believed to be: http://t.co/3tGXEIX4F2\\nhttps://t.co/aBSezf2QWS</td>\n      <td>2</td>\n      <td>7</td>\n      <td>5</td>\n      <td>1</td>\n      <td>4</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>1</td>\n      <td>2</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>PHOTO: Armed gunmen face police officers near #CharlieHebdo HQ in Paris http://t.co/3Jsosc7yl3 http://t.co/iOpVNO6Iq0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>2</td>\n      <td>5</td>\n      <td>6</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>2</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>...</th>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n    </tr>\n    <tr>\n      <th>5797</th>\n      <td>'I'll ride with you' http://t.co/llZnuCAzg5 Australia unites during #SydneySiege http://t.co/WIU22VPgkz</td>\n      <td>2</td>\n      <td>2</td>\n      <td>2</td>\n      <td>0</td>\n      <td>2</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>2</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>5798</th>\n      <td>Canada's thoughts and prayers are with our Australian friends. #MartinPlace #SydneySiege</td>\n      <td>0</td>\n      <td>1</td>\n      <td>2</td>\n      <td>3</td>\n      <td>4</td>\n      <td>0</td>\n      <td>1</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>5799</th>\n      <td>Every non-muslim in the world must watch this video https://t.co/sZdyhISoVh &amp;amp; show it every other non-muslim! #sydneysiege</td>\n      <td>0</td>\n      <td>3</td>\n      <td>3</td>\n      <td>2</td>\n      <td>4</td>\n      <td>1</td>\n      <td>1</td>\n      <td>3</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>1</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>5800</th>\n      <td>Suspect in Sydney cafe siege identified as Man Haron Monis, an Iranian granted asylum in Australia http://t.co/6Lrl9DEMXA</td>\n      <td>3</td>\n      <td>1</td>\n      <td>3</td>\n      <td>6</td>\n      <td>7</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>5801</th>\n      <td>Australians respond to racism by telling #Muslim community #illridewithyou. #sydneysiege #MartinPlace</td>\n      <td>2</td>\n      <td>2</td>\n      <td>3</td>\n      <td>3</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2</td>\n      <td>4</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n  </tbody>\n</table>\n<p>5802 rows × 18 columns</p>\n</div>",
      "text/plain": "                                                                                                                                             text  \\\n0                                                        BREAKING: Armed man takes hostage in kosher grocery east of Paris http://t.co/PBs3sMwhLt   \n1                                                                                           #CharlieHebdo killers dead, confirmed by gendarmerie.   \n2        Top French cartoonists Charb, Cabu, Wolinski, Tignous confirmed among dead in #Paris #CharlieHebdo attack. Editor is critically wounded.   \n3     Police have surrounded the area where the #CharlieHebdo attack suspects are believed to be: http://t.co/3tGXEIX4F2\\nhttps://t.co/aBSezf2QWS   \n4                           PHOTO: Armed gunmen face police officers near #CharlieHebdo HQ in Paris http://t.co/3Jsosc7yl3 http://t.co/iOpVNO6Iq0   \n...                                                                                                                                           ...   \n5797                                      'I'll ride with you' http://t.co/llZnuCAzg5 Australia unites during #SydneySiege http://t.co/WIU22VPgkz   \n5798                                                     Canada's thoughts and prayers are with our Australian friends. #MartinPlace #SydneySiege   \n5799               Every non-muslim in the world must watch this video https://t.co/sZdyhISoVh &amp; show it every other non-muslim! #sydneysiege   \n5800                    Suspect in Sydney cafe siege identified as Man Haron Monis, an Iranian granted asylum in Australia http://t.co/6Lrl9DEMXA   \n5801                                        Australians respond to racism by telling #Muslim community #illridewithyou. #sydneysiege #MartinPlace   \n\n      ^  v  Verb  n  Noun  o  Pronoun  a  Adjective  Adverb  r  #  HashTag  u  \\\n0     1  1     3  5     6  0        0  2          0       0  0  0        0  1   \n1     1  1     2  1     2  0        0  1          0       0  0  1        1  0   \n2     6  3     4  3     3  0        0  3          6       1  1  0        2  0   \n3     2  7     5  1     4  0        0  0          0       0  1  0        1  2   \n4     3  0     2  5     6  0        0  1          0       0  0  0        1  2   \n...  .. ..   ... ..   ... ..      ... ..        ...     ... .. ..      ... ..   \n5797  2  2     2  0     2  1        1  0          0       0  0  0        1  2   \n5798  0  1     2  3     4  0        1  1          1       0  0  2        2  0   \n5799  0  3     3  2     4  1        1  3          1       0  0  1        1  1   \n5800  3  1     3  6     7  0        0  2          2       0  0  0        0  1   \n5801  2  2     3  3     2  0        0  0          0       0  0  2        4  0   \n\n      URLcount  @  MentionCount  \n0            1  0             0  \n1            0  0             0  \n2            0  0             0  \n3            2  0             0  \n4            2  0             0  \n...        ... ..           ...  \n5797         2  0             0  \n5798         0  0             0  \n5799         1  0             0  \n5800         1  0             0  \n5801         0  0             0  \n\n[5802 rows x 18 columns]"
     },
     "execution_count": 238,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.describe()\n",
    "train[['text','^','v','Verb','n','Noun', 'o','Pronoun','a','Adjective','Adverb','r', '#', 'HashTag','u','URLcount','@','MentionCount']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 239,
   "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>@</th>\n      <th>MentionCount</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>462</th>\n      <td>Brilliant #CharlieHebdo satirists did not attack #Islam, but poked fun @ #intolerance &amp;amp; #extremism of all kind inc #racism &amp;amp; #islamophobia.</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>471</th>\n      <td>The shooting @ \"Charlie Hebdo\", in France, is a horrible tragedy. Ignorance, hatred and cowardice has no religion, no political allegiance.</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>1044</th>\n      <td>@FT's Tony Barber describes #CharlieHebdo as \"editorially foolish\" and \"stupid\"  http://t.co/dFan6rptkM http://t.co/cmiaUNA0eO</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>2104</th>\n      <td>Police have named the cop who shot Michael Brown. @alicesperi's latest report from #Ferguson: http://t.co/1K79nsOtx3 http://t.co/cV3VCkUR6L</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>2920</th>\n      <td>This is sheer insanity. Just look at @sebwalker's photo from #Ferguson. http://t.co/6f8dNxWBeO</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>3840</th>\n      <td>#RCMP news conference on #Ottawa shootings expected to begin momentarily. Watch live @ http://t.co/kngapKTSCe</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>4035</th>\n      <td>#RCMP to hold news conference on #Ottawa shootings at 2 pm ET, 11 am PT. Watch live coverage @ http://t.co/kngapKTSCe</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>4199</th>\n      <td>.@IvisonJ's first-hand account of the scene in Ottawa that will 'change this country forever' http://t.co/g5lhxHqNNs http://t.co/AP0stmObnn</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>4363</th>\n      <td>#RCMP asking for assistance if you have any photos or videos from #ottawa shooting send to: NatDiv_Media_DivNat@rcmp-grc.gc.ca</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>4374</th>\n      <td>#RCMP asking for assistance if you have any photos or videos from #ottawa shooting please send to: NatDiv_Media_DivNat@rcmp-grc.gc.ca</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>4397</th>\n      <td>#RCMP asking for assistance if you have any photos or videos from #ottawa shooting please send to: NatDiv_Media_DivNat@rcmp-grc.gc.ca</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>4442</th>\n      <td>The powerful War Memorial mouth-to-mouth image being shared is by @OttawaCitizen's Wayne Cuddington. #ottnews #ottawa http://t.co/FcSGb097G9</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>4462</th>\n      <td>If you have pictures or video of the suspect, email them NatDiv_Media_DivNat@rcmp-grc.gc.ca #ottawa #ottcity #ottnews</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>4478</th>\n      <td>Please forward any tips, photos or videos to NatDiv_Media_DivNat@rcmp-grc.gc.ca #OttawaShooting c: @RCMP_Nat_Div ^JT</td>\n      <td>1</td>\n      <td>2</td>\n    </tr>\n    <tr>\n      <th>4567</th>\n      <td>Ottawa youth: If you need to talk: 24/7 Crisis Line\\n613-260-2360 or 1-877-377-7775 (toll free for Eastern Ontario)\\ncrisis@ysb.on.ca</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>4735</th>\n      <td>#sydneysiege is a misnomer. It's a hostage situation. Inside a cafe. You want 2 know what being under siege looks like? look @ #Gaza.</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>5093</th>\n      <td>Hostage situation in Sydney is happening next door to @7NewsSydney's studio -- this is a live stream of the coverage https://t.co/uNXkJmspr8</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>5470</th>\n      <td>.@amcoren's report on what we know right now about the #sydneysiege: http://t.co/lkXAAgN5Rg</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n  </tbody>\n</table>\n</div>",
      "text/plain": "                                                                                                                                                     text  \\\n462   Brilliant #CharlieHebdo satirists did not attack #Islam, but poked fun @ #intolerance &amp; #extremism of all kind inc #racism &amp; #islamophobia.   \n471           The shooting @ \"Charlie Hebdo\", in France, is a horrible tragedy. Ignorance, hatred and cowardice has no religion, no political allegiance.   \n1044                       @FT's Tony Barber describes #CharlieHebdo as \"editorially foolish\" and \"stupid\"  http://t.co/dFan6rptkM http://t.co/cmiaUNA0eO   \n2104          Police have named the cop who shot Michael Brown. @alicesperi's latest report from #Ferguson: http://t.co/1K79nsOtx3 http://t.co/cV3VCkUR6L   \n2920                                                       This is sheer insanity. Just look at @sebwalker's photo from #Ferguson. http://t.co/6f8dNxWBeO   \n3840                                        #RCMP news conference on #Ottawa shootings expected to begin momentarily. Watch live @ http://t.co/kngapKTSCe   \n4035                                #RCMP to hold news conference on #Ottawa shootings at 2 pm ET, 11 am PT. Watch live coverage @ http://t.co/kngapKTSCe   \n4199          .@IvisonJ's first-hand account of the scene in Ottawa that will 'change this country forever' http://t.co/g5lhxHqNNs http://t.co/AP0stmObnn   \n4363                       #RCMP asking for assistance if you have any photos or videos from #ottawa shooting send to: NatDiv_Media_DivNat@rcmp-grc.gc.ca   \n4374                #RCMP asking for assistance if you have any photos or videos from #ottawa shooting please send to: NatDiv_Media_DivNat@rcmp-grc.gc.ca   \n4397                #RCMP asking for assistance if you have any photos or videos from #ottawa shooting please send to: NatDiv_Media_DivNat@rcmp-grc.gc.ca   \n4442         The powerful War Memorial mouth-to-mouth image being shared is by @OttawaCitizen's Wayne Cuddington. #ottnews #ottawa http://t.co/FcSGb097G9   \n4462                                If you have pictures or video of the suspect, email them NatDiv_Media_DivNat@rcmp-grc.gc.ca #ottawa #ottcity #ottnews   \n4478                                 Please forward any tips, photos or videos to NatDiv_Media_DivNat@rcmp-grc.gc.ca #OttawaShooting c: @RCMP_Nat_Div ^JT   \n4567                Ottawa youth: If you need to talk: 24/7 Crisis Line\\n613-260-2360 or 1-877-377-7775 (toll free for Eastern Ontario)\\ncrisis@ysb.on.ca   \n4735                #sydneysiege is a misnomer. It's a hostage situation. Inside a cafe. You want 2 know what being under siege looks like? look @ #Gaza.   \n5093         Hostage situation in Sydney is happening next door to @7NewsSydney's studio -- this is a live stream of the coverage https://t.co/uNXkJmspr8   \n5470                                                          .@amcoren's report on what we know right now about the #sydneysiege: http://t.co/lkXAAgN5Rg   \n\n      @  MentionCount  \n462   0             1  \n471   0             1  \n1044  0             1  \n2104  0             1  \n2920  0             1  \n3840  0             1  \n4035  0             1  \n4199  0             1  \n4363  0             1  \n4374  0             1  \n4397  0             1  \n4442  0             1  \n4462  0             1  \n4478  1             2  \n4567  0             1  \n4735  0             1  \n5093  0             1  \n5470  0             1  "
     },
     "execution_count": 239,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.loc[train.MentionCount != train['@']][['text','@','MentionCount']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 141,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": "0    5802\nName: isRT, dtype: int64"
     },
     "execution_count": 141,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.isRT.value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# CLF - Root"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Trad"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train_test(X_train, X_test, y_train, y_test, clf):\n",
    "    clf.fit(X_train, y_train)\n",
    "    result = clf.predict(X_test)\n",
    "    print(\"Accuracy:\\t\\t\",accuracy_score(y_test,result))\n",
    "    print('Precision Score:\\t', str(precision_score(y_test,result)))\n",
    "    print('Recall Score:\\t\\t' + str(recall_score(y_test,result)))\n",
    "    print(\"F1 Score:\\t\\t\",f1_score(y_test, result, average='macro', zero_division=True))\n",
    "    print(classification_report(y_test, result))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 337,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.naive_bayes import MultinomialNB\n",
    "\n",
    "tfidf_vectorizer = TfidfVectorizer(\n",
    "    max_features=5000, min_df=1, max_df=0.9, ngram_range=(1,2), analyzer='char')\n",
    "train_pos = tfidf_vectorizer.fit_transform(pheme_POS)\n",
    "test_pos = tfidf_vectorizer.transform(ext_POS)\n",
    "\n",
    "clf = MultinomialNB(alpha=0.1).fit(train_pos, pheme_y)\n",
    "predicted = clf.predict(test_pos)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy:\t\t 0.2556701030927835\n",
      "Precision Score:\t 0.6428571428571429\n",
      "Recall Score:\t\t0.04878048780487805\n",
      "F1 Score:\t\t 0.23033132437434337\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.23      0.91      0.37       116\n",
      "           1       0.64      0.05      0.09       369\n",
      "\n",
      "    accuracy                           0.26       485\n",
      "   macro avg       0.44      0.48      0.23       485\n",
      "weighted avg       0.54      0.26      0.16       485\n",
      "\n"
     ]
    }
   ],
   "source": [
    "clf = SVC()\n",
    "train_test(countvect_train[['$','v']], countvect_test[['$','v']], pheme_y, ext_y, clf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 183,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy:\t\t 0.2556701030927835\n",
      "Precision Score:\t 0.7\n",
      "Recall Score:\t\t0.037940379403794036\n",
      "F1 Score:\t\t 0.2253184607692614\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.24      0.95      0.38       116\n",
      "           1       0.70      0.04      0.07       369\n",
      "\n",
      "    accuracy                           0.26       485\n",
      "   macro avg       0.47      0.49      0.23       485\n",
      "weighted avg       0.59      0.26      0.15       485\n",
      "\n"
     ]
    }
   ],
   "source": [
    "clf = SVC()\n",
    "train_test(pheme_sparse_pos, ext_sparse_pos, pheme_y, ext_y, clf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 190,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy:\t\t 0.27422680412371137\n",
      "Precision Score:\t 0.7575757575757576\n",
      "Recall Score:\t\t0.06775067750677506\n",
      "F1 Score:\t\t 0.25232989979679066\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.24      0.93      0.38       116\n",
      "           1       0.76      0.07      0.12       369\n",
      "\n",
      "    accuracy                           0.27       485\n",
      "   macro avg       0.50      0.50      0.25       485\n",
      "weighted avg       0.63      0.27      0.19       485\n",
      "\n"
     ]
    }
   ],
   "source": [
    "clf = SVC()\n",
    "\n",
    "train_test(pheme_sparse_pos_pronouns[['Numeral','Verb']], ext_sparse_pos_pronouns[['Numeral','Verb']], pheme_y, ext_y, clf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 448,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy:\t\t 0.23917525773195877\n",
      "Precision Score:\t 0.0\n",
      "Recall Score:\t\t0.0\n",
      "F1 Score:\t\t 0.1930116472545757\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.24      1.00      0.39       116\n",
      "           1       0.00      0.00      0.00       369\n",
      "\n",
      "    accuracy                           0.24       485\n",
      "   macro avg       0.12      0.50      0.19       485\n",
      "weighted avg       0.06      0.24      0.09       485\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/june/miniconda3/envs/rosetta/lib/python3.8/site-packages/sklearn/metrics/_classification.py:1245: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Users/june/miniconda3/envs/rosetta/lib/python3.8/site-packages/sklearn/metrics/_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Users/june/miniconda3/envs/rosetta/lib/python3.8/site-packages/sklearn/metrics/_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Users/june/miniconda3/envs/rosetta/lib/python3.8/site-packages/sklearn/metrics/_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    }
   ],
   "source": [
    "clf = SVC()\n",
    "train_test(pheme_sparse_pronouns, ext_sparse_pronouns, pheme_y, ext_y, clf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 444,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy:\t\t 0.23917525773195877\n",
      "Precision Score:\t 0.0\n",
      "Recall Score:\t\t0.0\n",
      "F1 Score:\t\t 0.1930116472545757\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.24      1.00      0.39       116\n",
      "           1       0.00      0.00      0.00       369\n",
      "\n",
      "    accuracy                           0.24       485\n",
      "   macro avg       0.12      0.50      0.19       485\n",
      "weighted avg       0.06      0.24      0.09       485\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/june/miniconda3/envs/rosetta/lib/python3.8/site-packages/sklearn/metrics/_classification.py:1245: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Users/june/miniconda3/envs/rosetta/lib/python3.8/site-packages/sklearn/metrics/_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Users/june/miniconda3/envs/rosetta/lib/python3.8/site-packages/sklearn/metrics/_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Users/june/miniconda3/envs/rosetta/lib/python3.8/site-packages/sklearn/metrics/_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    }
   ],
   "source": [
    "clf = SVC()\n",
    "train_test(pheme_sparse_nopos, ext_sparse_nopos, pheme_y, ext_y, clf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 451,
   "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>Noun</th>\n      <th>Verb</th>\n      <th>Adjective</th>\n      <th>Pronoun</th>\n      <th>Adverb</th>\n      <th>Numeral</th>\n      <th>Conjunction_inj</th>\n      <th>Particle</th>\n      <th>Determiner</th>\n      <th>Modal</th>\n      <th>Whs</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>6</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>2</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>3</td>\n      <td>4</td>\n      <td>6</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>4</td>\n      <td>5</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>6</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>...</th>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n    </tr>\n    <tr>\n      <th>5797</th>\n      <td>2</td>\n      <td>2</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>5798</th>\n      <td>4</td>\n      <td>2</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>5799</th>\n      <td>4</td>\n      <td>3</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>4</td>\n      <td>1</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>5800</th>\n      <td>7</td>\n      <td>3</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>5801</th>\n      <td>2</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n  </tbody>\n</table>\n<p>5802 rows × 11 columns</p>\n</div>",
      "text/plain": "      Noun  Verb  Adjective  Pronoun  Adverb  Numeral  Conjunction_inj  \\\n0        6     3          0        0       0        0                2   \n1        2     2          0        0       0        0                1   \n2        3     4          6        0       1        0                2   \n3        4     5          0        0       0        0                0   \n4        6     2          0        0       0        0                2   \n...    ...   ...        ...      ...     ...      ...              ...   \n5797     2     2          0        1       0        0                2   \n5798     4     2          1        1       0        0                2   \n5799     4     3          1        1       0        0                1   \n5800     7     3          2        0       0        0                3   \n5801     2     3          0        0       0        0                1   \n\n      Particle  Determiner  Modal  Whs  \n0            0           0      0    0  \n1            0           0      0    0  \n2            0           0      0    0  \n3            0           2      0    1  \n4            0           0      0    0  \n...        ...         ...    ...  ...  \n5797         0           0      1    0  \n5798         0           0      0    0  \n5799         0           4      1    0  \n5800         0           1      0    0  \n5801         0           0      0    0  \n\n[5802 rows x 11 columns]"
     },
     "execution_count": 451,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pheme_sparse_pos"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 457,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy:\t\t 0.5752577319587628\n",
      "Precision Score:\t 0.8075471698113208\n",
      "Recall Score:\t\t0.5799457994579946\n",
      "F1 Score:\t\t 0.5309918131290371\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.30      0.56      0.39       116\n",
      "           1       0.81      0.58      0.68       369\n",
      "\n",
      "    accuracy                           0.58       485\n",
      "   macro avg       0.55      0.57      0.53       485\n",
      "weighted avg       0.69      0.58      0.61       485\n",
      "\n"
     ]
    }
   ],
   "source": [
    "train = pd.concat([countvect_train,pheme_sparse_pronouns],axis=1)\n",
    "test = pd.concat([countvect_test,ext_sparse_pronouns],axis=1)\n",
    "clf = GaussianNB()\n",
    "train_test(train, test, pheme_y, ext_y, clf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy:\t\t 0.24536082474226803\n",
      "Precision Score:\t 0.6666666666666666\n",
      "Recall Score:\t\t0.016260162601626018\n",
      "F1 Score:\t\t 0.20675139425139424\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.24      0.97      0.38       116\n",
      "           1       0.67      0.02      0.03       369\n",
      "\n",
      "    accuracy                           0.25       485\n",
      "   macro avg       0.45      0.50      0.21       485\n",
      "weighted avg       0.56      0.25      0.12       485\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# TweetNLP 결과 + 기존 PHEME POS tag의 pronouns 카운트 + POStagging을 제외한 기존 sparse\n",
    "train = pd.concat([countvect_train,pheme_sparse_nopos],axis=1)\n",
    "test = pd.concat([countvect_test,ext_sparse_nopos],axis=1)\n",
    "clf = GaussianNB()\n",
    "train_test(train, test, pheme_y, ext_y, clf)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# MLP"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Library"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torch.optim as optim\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "from torch.utils.data import TensorDataset, DataLoader\n",
    "import torchvision\n",
    "import torchvision.transforms as transforms\n",
    "from torch.utils.data import TensorDataset, DataLoader, RandomSampler, SequentialSampler\n",
    "from torch.utils.data.sampler import WeightedRandomSampler\n",
    "from torch.optim import lr_scheduler\n",
    "\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn.metrics import f1_score, precision_score, recall_score\n",
    "from transformers import AdamW, get_linear_schedule_with_warmup\n",
    "\n",
    "\n",
    "from fetchData import fetchdata, cv_events\n",
    "import __MLP\n",
    "# from __MLP import getSamplers, convert_df_to_unsqueezed_tensor, train_sequential, clf_report\n",
    "import random\n",
    "\n",
    "pd.set_option('display.max_columns', None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 244,
   "metadata": {},
   "outputs": [],
   "source": [
    "pheme_thread =  pd.read_csv('./data/_PHEME_thread_avg.csv')\n",
    "ext_thread =  pd.read_csv('./data/_PHEMEext_thread_avg.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Execution"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 246,
   "metadata": {},
   "outputs": [],
   "source": [
    "# New POS tags + Sparse\n",
    "train = pd.concat([countvect_train,pheme_sparse_nopos],axis=1)\n",
    "test = pd.concat([countvect_test,ext_sparse_nopos],axis=1)\n",
    "# Old POS tags + Sparse\n",
    "train = pd.concat([pheme_sparse_pos,pheme_sparse_nopos],axis=1)\n",
    "test = pd.concat([ext_sparse_pos,ext_sparse_nopos],axis=1)\n",
    "# Old POS tags + Sparse\n",
    "train = pd.concat([pheme_sparse_pos,pheme_sparse_pronouns],axis=1)\n",
    "test = pd.concat([ext_sparse_pos,ext_sparse_pronouns],axis=1)\n",
    "# # New POS tags + Sparse\n",
    "train = pd.concat([countvect_train,pheme_sparse_pronouns],axis=1)\n",
    "test = pd.concat([countvect_test,ext_sparse_pronouns],axis=1)\n",
    "\n",
    "train = pd.concat([countvect_train,pheme_sparse_pos,pheme_text,pheme_thread],axis=1)\n",
    "test = pd.concat([countvect_test,ext_sparse_pos,ext_thread],axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 247,
   "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>!</th>\n      <th>#</th>\n      <th>&amp;</th>\n      <th>,</th>\n      <th>@</th>\n      <th>^</th>\n      <th>a</th>\n      <th>d</th>\n      <th>g</th>\n      <th>l</th>\n      <th>n</th>\n      <th>o</th>\n      <th>p</th>\n      <th>r</th>\n      <th>s</th>\n      <th>t</th>\n      <th>u</th>\n      <th>v</th>\n      <th>x</th>\n      <th>z</th>\n      <th>~</th>\n      <th>Noun</th>\n      <th>Verb</th>\n      <th>Adjective</th>\n      <th>Pronoun</th>\n      <th>Adverb</th>\n      <th>Numeral</th>\n      <th>Conjunction_inj</th>\n      <th>Particle</th>\n      <th>Determiner</th>\n      <th>Modal</th>\n      <th>Whs</th>\n      <th>target</th>\n      <th>depth</th>\n      <th>SUM FriendsCount</th>\n      <th>AVG FriendsCount</th>\n      <th>AVG WordCount</th>\n      <th>SUM WordCount</th>\n      <th>AVG CharCount</th>\n      <th>AVG HashTag</th>\n      <th>SUM HashTag</th>\n      <th>Ratio HashTag</th>\n      <th>SUM Url</th>\n      <th>AVG Url</th>\n      <th>RATIO Url</th>\n      <th>SUM Mention</th>\n      <th>AVG Mention</th>\n      <th>Ratio Mention</th>\n      <th>AVG Statues</th>\n      <th>AVG Listed</th>\n      <th>AVG Follower</th>\n      <th>AVG followers/friend</th>\n      <th>AVG favorite</th>\n      <th>Tweets Count</th>\n      <th>Ratio Verified</th>\n      <th>SUM Verified</th>\n      <th>SUM RT</th>\n      <th>AVG RT</th>\n      <th>AVG AccAge</th>\n      <th>thread_time</th>\n      <th>AVG Emoji</th>\n      <th>RATIO Emoji</th>\n      <th>Ratio Media</th>\n      <th>RATIO Question</th>\n      <th>RATIO Exclaim</th>\n      <th>RATIO Period</th>\n      <th>AVG FPP</th>\n      <th>AVG SPP</th>\n      <th>AVG TPP</th>\n      <th>AVG Skepticism</th>\n      <th>Ratio Skepticism</th>\n      <th>test_auxiliary</th>\n      <th>test_tentat</th>\n      <th>test_certain</th>\n      <th>root_user_ratio</th>\n      <th>unique_user_ratio</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>count</th>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.00000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5.802000e+03</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5.802000e+03</td>\n      <td>5802.000000</td>\n      <td>5.802000e+03</td>\n      <td>5.802000e+03</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5.802000e+03</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n      <td>5802.000000</td>\n    </tr>\n    <tr>\n      <th>mean</th>\n      <td>0.026370</td>\n      <td>0.642709</td>\n      <td>0.292658</td>\n      <td>2.024647</td>\n      <td>0.165288</td>\n      <td>1.958290</td>\n      <td>0.933816</td>\n      <td>1.135643</td>\n      <td>0.030334</td>\n      <td>0.094450</td>\n      <td>3.524130</td>\n      <td>0.503102</td>\n      <td>2.217166</td>\n      <td>0.563771</td>\n      <td>0.023957</td>\n      <td>0.067046</td>\n      <td>0.875733</td>\n      <td>2.687349</td>\n      <td>0.021717</td>\n      <td>0.049293</td>\n      <td>0.048949</td>\n      <td>4.774560</td>\n      <td>2.784385</td>\n      <td>1.25991</td>\n      <td>0.500000</td>\n      <td>0.580834</td>\n      <td>0.286108</td>\n      <td>2.143571</td>\n      <td>0.073251</td>\n      <td>1.042572</td>\n      <td>0.108239</td>\n      <td>0.181834</td>\n      <td>0.339883</td>\n      <td>4.444847</td>\n      <td>2.633307e+04</td>\n      <td>1733.570913</td>\n      <td>14.125338</td>\n      <td>255.770079</td>\n      <td>98.171466</td>\n      <td>0.496359</td>\n      <td>5.682179</td>\n      <td>0.301478</td>\n      <td>2.000862</td>\n      <td>0.169802</td>\n      <td>0.159550</td>\n      <td>30.283006</td>\n      <td>1.410343</td>\n      <td>0.856806</td>\n      <td>2.811923e+04</td>\n      <td>1.188886</td>\n      <td>1.327177e+05</td>\n      <td>4.730672e+03</td>\n      <td>3661.967253</td>\n      <td>17.789038</td>\n      <td>0.090144</td>\n      <td>0.935539</td>\n      <td>593.678904</td>\n      <td>54.183651</td>\n      <td>1301.455119</td>\n      <td>4.204234e+04</td>\n      <td>0.051571</td>\n      <td>0.026130</td>\n      <td>0.150894</td>\n      <td>0.128175</td>\n      <td>0.115823</td>\n      <td>0.680897</td>\n      <td>0.280068</td>\n      <td>0.168354</td>\n      <td>0.364289</td>\n      <td>3.434852</td>\n      <td>0.913416</td>\n      <td>0.454008</td>\n      <td>0.013465</td>\n      <td>0.014904</td>\n      <td>0.160519</td>\n      <td>1.765709</td>\n    </tr>\n    <tr>\n      <th>std</th>\n      <td>0.172674</td>\n      <td>0.826882</td>\n      <td>0.548786</td>\n      <td>1.585656</td>\n      <td>0.447269</td>\n      <td>1.563124</td>\n      <td>0.978707</td>\n      <td>1.186479</td>\n      <td>0.191466</td>\n      <td>0.330134</td>\n      <td>1.765864</td>\n      <td>0.884898</td>\n      <td>1.332783</td>\n      <td>0.842226</td>\n      <td>0.158465</td>\n      <td>0.258928</td>\n      <td>0.748659</td>\n      <td>1.727758</td>\n      <td>0.146947</td>\n      <td>0.224320</td>\n      <td>0.302292</td>\n      <td>2.004439</td>\n      <td>1.745349</td>\n      <td>1.12685</td>\n      <td>0.856996</td>\n      <td>0.846888</td>\n      <td>0.638243</td>\n      <td>1.336208</td>\n      <td>0.272850</td>\n      <td>1.128870</td>\n      <td>0.343388</td>\n      <td>0.431313</td>\n      <td>0.473710</td>\n      <td>3.886665</td>\n      <td>8.265039e+04</td>\n      <td>8908.546640</td>\n      <td>3.310602</td>\n      <td>322.040398</td>\n      <td>20.474917</td>\n      <td>0.589339</td>\n      <td>6.190115</td>\n      <td>0.266754</td>\n      <td>2.683403</td>\n      <td>0.246276</td>\n      <td>0.226585</td>\n      <td>46.743261</td>\n      <td>0.650232</td>\n      <td>0.224672</td>\n      <td>3.792127e+04</td>\n      <td>0.624923</td>\n      <td>6.159211e+05</td>\n      <td>3.625373e+04</td>\n      <td>5009.969132</td>\n      <td>20.094870</td>\n      <td>0.166361</td>\n      <td>1.547161</td>\n      <td>3559.900906</td>\n      <td>253.023384</td>\n      <td>404.981863</td>\n      <td>1.059661e+05</td>\n      <td>0.223547</td>\n      <td>0.075204</td>\n      <td>0.232642</td>\n      <td>0.133937</td>\n      <td>0.135471</td>\n      <td>0.226197</td>\n      <td>0.289474</td>\n      <td>0.227774</td>\n      <td>0.336942</td>\n      <td>1.188174</td>\n      <td>0.125498</td>\n      <td>0.248231</td>\n      <td>0.040178</td>\n      <td>0.054032</td>\n      <td>0.236356</td>\n      <td>2.599912</td>\n    </tr>\n    <tr>\n      <th>min</th>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.00000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>1.000000</td>\n      <td>0.000000e+00</td>\n      <td>0.000000</td>\n      <td>4.000000</td>\n      <td>4.000000</td>\n      <td>32.052632</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>2.100000e+01</td>\n      <td>0.000000</td>\n      <td>1.400000e+01</td>\n      <td>7.650273e-02</td>\n      <td>0.000000</td>\n      <td>1.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>25.000000</td>\n      <td>0.803987</td>\n      <td>0.000000</td>\n      <td>0.000000e+00</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.002890</td>\n      <td>0.031792</td>\n    </tr>\n    <tr>\n      <th>25%</th>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>1.000000</td>\n      <td>0.000000</td>\n      <td>1.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>2.000000</td>\n      <td>0.000000</td>\n      <td>1.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>1.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>3.000000</td>\n      <td>1.000000</td>\n      <td>0.00000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>1.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>2.000000</td>\n      <td>5.072750e+03</td>\n      <td>550.987500</td>\n      <td>12.000000</td>\n      <td>89.000000</td>\n      <td>85.559524</td>\n      <td>0.142857</td>\n      <td>2.000000</td>\n      <td>0.105263</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>8.000000</td>\n      <td>1.000000</td>\n      <td>0.857143</td>\n      <td>1.145628e+04</td>\n      <td>0.798263</td>\n      <td>2.914953e+03</td>\n      <td>2.785679e+00</td>\n      <td>1064.941176</td>\n      <td>7.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>135.000000</td>\n      <td>11.363636</td>\n      <td>1050.459091</td>\n      <td>8.525000e+02</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.545455</td>\n      <td>0.066667</td>\n      <td>0.000000</td>\n      <td>0.125000</td>\n      <td>2.666667</td>\n      <td>0.863636</td>\n      <td>0.294118</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.047619</td>\n      <td>0.523810</td>\n    </tr>\n    <tr>\n      <th>50%</th>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>2.000000</td>\n      <td>0.000000</td>\n      <td>2.000000</td>\n      <td>1.000000</td>\n      <td>1.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>3.000000</td>\n      <td>0.000000</td>\n      <td>2.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>1.000000</td>\n      <td>2.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>5.000000</td>\n      <td>3.000000</td>\n      <td>1.00000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>2.000000</td>\n      <td>0.000000</td>\n      <td>1.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>3.000000</td>\n      <td>1.136250e+04</td>\n      <td>811.346816</td>\n      <td>14.032744</td>\n      <td>188.000000</td>\n      <td>97.699248</td>\n      <td>0.333333</td>\n      <td>4.000000</td>\n      <td>0.230769</td>\n      <td>1.000000</td>\n      <td>0.066667</td>\n      <td>0.066667</td>\n      <td>20.000000</td>\n      <td>1.333333</td>\n      <td>0.937500</td>\n      <td>1.921985e+04</td>\n      <td>1.059736</td>\n      <td>1.156036e+04</td>\n      <td>1.776215e+01</td>\n      <td>2285.028571</td>\n      <td>14.000000</td>\n      <td>0.043478</td>\n      <td>1.000000</td>\n      <td>214.000000</td>\n      <td>21.142857</td>\n      <td>1278.362500</td>\n      <td>6.901500e+03</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.050000</td>\n      <td>0.111111</td>\n      <td>0.090909</td>\n      <td>0.681818</td>\n      <td>0.225000</td>\n      <td>0.105263</td>\n      <td>0.321429</td>\n      <td>3.345804</td>\n      <td>0.964286</td>\n      <td>0.470588</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.071429</td>\n      <td>0.785714</td>\n    </tr>\n    <tr>\n      <th>75%</th>\n      <td>0.000000</td>\n      <td>1.000000</td>\n      <td>1.000000</td>\n      <td>3.000000</td>\n      <td>0.000000</td>\n      <td>3.000000</td>\n      <td>1.000000</td>\n      <td>2.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>5.000000</td>\n      <td>1.000000</td>\n      <td>3.000000</td>\n      <td>1.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>1.000000</td>\n      <td>4.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>6.000000</td>\n      <td>4.000000</td>\n      <td>2.00000</td>\n      <td>1.000000</td>\n      <td>1.000000</td>\n      <td>0.000000</td>\n      <td>3.000000</td>\n      <td>0.000000</td>\n      <td>2.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>1.000000</td>\n      <td>5.000000</td>\n      <td>2.308150e+04</td>\n      <td>1288.510417</td>\n      <td>16.000000</td>\n      <td>306.000000</td>\n      <td>109.600000</td>\n      <td>0.652174</td>\n      <td>8.000000</td>\n      <td>0.421053</td>\n      <td>3.000000</td>\n      <td>0.250000</td>\n      <td>0.227273</td>\n      <td>34.000000</td>\n      <td>1.737489</td>\n      <td>0.962963</td>\n      <td>3.180953e+04</td>\n      <td>1.424568</td>\n      <td>7.510993e+04</td>\n      <td>1.658462e+02</td>\n      <td>4544.706981</td>\n      <td>21.000000</td>\n      <td>0.105263</td>\n      <td>1.000000</td>\n      <td>486.000000</td>\n      <td>42.741071</td>\n      <td>1524.675439</td>\n      <td>3.090225e+04</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.200000</td>\n      <td>0.200000</td>\n      <td>0.166667</td>\n      <td>0.818182</td>\n      <td>0.400000</td>\n      <td>0.250000</td>\n      <td>0.521739</td>\n      <td>4.045455</td>\n      <td>1.000000</td>\n      <td>0.619048</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.142857</td>\n      <td>1.571429</td>\n    </tr>\n    <tr>\n      <th>max</th>\n      <td>3.000000</td>\n      <td>8.000000</td>\n      <td>4.000000</td>\n      <td>15.000000</td>\n      <td>5.000000</td>\n      <td>10.000000</td>\n      <td>7.000000</td>\n      <td>7.000000</td>\n      <td>3.000000</td>\n      <td>4.000000</td>\n      <td>11.000000</td>\n      <td>8.000000</td>\n      <td>7.000000</td>\n      <td>6.000000</td>\n      <td>2.000000</td>\n      <td>2.000000</td>\n      <td>3.000000</td>\n      <td>10.000000</td>\n      <td>2.000000</td>\n      <td>2.000000</td>\n      <td>6.000000</td>\n      <td>13.000000</td>\n      <td>10.000000</td>\n      <td>8.00000</td>\n      <td>7.000000</td>\n      <td>7.000000</td>\n      <td>10.000000</td>\n      <td>8.000000</td>\n      <td>3.000000</td>\n      <td>9.000000</td>\n      <td>3.000000</td>\n      <td>4.000000</td>\n      <td>1.000000</td>\n      <td>48.000000</td>\n      <td>2.783373e+06</td>\n      <td>423700.000000</td>\n      <td>37.750000</td>\n      <td>5320.000000</td>\n      <td>228.000000</td>\n      <td>7.500000</td>\n      <td>97.000000</td>\n      <td>1.000000</td>\n      <td>30.000000</td>\n      <td>2.666667</td>\n      <td>1.000000</td>\n      <td>972.000000</td>\n      <td>6.166667</td>\n      <td>1.000000</td>\n      <td>1.167536e+06</td>\n      <td>6.357100</td>\n      <td>2.171632e+07</td>\n      <td>1.078299e+06</td>\n      <td>165383.000000</td>\n      <td>346.000000</td>\n      <td>1.000000</td>\n      <td>40.000000</td>\n      <td>207563.000000</td>\n      <td>13837.533333</td>\n      <td>3462.666667</td>\n      <td>1.317255e+06</td>\n      <td>7.166667</td>\n      <td>1.000000</td>\n      <td>1.666667</td>\n      <td>1.000000</td>\n      <td>1.666667</td>\n      <td>1.666667</td>\n      <td>4.000000</td>\n      <td>4.000000</td>\n      <td>6.000000</td>\n      <td>12.000000</td>\n      <td>1.000000</td>\n      <td>1.000000</td>\n      <td>1.000000</td>\n      <td>1.000000</td>\n      <td>1.000000</td>\n      <td>11.000000</td>\n    </tr>\n  </tbody>\n</table>\n</div>",
      "text/plain": "                 !            #            &            ,            @  \\\ncount  5802.000000  5802.000000  5802.000000  5802.000000  5802.000000   \nmean      0.026370     0.642709     0.292658     2.024647     0.165288   \nstd       0.172674     0.826882     0.548786     1.585656     0.447269   \nmin       0.000000     0.000000     0.000000     0.000000     0.000000   \n25%       0.000000     0.000000     0.000000     1.000000     0.000000   \n50%       0.000000     0.000000     0.000000     2.000000     0.000000   \n75%       0.000000     1.000000     1.000000     3.000000     0.000000   \nmax       3.000000     8.000000     4.000000    15.000000     5.000000   \n\n                 ^            a            d            g            l  \\\ncount  5802.000000  5802.000000  5802.000000  5802.000000  5802.000000   \nmean      1.958290     0.933816     1.135643     0.030334     0.094450   \nstd       1.563124     0.978707     1.186479     0.191466     0.330134   \nmin       0.000000     0.000000     0.000000     0.000000     0.000000   \n25%       1.000000     0.000000     0.000000     0.000000     0.000000   \n50%       2.000000     1.000000     1.000000     0.000000     0.000000   \n75%       3.000000     1.000000     2.000000     0.000000     0.000000   \nmax      10.000000     7.000000     7.000000     3.000000     4.000000   \n\n                 n            o            p            r            s  \\\ncount  5802.000000  5802.000000  5802.000000  5802.000000  5802.000000   \nmean      3.524130     0.503102     2.217166     0.563771     0.023957   \nstd       1.765864     0.884898     1.332783     0.842226     0.158465   \nmin       0.000000     0.000000     0.000000     0.000000     0.000000   \n25%       2.000000     0.000000     1.000000     0.000000     0.000000   \n50%       3.000000     0.000000     2.000000     0.000000     0.000000   \n75%       5.000000     1.000000     3.000000     1.000000     0.000000   \nmax      11.000000     8.000000     7.000000     6.000000     2.000000   \n\n                 t            u            v            x            z  \\\ncount  5802.000000  5802.000000  5802.000000  5802.000000  5802.000000   \nmean      0.067046     0.875733     2.687349     0.021717     0.049293   \nstd       0.258928     0.748659     1.727758     0.146947     0.224320   \nmin       0.000000     0.000000     0.000000     0.000000     0.000000   \n25%       0.000000     0.000000     1.000000     0.000000     0.000000   \n50%       0.000000     1.000000     2.000000     0.000000     0.000000   \n75%       0.000000     1.000000     4.000000     0.000000     0.000000   \nmax       2.000000     3.000000    10.000000     2.000000     2.000000   \n\n                 ~         Noun         Verb   Adjective      Pronoun  \\\ncount  5802.000000  5802.000000  5802.000000  5802.00000  5802.000000   \nmean      0.048949     4.774560     2.784385     1.25991     0.500000   \nstd       0.302292     2.004439     1.745349     1.12685     0.856996   \nmin       0.000000     0.000000     0.000000     0.00000     0.000000   \n25%       0.000000     3.000000     1.000000     0.00000     0.000000   \n50%       0.000000     5.000000     3.000000     1.00000     0.000000   \n75%       0.000000     6.000000     4.000000     2.00000     1.000000   \nmax       6.000000    13.000000    10.000000     8.00000     7.000000   \n\n            Adverb      Numeral  Conjunction_inj     Particle   Determiner  \\\ncount  5802.000000  5802.000000      5802.000000  5802.000000  5802.000000   \nmean      0.580834     0.286108         2.143571     0.073251     1.042572   \nstd       0.846888     0.638243         1.336208     0.272850     1.128870   \nmin       0.000000     0.000000         0.000000     0.000000     0.000000   \n25%       0.000000     0.000000         1.000000     0.000000     0.000000   \n50%       0.000000     0.000000         2.000000     0.000000     1.000000   \n75%       1.000000     0.000000         3.000000     0.000000     2.000000   \nmax       7.000000    10.000000         8.000000     3.000000     9.000000   \n\n             Modal          Whs       target        depth  SUM FriendsCount  \\\ncount  5802.000000  5802.000000  5802.000000  5802.000000      5.802000e+03   \nmean      0.108239     0.181834     0.339883     4.444847      2.633307e+04   \nstd       0.343388     0.431313     0.473710     3.886665      8.265039e+04   \nmin       0.000000     0.000000     0.000000     1.000000      0.000000e+00   \n25%       0.000000     0.000000     0.000000     2.000000      5.072750e+03   \n50%       0.000000     0.000000     0.000000     3.000000      1.136250e+04   \n75%       0.000000     0.000000     1.000000     5.000000      2.308150e+04   \nmax       3.000000     4.000000     1.000000    48.000000      2.783373e+06   \n\n       AVG FriendsCount  AVG WordCount  SUM WordCount  AVG CharCount  \\\ncount       5802.000000    5802.000000    5802.000000    5802.000000   \nmean        1733.570913      14.125338     255.770079      98.171466   \nstd         8908.546640       3.310602     322.040398      20.474917   \nmin            0.000000       4.000000       4.000000      32.052632   \n25%          550.987500      12.000000      89.000000      85.559524   \n50%          811.346816      14.032744     188.000000      97.699248   \n75%         1288.510417      16.000000     306.000000     109.600000   \nmax       423700.000000      37.750000    5320.000000     228.000000   \n\n       AVG HashTag  SUM HashTag  Ratio HashTag      SUM Url      AVG Url  \\\ncount  5802.000000  5802.000000    5802.000000  5802.000000  5802.000000   \nmean      0.496359     5.682179       0.301478     2.000862     0.169802   \nstd       0.589339     6.190115       0.266754     2.683403     0.246276   \nmin       0.000000     0.000000       0.000000     0.000000     0.000000   \n25%       0.142857     2.000000       0.105263     0.000000     0.000000   \n50%       0.333333     4.000000       0.230769     1.000000     0.066667   \n75%       0.652174     8.000000       0.421053     3.000000     0.250000   \nmax       7.500000    97.000000       1.000000    30.000000     2.666667   \n\n         RATIO Url  SUM Mention  AVG Mention  Ratio Mention   AVG Statues  \\\ncount  5802.000000  5802.000000  5802.000000    5802.000000  5.802000e+03   \nmean      0.159550    30.283006     1.410343       0.856806  2.811923e+04   \nstd       0.226585    46.743261     0.650232       0.224672  3.792127e+04   \nmin       0.000000     0.000000     0.000000       0.000000  2.100000e+01   \n25%       0.000000     8.000000     1.000000       0.857143  1.145628e+04   \n50%       0.066667    20.000000     1.333333       0.937500  1.921985e+04   \n75%       0.227273    34.000000     1.737489       0.962963  3.180953e+04   \nmax       1.000000   972.000000     6.166667       1.000000  1.167536e+06   \n\n        AVG Listed  AVG Follower  AVG followers/friend   AVG favorite  \\\ncount  5802.000000  5.802000e+03          5.802000e+03    5802.000000   \nmean      1.188886  1.327177e+05          4.730672e+03    3661.967253   \nstd       0.624923  6.159211e+05          3.625373e+04    5009.969132   \nmin       0.000000  1.400000e+01          7.650273e-02       0.000000   \n25%       0.798263  2.914953e+03          2.785679e+00    1064.941176   \n50%       1.059736  1.156036e+04          1.776215e+01    2285.028571   \n75%       1.424568  7.510993e+04          1.658462e+02    4544.706981   \nmax       6.357100  2.171632e+07          1.078299e+06  165383.000000   \n\n       Tweets Count  Ratio Verified  SUM Verified         SUM RT  \\\ncount   5802.000000     5802.000000   5802.000000    5802.000000   \nmean      17.789038        0.090144      0.935539     593.678904   \nstd       20.094870        0.166361      1.547161    3559.900906   \nmin        1.000000        0.000000      0.000000      25.000000   \n25%        7.000000        0.000000      0.000000     135.000000   \n50%       14.000000        0.043478      1.000000     214.000000   \n75%       21.000000        0.105263      1.000000     486.000000   \nmax      346.000000        1.000000     40.000000  207563.000000   \n\n             AVG RT   AVG AccAge   thread_time    AVG Emoji  RATIO Emoji  \\\ncount   5802.000000  5802.000000  5.802000e+03  5802.000000  5802.000000   \nmean      54.183651  1301.455119  4.204234e+04     0.051571     0.026130   \nstd      253.023384   404.981863  1.059661e+05     0.223547     0.075204   \nmin        0.803987     0.000000  0.000000e+00     0.000000     0.000000   \n25%       11.363636  1050.459091  8.525000e+02     0.000000     0.000000   \n50%       21.142857  1278.362500  6.901500e+03     0.000000     0.000000   \n75%       42.741071  1524.675439  3.090225e+04     0.000000     0.000000   \nmax    13837.533333  3462.666667  1.317255e+06     7.166667     1.000000   \n\n       Ratio Media  RATIO Question  RATIO Exclaim  RATIO Period      AVG FPP  \\\ncount  5802.000000     5802.000000    5802.000000   5802.000000  5802.000000   \nmean      0.150894        0.128175       0.115823      0.680897     0.280068   \nstd       0.232642        0.133937       0.135471      0.226197     0.289474   \nmin       0.000000        0.000000       0.000000      0.000000     0.000000   \n25%       0.000000        0.000000       0.000000      0.545455     0.066667   \n50%       0.050000        0.111111       0.090909      0.681818     0.225000   \n75%       0.200000        0.200000       0.166667      0.818182     0.400000   \nmax       1.666667        1.000000       1.666667      1.666667     4.000000   \n\n           AVG SPP      AVG TPP  AVG Skepticism  Ratio Skepticism  \\\ncount  5802.000000  5802.000000     5802.000000       5802.000000   \nmean      0.168354     0.364289        3.434852          0.913416   \nstd       0.227774     0.336942        1.188174          0.125498   \nmin       0.000000     0.000000        0.000000          0.000000   \n25%       0.000000     0.125000        2.666667          0.863636   \n50%       0.105263     0.321429        3.345804          0.964286   \n75%       0.250000     0.521739        4.045455          1.000000   \nmax       4.000000     6.000000       12.000000          1.000000   \n\n       test_auxiliary  test_tentat  test_certain  root_user_ratio  \\\ncount     5802.000000  5802.000000   5802.000000      5802.000000   \nmean         0.454008     0.013465      0.014904         0.160519   \nstd          0.248231     0.040178      0.054032         0.236356   \nmin          0.000000     0.000000      0.000000         0.002890   \n25%          0.294118     0.000000      0.000000         0.047619   \n50%          0.470588     0.000000      0.000000         0.071429   \n75%          0.619048     0.000000      0.000000         0.142857   \nmax          1.000000     1.000000      1.000000         1.000000   \n\n       unique_user_ratio  \ncount        5802.000000  \nmean            1.765709  \nstd             2.599912  \nmin             0.031792  \n25%             0.523810  \n50%             0.785714  \n75%             1.571429  \nmax            11.000000  "
     },
     "execution_count": 247,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "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>v</th>\n      <th>Verb</th>\n      <th>n</th>\n      <th>Noun</th>\n      <th>p</th>\n      <th>Pronoun</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>BREAKING: Armed man takes hostage in kosher grocery east of Paris http://t.co/PBs3sMwhLt</td>\n      <td>1</td>\n      <td>3</td>\n      <td>5</td>\n      <td>6</td>\n      <td>2</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>#CharlieHebdo killers dead, confirmed by gendarmerie.</td>\n      <td>1</td>\n      <td>2</td>\n      <td>1</td>\n      <td>2</td>\n      <td>1</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>Top French cartoonists Charb, Cabu, Wolinski, Tignous confirmed among dead in #Paris #CharlieHebdo attack. Editor is critically wounded.</td>\n      <td>3</td>\n      <td>4</td>\n      <td>3</td>\n      <td>3</td>\n      <td>2</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>Police have surrounded the area where the #CharlieHebdo attack suspects are believed to be: http://t.co/3tGXEIX4F2\\nhttps://t.co/aBSezf2QWS</td>\n      <td>7</td>\n      <td>5</td>\n      <td>1</td>\n      <td>4</td>\n      <td>1</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>PHOTO: Armed gunmen face police officers near #CharlieHebdo HQ in Paris http://t.co/3Jsosc7yl3 http://t.co/iOpVNO6Iq0</td>\n      <td>0</td>\n      <td>2</td>\n      <td>5</td>\n      <td>6</td>\n      <td>2</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>...</th>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n    </tr>\n    <tr>\n      <th>5797</th>\n      <td>'I'll ride with you' http://t.co/llZnuCAzg5 Australia unites during #SydneySiege http://t.co/WIU22VPgkz</td>\n      <td>2</td>\n      <td>2</td>\n      <td>0</td>\n      <td>2</td>\n      <td>2</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>5798</th>\n      <td>Canada's thoughts and prayers are with our Australian friends. #MartinPlace #SydneySiege</td>\n      <td>1</td>\n      <td>2</td>\n      <td>3</td>\n      <td>4</td>\n      <td>1</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>5799</th>\n      <td>Every non-muslim in the world must watch this video https://t.co/sZdyhISoVh &amp;amp; show it every other non-muslim! #sydneysiege</td>\n      <td>3</td>\n      <td>3</td>\n      <td>2</td>\n      <td>4</td>\n      <td>1</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>5800</th>\n      <td>Suspect in Sydney cafe siege identified as Man Haron Monis, an Iranian granted asylum in Australia http://t.co/6Lrl9DEMXA</td>\n      <td>1</td>\n      <td>3</td>\n      <td>6</td>\n      <td>7</td>\n      <td>3</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>5801</th>\n      <td>Australians respond to racism by telling #Muslim community #illridewithyou. #sydneysiege #MartinPlace</td>\n      <td>2</td>\n      <td>3</td>\n      <td>3</td>\n      <td>2</td>\n      <td>2</td>\n      <td>0</td>\n    </tr>\n  </tbody>\n</table>\n<p>5802 rows × 7 columns</p>\n</div>",
      "text/plain": "                                                                                                                                             text  \\\n0                                                        BREAKING: Armed man takes hostage in kosher grocery east of Paris http://t.co/PBs3sMwhLt   \n1                                                                                           #CharlieHebdo killers dead, confirmed by gendarmerie.   \n2        Top French cartoonists Charb, Cabu, Wolinski, Tignous confirmed among dead in #Paris #CharlieHebdo attack. Editor is critically wounded.   \n3     Police have surrounded the area where the #CharlieHebdo attack suspects are believed to be: http://t.co/3tGXEIX4F2\\nhttps://t.co/aBSezf2QWS   \n4                           PHOTO: Armed gunmen face police officers near #CharlieHebdo HQ in Paris http://t.co/3Jsosc7yl3 http://t.co/iOpVNO6Iq0   \n...                                                                                                                                           ...   \n5797                                      'I'll ride with you' http://t.co/llZnuCAzg5 Australia unites during #SydneySiege http://t.co/WIU22VPgkz   \n5798                                                     Canada's thoughts and prayers are with our Australian friends. #MartinPlace #SydneySiege   \n5799               Every non-muslim in the world must watch this video https://t.co/sZdyhISoVh &amp; show it every other non-muslim! #sydneysiege   \n5800                    Suspect in Sydney cafe siege identified as Man Haron Monis, an Iranian granted asylum in Australia http://t.co/6Lrl9DEMXA   \n5801                                        Australians respond to racism by telling #Muslim community #illridewithyou. #sydneysiege #MartinPlace   \n\n      v  Verb  n  Noun  p  Pronoun  \n0     1     3  5     6  2        0  \n1     1     2  1     2  1        0  \n2     3     4  3     3  2        0  \n3     7     5  1     4  1        0  \n4     0     2  5     6  2        0  \n...  ..   ... ..   ... ..      ...  \n5797  2     2  0     2  2        1  \n5798  1     2  3     4  1        1  \n5799  3     3  2     4  1        1  \n5800  1     3  6     7  3        0  \n5801  2     3  3     2  2        0  \n\n[5802 rows x 7 columns]"
     },
     "execution_count": 131,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train[['text','v','Verb','n','Noun', 'p','Pronoun']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 251,
   "metadata": {},
   "outputs": [
    {
     "ename": "ValueError",
     "evalue": "could not determine the shape of object type 'DataFrame'",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mValueError\u001b[0m                                Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-251-4c0e0150aa7a>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mtensor_x1\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtensor_y1\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtensor_x2\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtensor_y2\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m__MLP\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mconvert_df_to_unsqueezed_tensor\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrain\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mpheme_y\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtest\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mext_y\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[0mtrain_dataset\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mTensorDataset\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtensor_x1\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mtensor_y1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0mtest_dataset\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mTensorDataset\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtensor_x2\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mtensor_y2\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0mbatch_size\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;36m16\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/Documents/Code/FYP/code_data/__MLP.py\u001b[0m in \u001b[0;36mconvert_df_to_unsqueezed_tensor\u001b[0;34m(trainX, trainY, testX, testY, TrainX2, TestX2)\u001b[0m\n\u001b[1;32m    542\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    543\u001b[0m \u001b[0;32mdef\u001b[0m \u001b[0mconvert_df_to_unsqueezed_tensor\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrainX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtrainY\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtestX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtestY\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mTrainX2\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mTestX2\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 544\u001b[0;31m     \u001b[0mtensor_x1\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTensor\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrainX\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0munsqueeze\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    545\u001b[0m     \u001b[0mtensor_y1\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTensor\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrainY\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0munsqueeze\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    546\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mValueError\u001b[0m: could not determine the shape of object type 'DataFrame'"
     ]
    }
   ],
   "source": [
    "tensor_x1, tensor_y1, tensor_x2, tensor_y2 = __MLP.convert_df_to_unsqueezed_tensor(train, pheme_y, test, ext_y)\n",
    "train_dataset = TensorDataset(tensor_x1,tensor_y1)\n",
    "test_dataset = TensorDataset(tensor_x2,tensor_y2)\n",
    "\n",
    "batch_size = 16\n",
    "\n",
    "train_sampler, test_sampler = __MLP.getSamplers(pheme_y, tensor_x2)\n",
    "train_dataloader = DataLoader(train_dataset, batch_size=batch_size, sampler=train_sampler, num_workers=2)\n",
    "test_dataloader = DataLoader(test_dataset, batch_size=batch_size, shuffle=False, num_workers=2)\n",
    "\n",
    "data = next(iter(train_dataloader))\n",
    "print(\"mean: %s, std: %s\" %(data[0].mean(), data[0].std()))\n",
    "\n",
    "train_size = int(tensor_y1.size(0))\n",
    "test_size = int(tensor_y2.size(0))\n",
    "\n",
    "print(tensor_x1.size(),tensor_x2.size())\n",
    "print(tensor_y1.size(),tensor_y2.size())\n",
    "print(\"Train Size\",train_size,\"Test Size\",test_size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [],
   "source": [
    "class POS_net(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(POS_net, self).__init__() # 1*20\n",
    "        self.fc1 = nn.Linear(26, 3, bias=True) # 420\n",
    "        # self.fc2 = nn.Linear(12, 1, bias=True)\n",
    "        self.fc3 = nn.Linear(3, 1)\n",
    "\n",
    "        self.drop_3 = nn.Dropout(0.3)\n",
    "        self.drop_4 = nn.Dropout(0.4)\n",
    "        self.drop_2 = nn.Dropout(0.2)\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = F.elu(self.fc1(x))\n",
    "        # x = F.elu(self.fc3(x))\n",
    "        x = self.fc3(x)\n",
    "        return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {},
   "outputs": [],
   "source": [
    "epochs=100\n",
    "model_bert = POS_net()\n",
    "\n",
    "criterion = nn.BCEWithLogitsLoss()\n",
    "optimizer = optim.Adam(model_bert.parameters(), lr=5e-4, eps=1e-8, weight_decay=1e-5)\n",
    "train_loss = []\n",
    "train_accuracy = []\n",
    "total_steps = len(train_dataloader) * epochs\n",
    "scheduler = get_linear_schedule_with_warmup(optimizer,\n",
    "                                            num_warmup_steps=0,  # Default value\n",
    "                                            num_training_steps=total_steps)\n",
    "\n",
    "PATH = \"./Model/temp_model.pt\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 0/149\tTrain) Acc: 0.5681, Loss: 0.6886\n",
      "\t\tValidation) Acc: 0.6392 Loss:0.6546\n",
      "\t\t\tSaving the best model w/ val loss 0.6546\n",
      "Epoch 2/149\tTrain) Acc: 0.6534, Loss: 0.6430\n",
      "\t\tValidation) Acc: 0.6000 Loss:0.6685\n",
      "Epoch 4/149\tTrain) Acc: 0.6680, Loss: 0.6169\n",
      "\t\tValidation) Acc: 0.5526 Loss:0.7153\n",
      "Epoch 6/149\tTrain) Acc: 0.6667, Loss: 0.6162\n",
      "\t\tValidation) Acc: 0.5608 Loss:0.7215\n",
      "Epoch 8/149\tTrain) Acc: 0.6679, Loss: 0.6139\n",
      "\t\tValidation) Acc: 0.5402 Loss:0.7359\n",
      "Epoch 10/149\tTrain) Acc: 0.6779, Loss: 0.6035\n",
      "\t\tValidation) Acc: 0.5443 Loss:0.7411\n",
      "Epoch 12/149\tTrain) Acc: 0.6751, Loss: 0.5974\n",
      "\t\tValidation) Acc: 0.5175 Loss:0.7608\n",
      "Epoch 14/149\tTrain) Acc: 0.6813, Loss: 0.6031\n",
      "\t\tValidation) Acc: 0.5691 Loss:0.7297\n",
      "Epoch 16/149\tTrain) Acc: 0.6839, Loss: 0.6016\n",
      "\t\tValidation) Acc: 0.5402 Loss:0.7564\n",
      "Epoch 18/149\tTrain) Acc: 0.6975, Loss: 0.5910\n",
      "\t\tValidation) Acc: 0.5691 Loss:0.7362\n",
      "Epoch 20/149\tTrain) Acc: 0.6732, Loss: 0.6076\n",
      "\t\tValidation) Acc: 0.5278 Loss:0.7589\n",
      "Epoch 22/149\tTrain) Acc: 0.7015, Loss: 0.5880\n",
      "\t\tValidation) Acc: 0.5361 Loss:0.7621\n",
      "Epoch 24/149\tTrain) Acc: 0.6917, Loss: 0.5950\n",
      "\t\tValidation) Acc: 0.5464 Loss:0.7627\n",
      "Epoch 26/149\tTrain) Acc: 0.6779, Loss: 0.6033\n",
      "\t\tValidation) Acc: 0.5093 Loss:0.7882\n",
      "Epoch 28/149\tTrain) Acc: 0.6837, Loss: 0.5994\n",
      "\t\tValidation) Acc: 0.5629 Loss:0.7443\n",
      "Epoch 30/149\tTrain) Acc: 0.6830, Loss: 0.5961\n",
      "\t\tValidation) Acc: 0.5546 Loss:0.7555\n",
      "Finishing the Model: Val Loss is not decreasing...\n",
      "[0.7797656726591365, 0.788154982291546, 0.7616060107024675, 0.7443466557669886, 0.7505151243553948]\n"
     ]
    }
   ],
   "source": [
    "train_acc, train_loss, val_acc, val_loss_list = __MLP.train_sequential(model=model_bert, num_epochs=150, patience=30, criterion=criterion, optimizer=optimizer, scheduler=scheduler, train_loader=train_dataloader,train_size=train_size, test_loader=test_dataloader, test_size=test_size, PATH=PATH)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": "<Figure size 864x576 with 2 Axes>"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Min of Training Loss: 0.587970\n",
      "Max of Training Accuracy: 0.701482\n",
      "Mean of Training Loss: 0.610001\n",
      "Mean of Training Accuracy: 0.672087\n",
      "------------\n",
      "Max of Testing Accuracy: 0.639175\n",
      "Min of Testing Loss: 0.654601\n",
      "Mean of Testing Loss: 0.738705\n",
      "Mean of Testing Accuracy: 0.550848\n",
      "------------\n"
     ]
    }
   ],
   "source": [
    "__MLP.clf_report(train_loss, train_acc, val_loss_list, val_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "accuracy_score:\t\t0.6392\n",
      "Precision Score:\t0.8464\n",
      "Recall Score:\t\t0.6423\n",
      "f1_score:\t\t0.5996\n",
      "Test_loss:\t\t0.6546\n"
     ]
    }
   ],
   "source": [
    "model = POS_net()\n",
    "PATH = './Model/temp_model.pt'\n",
    "model.load_state_dict(torch.load(PATH))\n",
    "__MLP.predict(model, criterion, test_dataloader, test_size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.8.8 64-bit (conda)",
   "name": "python388jvsc74a57bd0b3e779290c3971bcb91630500d66c3c3cecd721489b4b277b4ac4fef67ef773c"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.8"
  },
  "metadata": {
   "interpreter": {
    "hash": "b3e779290c3971bcb91630500d66c3c3cecd721489b4b277b4ac4fef67ef773c"
   }
  },
  "orig_nbformat": 2
 },
 "nbformat": 4,
 "nbformat_minor": 2
}