NLP - Predicting Disaster Tweets Using Logistic Regression.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Predict whether the Tweet is a real Disaster or Not?\n",
    "### Target = 1 means Real Disaster Tweet or else Target =0 Not Real Disaster Tweet"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "import re,string\n",
    "import category_encoders as ce\n",
    "\n",
    "\n",
    "import nltk\n",
    "from nltk import FreqDist,word_tokenize\n",
    "from wordcloud import STOPWORDS\n",
    "from nltk import bigrams\n",
    "\n",
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "from sklearn.feature_extraction.text import TfidfVectorizer\n",
    "\n",
    "from sklearn.metrics import f1_score\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.feature_selection import RFECV"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Exploratory Data Analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Length of Train Data 7613\n",
      "Length of Test Data 3263\n"
     ]
    }
   ],
   "source": [
    "train_data = pd.read_csv(\"train.csv\")\n",
    "test_data = pd.read_csv(\"test.csv\")\n",
    "submission = pd.read_csv(\"sample_submission.csv\")\n",
    "print(\"Length of Train Data\",len(train_data))\n",
    "print(\"Length of Test Data\",len(test_data))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "There are 7613 rows and 5 columns in train\n",
      "There are 3263 rows and 4 columns in train\n"
     ]
    }
   ],
   "source": [
    "print('There are {} rows and {} columns in train'.format(train_data.shape[0],train_data.shape[1]))\n",
    "print('There are {} rows and {} columns in train'.format(test_data.shape[0],test_data.shape[1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check for Duplicates\n",
    "train_data.duplicated().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id</th>\n",
       "      <th>keyword</th>\n",
       "      <th>location</th>\n",
       "      <th>text</th>\n",
       "      <th>target</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Our Deeds are the Reason of this #earthquake M...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Forest fire near La Ronge Sask. Canada</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>All residents asked to 'shelter in place' are ...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>6</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>13,000 people receive #wildfires evacuation or...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>7</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Just got sent this photo from Ruby #Alaska as ...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7608</th>\n",
       "      <td>10869</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Two giant cranes holding a bridge collapse int...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7609</th>\n",
       "      <td>10870</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>@aria_ahrary @TheTawniest The out of control w...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7610</th>\n",
       "      <td>10871</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>M1.94 [01:04 UTC]?5km S of Volcano Hawaii. htt...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7611</th>\n",
       "      <td>10872</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Police investigating after an e-bike collided ...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7612</th>\n",
       "      <td>10873</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>The Latest: More Homes Razed by Northern Calif...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>7613 rows × 5 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         id keyword location  \\\n",
       "0         1     NaN      NaN   \n",
       "1         4     NaN      NaN   \n",
       "2         5     NaN      NaN   \n",
       "3         6     NaN      NaN   \n",
       "4         7     NaN      NaN   \n",
       "...     ...     ...      ...   \n",
       "7608  10869     NaN      NaN   \n",
       "7609  10870     NaN      NaN   \n",
       "7610  10871     NaN      NaN   \n",
       "7611  10872     NaN      NaN   \n",
       "7612  10873     NaN      NaN   \n",
       "\n",
       "                                                   text  target  \n",
       "0     Our Deeds are the Reason of this #earthquake M...       1  \n",
       "1                Forest fire near La Ronge Sask. Canada       1  \n",
       "2     All residents asked to 'shelter in place' are ...       1  \n",
       "3     13,000 people receive #wildfires evacuation or...       1  \n",
       "4     Just got sent this photo from Ruby #Alaska as ...       1  \n",
       "...                                                 ...     ...  \n",
       "7608  Two giant cranes holding a bridge collapse int...       1  \n",
       "7609  @aria_ahrary @TheTawniest The out of control w...       1  \n",
       "7610  M1.94 [01:04 UTC]?5km S of Volcano Hawaii. htt...       1  \n",
       "7611  Police investigating after an e-bike collided ...       1  \n",
       "7612  The Latest: More Homes Razed by Northern Calif...       1  \n",
       "\n",
       "[7613 rows x 5 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data = train_data.drop_duplicates().reset_index(drop=True)\n",
    "train_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Class balance\n",
    "# train.target.value_counts()\n",
    "sns.countplot(y=train_data.target);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Data id             0\n",
      "keyword       61\n",
      "location    2533\n",
      "text           0\n",
      "target         0\n",
      "dtype: int64\n",
      "Test Data id             0\n",
      "keyword       26\n",
      "location    1105\n",
      "text           0\n",
      "dtype: int64\n"
     ]
    }
   ],
   "source": [
    "\"\"\"\n",
    "Text is all non-null. Only a small percentage of tweets have no keyword. Location has much more null values.\n",
    "\"\"\"\n",
    "print(\"Train Data\",train_data.isnull().sum())\n",
    "print(\"Test Data\",test_data.isnull().sum())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "         id  keyword                       location  \\\n",
      "31       48   ablaze                     Birmingham   \n",
      "32       49   ablaze  Est. September 2012 - Bristol   \n",
      "33       50   ablaze                         AFRICA   \n",
      "34       52   ablaze               Philadelphia, PA   \n",
      "35       53   ablaze                     London, UK   \n",
      "...     ...      ...                            ...   \n",
      "7578  10830  wrecked                            NaN   \n",
      "7579  10831  wrecked              Vancouver, Canada   \n",
      "7580  10832  wrecked                        London    \n",
      "7581  10833  wrecked                        Lincoln   \n",
      "7582  10834  wrecked                            NaN   \n",
      "\n",
      "                                                   text  target  \n",
      "31    @bbcmtd Wholesale Markets ablaze http://t.co/l...       1  \n",
      "32    We always try to bring the heavy. #metal #RT h...       0  \n",
      "33    #AFRICANBAZE: Breaking news:Nigeria flag set a...       1  \n",
      "34                   Crying out for more! Set me ablaze       0  \n",
      "35    On plus side LOOK AT THE SKY LAST NIGHT IT WAS...       0  \n",
      "...                                                 ...     ...  \n",
      "7578   @jt_ruff23 @cameronhacker and I wrecked you both       0  \n",
      "7579  Three days off from work and they've pretty mu...       0  \n",
      "7580  #FX #forex #trading Cramer: Iger's 3 words tha...       0  \n",
      "7581  @engineshed Great atmosphere at the British Li...       0  \n",
      "7582  Cramer: Iger's 3 words that wrecked Disney's s...       0  \n",
      "\n",
      "[7552 rows x 5 columns]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'samples')"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(train_data[~train_data['keyword'].isnull()])\n",
    "\"\"\"\n",
    "Class Distribution of the Target Variable. There are more tweets with class 0 ( No disaster) \n",
    "than class 1 ( disaster tweets)\n",
    "\"\"\"\n",
    "x = train_data.target.value_counts()\n",
    "sns.barplot(x.index,x)\n",
    "plt.gca().set_ylabel('samples')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "221 221\n",
      "set()\n"
     ]
    }
   ],
   "source": [
    "# Check number of unique keywords, and whether they are the same for train and test sets\n",
    "print (train_data.keyword.nunique(), test_data.keyword.nunique())\n",
    "print (set(train_data.keyword.unique()) - set(test_data.keyword.unique()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 648x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(9,6))\n",
    "sns.countplot(y=train_data.keyword, order = train_data.keyword.value_counts().iloc[:15].index)\n",
    "plt.title('Top 15 keywords')\n",
    "plt.show()\n",
    "# train.keyword.value_counts().head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 936x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\"\"\"\n",
    "There is no common top 10 keywords between Disaster and Non-Disaster Tweets\n",
    "\"\"\"\n",
    "diasater_keyword = train_data[train_data['target'] == 1].keyword.value_counts().head(10)\n",
    "nonDisaster_keyword = train_data[train_data['target'] == 0].keyword.value_counts().head(10)\n",
    "\n",
    "plt.figure(figsize=(13,5))\n",
    "plt.subplot(121)\n",
    "sns.barplot(diasater_keyword, diasater_keyword.index)\n",
    "plt.title('Top keywords for disaster tweets')\n",
    "plt.subplot(122)\n",
    "sns.barplot(nonDisaster_keyword, nonDisaster_keyword.index)\n",
    "plt.title('Top keywords for non-disaster tweets')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 936x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "top_d = train_data.groupby('keyword').mean()['target'].sort_values(ascending=False).head(10)\n",
    "top_nd = train_data.groupby('keyword').mean()['target'].sort_values().head(10)\n",
    "\n",
    "plt.figure(figsize=(13,5))\n",
    "plt.subplot(121)\n",
    "sns.barplot(top_d, top_d.index, color='pink')\n",
    "plt.title('Keywords with highest % of disaster tweets')\n",
    "plt.subplot(122)\n",
    "sns.barplot(top_nd, top_nd.index, color='yellow')\n",
    "plt.title('Keywords with lowest % of disaster tweets')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data Cleaning & Creating Seperate Hashtags , Mentions and Links"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id</th>\n",
       "      <th>keyword</th>\n",
       "      <th>location</th>\n",
       "      <th>text</th>\n",
       "      <th>target</th>\n",
       "      <th>text_clean</th>\n",
       "      <th>hashtags</th>\n",
       "      <th>mentions</th>\n",
       "      <th>links</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Our Deeds are the Reason of this #earthquake M...</td>\n",
       "      <td>1</td>\n",
       "      <td>Our Deeds are the Reason of this #earthquake M...</td>\n",
       "      <td>earthquake</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Forest fire near La Ronge Sask. Canada</td>\n",
       "      <td>1</td>\n",
       "      <td>Forest fire near La Ronge Sask. Canada</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>All residents asked to 'shelter in place' are ...</td>\n",
       "      <td>1</td>\n",
       "      <td>All residents asked to 'shelter in place' are ...</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>6</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>13,000 people receive #wildfires evacuation or...</td>\n",
       "      <td>1</td>\n",
       "      <td>13,000 people receive #wildfires evacuation or...</td>\n",
       "      <td>wildfires</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>7</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Just got sent this photo from Ruby #Alaska as ...</td>\n",
       "      <td>1</td>\n",
       "      <td>Just got sent this photo from Ruby #Alaska as ...</td>\n",
       "      <td>Alaska wildfires</td>\n",
       "      <td>no</td>\n",
       "      <td>no</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7608</th>\n",
       "      <td>10869</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Two giant cranes holding a bridge collapse int...</td>\n",
       "      <td>1</td>\n",
       "      <td>Two giant cranes holding a bridge collapse int...</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "      <td>http://t.co/STfMbbZFB5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7609</th>\n",
       "      <td>10870</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>@aria_ahrary @TheTawniest The out of control w...</td>\n",
       "      <td>1</td>\n",
       "      <td>@aria_ahrary @TheTawniest The out of control w...</td>\n",
       "      <td>no</td>\n",
       "      <td>aria_ahrary TheTawniest</td>\n",
       "      <td>no</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7610</th>\n",
       "      <td>10871</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>M1.94 [01:04 UTC]?5km S of Volcano Hawaii. htt...</td>\n",
       "      <td>1</td>\n",
       "      <td>M1.94 [01:04 UTC]?5km S of Volcano Hawaii.</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "      <td>http://t.co/zDtoyd8EbJ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7611</th>\n",
       "      <td>10872</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Police investigating after an e-bike collided ...</td>\n",
       "      <td>1</td>\n",
       "      <td>Police investigating after an e-bike collided ...</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7612</th>\n",
       "      <td>10873</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>The Latest: More Homes Razed by Northern Calif...</td>\n",
       "      <td>1</td>\n",
       "      <td>The Latest: More Homes Razed by Northern Calif...</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "      <td>http://t.co/YmY4rSkQ3d</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>7613 rows × 9 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         id keyword location  \\\n",
       "0         1     NaN      NaN   \n",
       "1         4     NaN      NaN   \n",
       "2         5     NaN      NaN   \n",
       "3         6     NaN      NaN   \n",
       "4         7     NaN      NaN   \n",
       "...     ...     ...      ...   \n",
       "7608  10869     NaN      NaN   \n",
       "7609  10870     NaN      NaN   \n",
       "7610  10871     NaN      NaN   \n",
       "7611  10872     NaN      NaN   \n",
       "7612  10873     NaN      NaN   \n",
       "\n",
       "                                                   text  target  \\\n",
       "0     Our Deeds are the Reason of this #earthquake M...       1   \n",
       "1                Forest fire near La Ronge Sask. Canada       1   \n",
       "2     All residents asked to 'shelter in place' are ...       1   \n",
       "3     13,000 people receive #wildfires evacuation or...       1   \n",
       "4     Just got sent this photo from Ruby #Alaska as ...       1   \n",
       "...                                                 ...     ...   \n",
       "7608  Two giant cranes holding a bridge collapse int...       1   \n",
       "7609  @aria_ahrary @TheTawniest The out of control w...       1   \n",
       "7610  M1.94 [01:04 UTC]?5km S of Volcano Hawaii. htt...       1   \n",
       "7611  Police investigating after an e-bike collided ...       1   \n",
       "7612  The Latest: More Homes Razed by Northern Calif...       1   \n",
       "\n",
       "                                             text_clean          hashtags  \\\n",
       "0     Our Deeds are the Reason of this #earthquake M...        earthquake   \n",
       "1                Forest fire near La Ronge Sask. Canada                no   \n",
       "2     All residents asked to 'shelter in place' are ...                no   \n",
       "3     13,000 people receive #wildfires evacuation or...         wildfires   \n",
       "4     Just got sent this photo from Ruby #Alaska as ...  Alaska wildfires   \n",
       "...                                                 ...               ...   \n",
       "7608  Two giant cranes holding a bridge collapse int...                no   \n",
       "7609  @aria_ahrary @TheTawniest The out of control w...                no   \n",
       "7610         M1.94 [01:04 UTC]?5km S of Volcano Hawaii.                no   \n",
       "7611  Police investigating after an e-bike collided ...                no   \n",
       "7612  The Latest: More Homes Razed by Northern Calif...                no   \n",
       "\n",
       "                     mentions                   links  \n",
       "0                          no                      no  \n",
       "1                          no                      no  \n",
       "2                          no                      no  \n",
       "3                          no                      no  \n",
       "4                          no                      no  \n",
       "...                       ...                     ...  \n",
       "7608                       no  http://t.co/STfMbbZFB5  \n",
       "7609  aria_ahrary TheTawniest                      no  \n",
       "7610                       no  http://t.co/zDtoyd8EbJ  \n",
       "7611                       no                      no  \n",
       "7612                       no  http://t.co/YmY4rSkQ3d  \n",
       "\n",
       "[7613 rows x 9 columns]"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def clean_text(text):\n",
    "    text = re.sub(r'https?://\\S+', '', text) # Remove link\n",
    "    text = re.sub(r'\\n',' ', text) # Remove line breaks\n",
    "    text = re.sub('\\s+', ' ', text).strip() # Remove leading, trailing, and extra spaces\n",
    "    return text\n",
    "\n",
    "\n",
    "def find_hashtags(tweet):\n",
    "    return \" \".join([match.group(0)[1:] for match in re.finditer(r\"#\\w+\", tweet)]) or 'no'\n",
    "\n",
    "def find_mentions(tweet):\n",
    "    return \" \".join([match.group(0)[1:] for match in re.finditer(r\"@\\w+\", tweet)]) or 'no'\n",
    "\n",
    "def find_links(tweet):\n",
    "    return \" \".join([match.group(0)[:] for match in re.finditer(r\"https?://\\S+\", tweet)]) or 'no'\n",
    "\n",
    "def process_text(df):\n",
    "    \n",
    "    df['text_clean'] = df['text'].apply(lambda x: clean_text(x))\n",
    "    df['hashtags'] = df['text'].apply(lambda x: find_hashtags(x))\n",
    "    df['mentions'] = df['text'].apply(lambda x: find_mentions(x))\n",
    "    df['links'] = df['text'].apply(lambda x: find_links(x))\n",
    "    # df['hashtags'].fillna(value='no', inplace=True)\n",
    "    # df['mentions'].fillna(value='no', inplace=True)\n",
    "    \n",
    "    return df\n",
    "    \n",
    "train_data = process_text(train_data)\n",
    "test_data = process_text(test_data)\n",
    "\n",
    "train_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(7613, 18) (3263, 17)\n"
     ]
    }
   ],
   "source": [
    "def create_stat(df):\n",
    "    # Tweet length\n",
    "    df['text_len'] = df['text_clean'].apply(len)\n",
    "    # Word count\n",
    "    df['word_count'] = df[\"text_clean\"].apply(lambda x: len(str(x).split()))\n",
    "    # Stopword count\n",
    "    df['stop_word_count'] = df['text_clean'].apply(lambda x: len([w for w in str(x).lower().split() if w in STOPWORDS]))\n",
    "    # Punctuation count\n",
    "    df['punctuation_count'] = df['text_clean'].apply(lambda x: len([c for c in str(x) if c in string.punctuation]))\n",
    "    # Count of hashtags (#)\n",
    "    df['hashtag_count'] = df['hashtags'].apply(lambda x: len(str(x).split()))\n",
    "    # Count of mentions (@)\n",
    "    df['mention_count'] = df['mentions'].apply(lambda x: len(str(x).split()))\n",
    "    # Count of links\n",
    "    df['link_count'] = df['links'].apply(lambda x: len(str(x).split()))\n",
    "    # Count of uppercase letters\n",
    "    df['caps_count'] = df['text_clean'].apply(lambda x: sum(1 for c in str(x) if c.isupper()))\n",
    "    # Ratio of uppercase letters\n",
    "    df['caps_ratio'] = df['caps_count'] / df['text_len']\n",
    "    return df\n",
    "\n",
    "train_data = create_stat(train_data)\n",
    "test_data = create_stat(test_data)\n",
    "\n",
    "print(train_data.shape, test_data.shape)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "stop_word_count     -0.111250\n",
       "mention_count       -0.049654\n",
       "caps_ratio          -0.014970\n",
       "punctuation_count   -0.012535\n",
       "word_count           0.017081\n",
       "link_count           0.020244\n",
       "caps_count           0.027808\n",
       "hashtag_count        0.032853\n",
       "id                   0.060781\n",
       "text_len             0.096435\n",
       "Name: target, dtype: float64"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Correlation\n",
    "# All the features are having lower correlation with the Target Variable\n",
    "\n",
    "train_data.corr()['target'].drop('target').sort_values()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\"\"\"\n",
    "Most Frequent Words amnd Bigrams\n",
    "\"\"\"\n",
    "stopwords = set(STOPWORDS)\n",
    "word_freq = FreqDist(w for w in word_tokenize(' '.join(train_data['text']).lower()) if \n",
    "                     (w not in stopwords) & (w.isalpha()))\n",
    "\n",
    "df_word_freq = pd.DataFrame.from_dict(word_freq, orient='index', columns=['count'])\n",
    "top20w = df_word_freq.sort_values('count',ascending=False).head(20)\n",
    "plt.figure(figsize=(8,6))\n",
    "sns.barplot(top20w['count'], top20w.index)\n",
    "plt.title('Top 20 words')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "\"\"\"\n",
    "Observation\n",
    "Top two words in disaster tweets: 'fire' and 'news'.\n",
    "Words are more specific for real disaster tweets (e.g. 'califonia', 'hiroshima', 'fire', 'police', 'suicide', 'bomb').\n",
    "\"\"\"\n",
    "plt.subplot(121)\n",
    "freq_d = FreqDist(w for w in word_tokenize(' '.join(train_data.loc[train_data.target==1, 'text']).lower()) if \n",
    "                     (w not in stopwords) & (w.isalpha()))\n",
    "df_d = pd.DataFrame.from_dict(freq_d, orient='index', columns=['count'])\n",
    "top20_d = df_d.sort_values('count',ascending=False).head(20)\n",
    "sns.barplot(top20_d['count'], top20_d.index, color='c')\n",
    "plt.title('Top words in disaster tweets')\n",
    "plt.subplot(122)\n",
    "\n",
    "\n",
    "freq_nd = FreqDist(w for w in word_tokenize(' '.join(train_data.loc[train_data.target==0, 'text']).lower()) if \n",
    "                     (w not in stopwords) & (w.isalpha()))\n",
    "df_nd = pd.DataFrame.from_dict(freq_nd, orient='index', columns=['count'])\n",
    "top20_nd = df_nd.sort_values('count',ascending=False).head(20)\n",
    "sns.barplot(top20_nd['count'], top20_nd.index, color='y')\n",
    "plt.title('Top words in non-disaster tweets')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x504 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Bigrams\n",
    "\n",
    "plt.figure(figsize=(16,7))\n",
    "plt.subplot(121)\n",
    "bigram_d = list(bigrams([w for w in word_tokenize(' '.join(train_data.loc[train_data.target==1, 'text']).lower()) if \n",
    "              (w not in stopwords) & (w.isalpha())]))\n",
    "d_fq = FreqDist(bg for bg in bigram_d)\n",
    "bgdf_d = pd.DataFrame.from_dict(d_fq, orient='index', columns=['count'])\n",
    "bgdf_d.index = bgdf_d.index.map(lambda x: ' '.join(x))\n",
    "bgdf_d = bgdf_d.sort_values('count',ascending=False)\n",
    "sns.barplot(bgdf_d.head(20)['count'], bgdf_d.index[:20])\n",
    "plt.title('Top bigrams in disaster tweets')\n",
    "\n",
    "plt.subplot(122)\n",
    "bigram_nd = list(bigrams([w for w in word_tokenize(' '.join(train_data.loc[train_data.target==0, 'text']).lower()) if \n",
    "              (w not in stopwords) & (w.isalpha())]))\n",
    "nd_fq = FreqDist(bg for bg in bigram_nd)\n",
    "bgdf_nd = pd.DataFrame.from_dict(nd_fq, orient='index', columns=['count'])\n",
    "bgdf_nd.index = bgdf_nd.index.map(lambda x: ' '.join(x))\n",
    "bgdf_nd = bgdf_nd.sort_values('count',ascending=False)\n",
    "sns.barplot(bgdf_nd.head(20)['count'], bgdf_nd.index[:20])\n",
    "plt.title('Top bigrams in non-disaster tweets')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Encoding & Vectorizers\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "features = ['keyword']\n",
    "encoder = ce.TargetEncoder(cols=features)\n",
    "encoder.fit(train_data[features],train_data['target'])\n",
    "\n",
    "train_data = train_data.join(encoder.transform(train_data[features]).add_suffix('_target'))\n",
    "test_data = test_data.join(encoder.transform(test_data[features]).add_suffix('_target'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "id                      0\n",
       "keyword                61\n",
       "location             2533\n",
       "text                    0\n",
       "target                  0\n",
       "text_clean              0\n",
       "hashtags                0\n",
       "mentions                0\n",
       "links                   0\n",
       "text_len                0\n",
       "word_count              0\n",
       "stop_word_count         0\n",
       "punctuation_count       0\n",
       "hashtag_count           0\n",
       "mention_count           0\n",
       "link_count              0\n",
       "caps_count              0\n",
       "caps_ratio              0\n",
       "keyword_target          0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ap</th>\n",
       "      <th>arianagrande</th>\n",
       "      <th>change</th>\n",
       "      <th>djicemoon</th>\n",
       "      <th>emmerdale</th>\n",
       "      <th>foxnews</th>\n",
       "      <th>invalid</th>\n",
       "      <th>justinbieber</th>\n",
       "      <th>mikeparractor</th>\n",
       "      <th>no</th>\n",
       "      <th>potus</th>\n",
       "      <th>stretcher</th>\n",
       "      <th>towel</th>\n",
       "      <th>usagov</th>\n",
       "      <th>usatoday</th>\n",
       "      <th>viralspell</th>\n",
       "      <th>youngheroesid</th>\n",
       "      <th>youtube</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\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>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",
       "      <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>0</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>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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\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>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",
       "      <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>0</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>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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\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>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",
       "      <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",
       "      <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>7608</th>\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>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",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7609</th>\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>0</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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7610</th>\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>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",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7611</th>\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>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",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7612</th>\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>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",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>7613 rows × 18 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      ap  arianagrande  change  djicemoon  emmerdale  foxnews  invalid  \\\n",
       "0      0             0       0          0          0        0        0   \n",
       "1      0             0       0          0          0        0        0   \n",
       "2      0             0       0          0          0        0        0   \n",
       "3      0             0       0          0          0        0        0   \n",
       "4      0             0       0          0          0        0        0   \n",
       "...   ..           ...     ...        ...        ...      ...      ...   \n",
       "7608   0             0       0          0          0        0        0   \n",
       "7609   0             0       0          0          0        0        0   \n",
       "7610   0             0       0          0          0        0        0   \n",
       "7611   0             0       0          0          0        0        0   \n",
       "7612   0             0       0          0          0        0        0   \n",
       "\n",
       "      justinbieber  mikeparractor  no  potus  stretcher  towel  usagov  \\\n",
       "0                0              0   1      0          0      0       0   \n",
       "1                0              0   1      0          0      0       0   \n",
       "2                0              0   1      0          0      0       0   \n",
       "3                0              0   1      0          0      0       0   \n",
       "4                0              0   1      0          0      0       0   \n",
       "...            ...            ...  ..    ...        ...    ...     ...   \n",
       "7608             0              0   1      0          0      0       0   \n",
       "7609             0              0   0      0          0      0       0   \n",
       "7610             0              0   1      0          0      0       0   \n",
       "7611             0              0   1      0          0      0       0   \n",
       "7612             0              0   1      0          0      0       0   \n",
       "\n",
       "      usatoday  viralspell  youngheroesid  youtube  \n",
       "0            0           0              0        0  \n",
       "1            0           0              0        0  \n",
       "2            0           0              0        0  \n",
       "3            0           0              0        0  \n",
       "4            0           0              0        0  \n",
       "...        ...         ...            ...      ...  \n",
       "7608         0           0              0        0  \n",
       "7609         0           0              0        0  \n",
       "7610         0           0              0        0  \n",
       "7611         0           0              0        0  \n",
       "7612         0           0              0        0  \n",
       "\n",
       "[7613 rows x 18 columns]"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Mentions\n",
    "vec_men = CountVectorizer(min_df = 5)\n",
    "men_vec = vec_men.fit_transform(train_data['mentions'])\n",
    "X_train_men = pd.DataFrame(men_vec.toarray(), columns=vec_men.get_feature_names())\n",
    "\n",
    "men_vec_test = vec_men.transform(test_data['mentions'])\n",
    "X_test_men = pd.DataFrame(men_vec_test.toarray(), columns=vec_men.get_feature_names())\n",
    "\n",
    "X_train_men"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "scrolled": true
   },
   "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>http://t.co/cybksxhf7d</th>\n",
       "      <th>http://t.co/encmhz6y34</th>\n",
       "      <th>http://t.co/ksawlyux02</th>\n",
       "      <th>http://t.co/q2eblokeve</th>\n",
       "      <th>http://t.co/qew4c5m1xd</th>\n",
       "      <th>http://t.co/vvplfqv58p</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\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",
       "    </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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7608</th>\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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7609</th>\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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7610</th>\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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7611</th>\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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7612</th>\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",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>7613 rows × 6 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      http://t.co/cybksxhf7d  http://t.co/encmhz6y34  http://t.co/ksawlyux02  \\\n",
       "0                          0                       0                       0   \n",
       "1                          0                       0                       0   \n",
       "2                          0                       0                       0   \n",
       "3                          0                       0                       0   \n",
       "4                          0                       0                       0   \n",
       "...                      ...                     ...                     ...   \n",
       "7608                       0                       0                       0   \n",
       "7609                       0                       0                       0   \n",
       "7610                       0                       0                       0   \n",
       "7611                       0                       0                       0   \n",
       "7612                       0                       0                       0   \n",
       "\n",
       "      http://t.co/q2eblokeve  http://t.co/qew4c5m1xd  http://t.co/vvplfqv58p  \n",
       "0                          0                       0                       0  \n",
       "1                          0                       0                       0  \n",
       "2                          0                       0                       0  \n",
       "3                          0                       0                       0  \n",
       "4                          0                       0                       0  \n",
       "...                      ...                     ...                     ...  \n",
       "7608                       0                       0                       0  \n",
       "7609                       0                       0                       0  \n",
       "7610                       0                       0                       0  \n",
       "7611                       0                       0                       0  \n",
       "7612                       0                       0                       0  \n",
       "\n",
       "[7613 rows x 6 columns]"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Links\n",
    "vec_links = CountVectorizer(min_df = 5, analyzer = 'word', token_pattern = r'https?://\\S+') # Only include those >=5 occurrences\n",
    "link_vec = vec_links.fit_transform(train_data['links'])\n",
    "link_vec_test = vec_links.transform(test_data['links'])\n",
    "X_train_link = pd.DataFrame(link_vec.toarray(), columns=vec_links.get_feature_names())\n",
    "X_test_link = pd.DataFrame(link_vec_test.toarray(), columns=vec_links.get_feature_names())\n",
    "\n",
    "X_train_link"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(7613, 6) (7613, 18) (7613, 107)\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>abstorm</th>\n",
       "      <th>africa</th>\n",
       "      <th>afterlife</th>\n",
       "      <th>allah</th>\n",
       "      <th>animalrescue</th>\n",
       "      <th>antioch</th>\n",
       "      <th>armageddon</th>\n",
       "      <th>art</th>\n",
       "      <th>bb17</th>\n",
       "      <th>bbc</th>\n",
       "      <th>...</th>\n",
       "      <th>usa</th>\n",
       "      <th>usgs</th>\n",
       "      <th>wildfire</th>\n",
       "      <th>wildfires</th>\n",
       "      <th>wmata</th>\n",
       "      <th>womens</th>\n",
       "      <th>world</th>\n",
       "      <th>worldnews</th>\n",
       "      <th>wx</th>\n",
       "      <th>yyc</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</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",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</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",
       "      <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",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7608</th>\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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7609</th>\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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7610</th>\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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7611</th>\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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7612</th>\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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>7613 rows × 107 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      abstorm  africa  afterlife  allah  animalrescue  antioch  armageddon  \\\n",
       "0           0       0          0      0             0        0           0   \n",
       "1           0       0          0      0             0        0           0   \n",
       "2           0       0          0      0             0        0           0   \n",
       "3           0       0          0      0             0        0           0   \n",
       "4           0       0          0      0             0        0           0   \n",
       "...       ...     ...        ...    ...           ...      ...         ...   \n",
       "7608        0       0          0      0             0        0           0   \n",
       "7609        0       0          0      0             0        0           0   \n",
       "7610        0       0          0      0             0        0           0   \n",
       "7611        0       0          0      0             0        0           0   \n",
       "7612        0       0          0      0             0        0           0   \n",
       "\n",
       "      art  bb17  bbc  ...  usa  usgs  wildfire  wildfires  wmata  womens  \\\n",
       "0       0     0    0  ...    0     0         0          0      0       0   \n",
       "1       0     0    0  ...    0     0         0          0      0       0   \n",
       "2       0     0    0  ...    0     0         0          0      0       0   \n",
       "3       0     0    0  ...    0     0         0          1      0       0   \n",
       "4       0     0    0  ...    0     0         0          1      0       0   \n",
       "...   ...   ...  ...  ...  ...   ...       ...        ...    ...     ...   \n",
       "7608    0     0    0  ...    0     0         0          0      0       0   \n",
       "7609    0     0    0  ...    0     0         0          0      0       0   \n",
       "7610    0     0    0  ...    0     0         0          0      0       0   \n",
       "7611    0     0    0  ...    0     0         0          0      0       0   \n",
       "7612    0     0    0  ...    0     0         0          0      0       0   \n",
       "\n",
       "      world  worldnews  wx  yyc  \n",
       "0         0          0   0    0  \n",
       "1         0          0   0    0  \n",
       "2         0          0   0    0  \n",
       "3         0          0   0    0  \n",
       "4         0          0   0    0  \n",
       "...     ...        ...  ..  ...  \n",
       "7608      0          0   0    0  \n",
       "7609      0          0   0    0  \n",
       "7610      0          0   0    0  \n",
       "7611      0          0   0    0  \n",
       "7612      0          0   0    0  \n",
       "\n",
       "[7613 rows x 107 columns]"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Hashtags\n",
    "vec_hash = CountVectorizer(min_df = 5)\n",
    "hash_vec = vec_hash.fit_transform(train_data['hashtags'])\n",
    "hash_vec_test = vec_hash.transform(test_data['hashtags'])\n",
    "X_train_hash = pd.DataFrame(hash_vec.toarray(), columns=vec_hash.get_feature_names())\n",
    "X_test_hash = pd.DataFrame(hash_vec_test.toarray(), columns=vec_hash.get_feature_names())\n",
    "print (X_train_link.shape, X_train_men.shape, X_train_hash.shape)\n",
    "X_train_hash\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(7613, 1691)\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>00</th>\n",
       "      <th>01</th>\n",
       "      <th>04</th>\n",
       "      <th>05</th>\n",
       "      <th>06</th>\n",
       "      <th>07</th>\n",
       "      <th>08</th>\n",
       "      <th>08 05</th>\n",
       "      <th>08 06</th>\n",
       "      <th>10</th>\n",
       "      <th>...</th>\n",
       "      <th>û_</th>\n",
       "      <th>ûª</th>\n",
       "      <th>ûªs</th>\n",
       "      <th>ûªt</th>\n",
       "      <th>ûªve</th>\n",
       "      <th>ûï</th>\n",
       "      <th>ûïwhen</th>\n",
       "      <th>ûïwhen saw</th>\n",
       "      <th>ûò</th>\n",
       "      <th>ûó</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</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",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7608</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7609</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7610</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.340738</td>\n",
       "      <td>0.361616</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7611</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7612</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>7613 rows × 1691 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       00        01        04   05   06   07   08  08 05  08 06   10  ...  \\\n",
       "0     0.0  0.000000  0.000000  0.0  0.0  0.0  0.0    0.0    0.0  0.0  ...   \n",
       "1     0.0  0.000000  0.000000  0.0  0.0  0.0  0.0    0.0    0.0  0.0  ...   \n",
       "2     0.0  0.000000  0.000000  0.0  0.0  0.0  0.0    0.0    0.0  0.0  ...   \n",
       "3     0.0  0.000000  0.000000  0.0  0.0  0.0  0.0    0.0    0.0  0.0  ...   \n",
       "4     0.0  0.000000  0.000000  0.0  0.0  0.0  0.0    0.0    0.0  0.0  ...   \n",
       "...   ...       ...       ...  ...  ...  ...  ...    ...    ...  ...  ...   \n",
       "7608  0.0  0.000000  0.000000  0.0  0.0  0.0  0.0    0.0    0.0  0.0  ...   \n",
       "7609  0.0  0.000000  0.000000  0.0  0.0  0.0  0.0    0.0    0.0  0.0  ...   \n",
       "7610  0.0  0.340738  0.361616  0.0  0.0  0.0  0.0    0.0    0.0  0.0  ...   \n",
       "7611  0.0  0.000000  0.000000  0.0  0.0  0.0  0.0    0.0    0.0  0.0  ...   \n",
       "7612  0.0  0.000000  0.000000  0.0  0.0  0.0  0.0    0.0    0.0  0.0  ...   \n",
       "\n",
       "       û_   ûª  ûªs  ûªt  ûªve   ûï  ûïwhen  ûïwhen saw   ûò   ûó  \n",
       "0     0.0  0.0  0.0  0.0   0.0  0.0     0.0         0.0  0.0  0.0  \n",
       "1     0.0  0.0  0.0  0.0   0.0  0.0     0.0         0.0  0.0  0.0  \n",
       "2     0.0  0.0  0.0  0.0   0.0  0.0     0.0         0.0  0.0  0.0  \n",
       "3     0.0  0.0  0.0  0.0   0.0  0.0     0.0         0.0  0.0  0.0  \n",
       "4     0.0  0.0  0.0  0.0   0.0  0.0     0.0         0.0  0.0  0.0  \n",
       "...   ...  ...  ...  ...   ...  ...     ...         ...  ...  ...  \n",
       "7608  0.0  0.0  0.0  0.0   0.0  0.0     0.0         0.0  0.0  0.0  \n",
       "7609  0.0  0.0  0.0  0.0   0.0  0.0     0.0         0.0  0.0  0.0  \n",
       "7610  0.0  0.0  0.0  0.0   0.0  0.0     0.0         0.0  0.0  0.0  \n",
       "7611  0.0  0.0  0.0  0.0   0.0  0.0     0.0         0.0  0.0  0.0  \n",
       "7612  0.0  0.0  0.0  0.0   0.0  0.0     0.0         0.0  0.0  0.0  \n",
       "\n",
       "[7613 rows x 1691 columns]"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\"\"\"\n",
    "TF-IDF Vectoriser\n",
    "\"\"\"\n",
    "\n",
    "vec_text = TfidfVectorizer(min_df = 10, ngram_range = (1,2), stop_words='english') \n",
    "# Only include >=10 occurrences\n",
    "# Have unigrams and bigrams\n",
    "text_vec = vec_text.fit_transform(train_data['text_clean'])\n",
    "text_vec_test = vec_text.transform(test_data['text_clean'])\n",
    "X_train_text = pd.DataFrame(text_vec.toarray(), columns=vec_text.get_feature_names())\n",
    "X_test_text = pd.DataFrame(text_vec_test.toarray(), columns=vec_text.get_feature_names())\n",
    "print (X_train_text.shape)\n",
    "X_train_text"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>abstorm</th>\n",
       "      <th>africa</th>\n",
       "      <th>afterlife</th>\n",
       "      <th>allah</th>\n",
       "      <th>animalrescue</th>\n",
       "      <th>antioch</th>\n",
       "      <th>armageddon</th>\n",
       "      <th>art</th>\n",
       "      <th>bb17</th>\n",
       "      <th>bbc</th>\n",
       "      <th>...</th>\n",
       "      <th>usa</th>\n",
       "      <th>usgs</th>\n",
       "      <th>wildfire</th>\n",
       "      <th>wildfires</th>\n",
       "      <th>wmata</th>\n",
       "      <th>womens</th>\n",
       "      <th>world</th>\n",
       "      <th>worldnews</th>\n",
       "      <th>wx</th>\n",
       "      <th>yyc</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</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",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</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",
       "      <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",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7608</th>\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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7609</th>\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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7610</th>\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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7611</th>\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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7612</th>\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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</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>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>7613 rows × 107 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      abstorm  africa  afterlife  allah  animalrescue  antioch  armageddon  \\\n",
       "0           0       0          0      0             0        0           0   \n",
       "1           0       0          0      0             0        0           0   \n",
       "2           0       0          0      0             0        0           0   \n",
       "3           0       0          0      0             0        0           0   \n",
       "4           0       0          0      0             0        0           0   \n",
       "...       ...     ...        ...    ...           ...      ...         ...   \n",
       "7608        0       0          0      0             0        0           0   \n",
       "7609        0       0          0      0             0        0           0   \n",
       "7610        0       0          0      0             0        0           0   \n",
       "7611        0       0          0      0             0        0           0   \n",
       "7612        0       0          0      0             0        0           0   \n",
       "\n",
       "      art  bb17  bbc  ...  usa  usgs  wildfire  wildfires  wmata  womens  \\\n",
       "0       0     0    0  ...    0     0         0          0      0       0   \n",
       "1       0     0    0  ...    0     0         0          0      0       0   \n",
       "2       0     0    0  ...    0     0         0          0      0       0   \n",
       "3       0     0    0  ...    0     0         0          1      0       0   \n",
       "4       0     0    0  ...    0     0         0          1      0       0   \n",
       "...   ...   ...  ...  ...  ...   ...       ...        ...    ...     ...   \n",
       "7608    0     0    0  ...    0     0         0          0      0       0   \n",
       "7609    0     0    0  ...    0     0         0          0      0       0   \n",
       "7610    0     0    0  ...    0     0         0          0      0       0   \n",
       "7611    0     0    0  ...    0     0         0          0      0       0   \n",
       "7612    0     0    0  ...    0     0         0          0      0       0   \n",
       "\n",
       "      world  worldnews  wx  yyc  \n",
       "0         0          0   0    0  \n",
       "1         0          0   0    0  \n",
       "2         0          0   0    0  \n",
       "3         0          0   0    0  \n",
       "4         0          0   0    0  \n",
       "...     ...        ...  ..  ...  \n",
       "7608      0          0   0    0  \n",
       "7609      0          0   0    0  \n",
       "7610      0          0   0    0  \n",
       "7611      0          0   0    0  \n",
       "7612      0          0   0    0  \n",
       "\n",
       "[7613 rows x 107 columns]"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_hash"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(7613, 1841) (3263, 1840)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Index(['id', 'keyword', 'location', 'text', 'target', 'text_clean', 'hashtags',\n",
       "       'mentions', 'links', 'text_len',\n",
       "       ...\n",
       "       'û_', 'ûª', 'ûªs', 'ûªt', 'ûªve', 'ûï', 'ûïwhen', 'ûïwhen saw', 'ûò',\n",
       "       'ûó'],\n",
       "      dtype='object', length=1841)"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data = train_data.join(X_train_link)\n",
    "train_data = train_data.join(X_train_men)\n",
    "train_data = train_data.join(X_train_hash,rsuffix='_hashtag')\n",
    "train_data = train_data.join(X_train_text,rsuffix='_text')\n",
    "\n",
    "test_data = test_data.join(X_test_link)\n",
    "test_data = test_data.join(X_test_men)\n",
    "test_data = test_data.join(X_test_hash,rsuffix='_hashtag')\n",
    "test_data = test_data.join(X_test_text,rsuffix='_text')\n",
    "print (train_data.shape, test_data.shape)\n",
    "train_data.columns\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'X_test' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-34-c5e9ad29dd00>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"Test Shape\"\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mX_test\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshape\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[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"Train Shape\"\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mX_train\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshape\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[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0mX_test\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcolumns\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mNameError\u001b[0m: name 'X_test' is not defined"
     ]
    }
   ],
   "source": [
    "print(\"Test Shape\",X_test.shape)\n",
    "print(\"Train Shape\",X_train.shape)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "\n",
    "features_to_drop = ['id', 'keyword','location','text','text_clean', 'hashtags', 'mentions','links']\n",
    "scaler = MinMaxScaler()\n",
    "\n",
    "X_train = train_data.drop(columns = features_to_drop + ['target'])\n",
    "X_test = test_data.drop(columns = features_to_drop)\n",
    "\n",
    "y_train = train_data.target\n",
    "\n",
    "lr = LogisticRegression(solver='liblinear', random_state=123) # Other solvers have failure to converge problem\n",
    "\n",
    "pipeline = Pipeline([('scale',scaler), ('lr', lr),])\n",
    "\n",
    "pipeline.fit(X_train, y_train)\n",
    "y_test = pipeline.predict(X_test)\n",
    "\n",
    "submit = submission.copy()\n",
    "submit.target = y_test\n",
    "submit.to_csv('submit_lr.csv',index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training accuracy: 0.8529\n"
     ]
    }
   ],
   "source": [
    "print ('Training accuracy: %.4f' % pipeline.score(X_train, y_train))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training f-1 score: 0.8180\n"
     ]
    }
   ],
   "source": [
    "print ('Training f-1 score: %.4f' % f1_score(y_train, pipeline.predict(X_train)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>3976</td>\n",
       "      <td>366</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>754</td>\n",
       "      <td>2517</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      0     1\n",
       "0  3976   366\n",
       "1   754  2517"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Confusion matrix\n",
    "\n",
    "pd.DataFrame(confusion_matrix(y_train, pipeline.predict(X_train)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model Evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cross validation F-1 score: 0.753\n"
     ]
    }
   ],
   "source": [
    "\"\"\"\n",
    "Cross Validation\n",
    "\"\"\"\n",
    "\n",
    "from sklearn.model_selection import cross_val_score\n",
    "from sklearn.model_selection import ShuffleSplit\n",
    "\n",
    "cv = ShuffleSplit(n_splits=5, test_size=0.2, random_state=123)\n",
    "cv_score = cross_val_score(pipeline, X_train, y_train, cv=cv, scoring='f1')\n",
    "print('Cross validation F-1 score: %.3f' %np.mean(cv_score))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x504 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\"\"\"\n",
    "Feature Selection - Identifying the Top Features\n",
    "\"\"\"\n",
    "\n",
    "plt.figure(figsize=(16,7))\n",
    "s1 = pd.Series(np.transpose(lr.coef_[0]), index=X_train.columns).sort_values(ascending=False)[:20]\n",
    "s2 = pd.Series(np.transpose(lr.coef_[0]), index=X_train.columns).sort_values()[:20]\n",
    "plt.subplot(121)\n",
    "sns.barplot(y=s1.index, x=s1)\n",
    "plt.title('Top positive coefficients')\n",
    "plt.subplot(122)\n",
    "sns.barplot(y=s2.index, x=s2)\n",
    "plt.title('Top negative coefficients')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Findings:\n",
    "\n",
    "'keyword_target' is the top positive coefficient, meaning the keyword column made a good feature\n",
    "hiroshima both as text and hashtag made the top 20 positive coefficients\n",
    "Punctuation count and stop word count are among top 20 negative coefficients\n",
    "None of the bigrams made the top features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Pipeline(memory=None,\n",
       "         steps=[('scale', MinMaxScaler(copy=True, feature_range=(0, 1))),\n",
       "                ('rfecv',\n",
       "                 RFECV(cv=ShuffleSplit(n_splits=5, random_state=123, test_size=0.2, train_size=None),\n",
       "                       estimator=LogisticRegression(C=1.0, class_weight=None,\n",
       "                                                    dual=False,\n",
       "                                                    fit_intercept=True,\n",
       "                                                    intercept_scaling=1,\n",
       "                                                    l1_ratio=None, max_iter=100,\n",
       "                                                    multi_class='auto',\n",
       "                                                    n_jobs=None, penalty='l2',\n",
       "                                                    random_state=777,\n",
       "                                                    solver='liblinear',\n",
       "                                                    tol=0.0001, verbose=0,\n",
       "                                                    warm_start=False),\n",
       "                       min_features_to_select=1, n_jobs=None, scoring='f1',\n",
       "                       step=10, verbose=0))],\n",
       "         verbose=False)"
      ]
     },
     "execution_count": 87,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\"\"\"\n",
    "step parameter in RFECV corresponds to the (integer) number of features to remove at each iteration.\n",
    "\"\"\"\n",
    "steps = 10\n",
    "rfecv = RFECV(estimator=lr, step=steps, cv=cv, scoring='f1')\n",
    "pipeline2 = Pipeline([('scale',scaler), ('rfecv', rfecv)])\n",
    "pipeline2.fit(X_train, y_train)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(7613, 1132)"
      ]
     },
     "execution_count": 101,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\"\"\"\n",
    "RFECV exposes support_ which is another attribute to find out the features which contribute the most to predicting. \n",
    "In order to find out which features are selected we can use the following code.\n",
    "\"\"\"\n",
    "\n",
    "selected_features = X_train.columns[rfecv.support_ == True]\n",
    "X_train2 = X_train[selected_features]\n",
    "X_test2 = X_test[selected_features]\n",
    "\n",
    "X_train2.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmcAAAFzCAYAAAB7Ha4BAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+j8jraAAAgAElEQVR4nOzdd3iV9fnH8fedkAFZzLD33isyFNyzDtziat1aZ7W21U617a9a666txWqt4EZcqDiKIKgIhL1HGAl7ZZB9zvn+/shJDBCSB3IOyYHP67rOlTzfZ5z7RCE393eZcw4RERERqR+i6joAEREREfmBkjMRERGRekTJmYiIiEg9ouRMREREpB5RciYiIiJSjyg5ExEREalHGtR1AKHSvHlz16lTp7oOQ0RERKRG6enpO51zLao6d9QkZ506dWLu3Ll1HYaIiIhIjcxsw8HOqVtTREREpB5RciYiIiJSjyg5ExEREalHlJyJiIiI1CNKzkRERETqESVnIiIiIvWIkjMRERGRekTJmYiIiEg9ouRMREREpB5RciYiIiJSjyg5ExEREalHlJyJiEiF3fklzNu4h73FvroOReSYddRsfC4iIrUTCDiuf2UOCzOzAejQtBEPntOLc/q3ruPIRI4tqpyJiAgA76RnsjAzm7tO7cb9Z/agQbTx8EfLKPUH6jq0eiEQcCzdnEPGjr34A+6g1+zJL2Hn3mK25xVR4tPPTg6dKmciIhEkEHBERVnIn5tdUMJjU1ZyXKcm3HdGD8yMXq2SuenVuUxZspXzB7YJ+XtGipVb8xg/az1fLNvGttxiAOJjoujdOplbT+zKWX1bYmas3JrHPW/OZ8XWvIp7o6OMdk0a0qNlEr8/rw/tmzaqq49RwecPMG9jNv3bptAwNhqAjbsKmPD9Bto2bsi1IzqG5f8x8U7JmYhIBPAHHL/7YAkT52YxtGMTzujTks7NE9iWW8T2vGKaJMTSp3UyvVolkRB36H+1P/nFKrILSnj4guGYlf1iPrVXKh2bNeKVb9cf0eRsU3Yhm7MLOa5T0yP2ngezbmc+l//rO0p8AU7q0YLT+7TEOceKrXlMX7WD2yakc3LPFgzv3Iynv1xFUnwDHjinF41iozEztucWkbEzn2krtnPvWwt469aRRIc48XHOMXvdblZszeOaER1rfP7fPl/FC9PX0jAmmlN7pRIVZXy8aDMAAQdfLNvGk5cPJDU5PqRxindKzkRE6rlSf4D731nIBws2c3bfVmTs3Msjk5dVeW1MtPHk5YMOKZlK37CHCbM2cO2IjvRpk1zRHhVl/HhkJ/44eRmLsrIZ0K5xrT9LTZZvyeXal75nT0Epn997Il1bJIb9PQ8mp7CUG/87hyiDz+898YCq1wP+AP/9dj1PfbGKaSt3cErPFvz10oG0SIo74FmT5mVx39sLeXnmOm4+sUtI4iv2+ZkwayOvfb+BjB35ADRuFMOYQW0Pes/CzGzGfb2Ws/q2pEVSHFOWbKOwxMdNo7twwwmd+Wrldh7+aCnnPDOD+87swSVD2hEfEx2SeMU7c67qfvNIk5aW5ubOnVvXYYiIhFSpP8Cdr8/js6Xb+OXZPbn95G4AbNiVz869xbRMjic1KZ4de4tZtjmXf05bw9LNubx160gGta85mZqxege3jU+nSUIsH981mpRGMfuczy0qZeT//Y8z+7biqSsGheUzlluQmc1PXp5Nw5ho9hb7OKFbM/51bVrF+YISHw1joisqe7VRWOJn+qodTFmyhakrttMiKY5R3ZozsmtzWqfEk9Iwht99sIRZGbuYcONwhndpdtBnbcstYsmmHE7tlXrQ2Jxz3DI+nemrdvDxXaPo3jLpgPNZewpZtiWXNdv3UlTqxxdwJMY14Irj2tM8cd+EL33DHn717iLWbN/LkA6NGTusA/+ekUHAwec/O7HKbslin58LnvuGnMJSPr/vRJLjY/AHHAHniIn+YQj66m15/GLiIhZkZtMsIZarh3fgtN4t6dsmmQbRR2aoemGJn1nrdtEmpSE9WiZW/FzXbM9jUVYOXlOXxPgGjOrW/LCqyeFmZunOubQqzyk5ExGpv8Z/t57ffbCU357bm5tG11xx2bW3mDHPf0OJL8CHd46iVcrBu6Y+XLiZn7+9gK4tEvnvDcNoeZBurIc+XMpr32/g03tG0y01qcpramv1tjwufP4bmiXG8dpNw3l//iae+GIVE28bSVqnpkxftYNbx89lZJdmPHn5IJokxB7ye2zNKeKlmRnMWb+HJZty8AUcTRrFcGqvluzKL+b7jN0Ulvr3ueexS/pzxXEdQvIZd+QVc+ZT02nXpBHjbxxG40Zln2Htjr3c/OrciuoXQJRBg6goSvwBGsVGc/0JnTitd0tWbMljzvrdvL9gE62T4/nzxf05pWcqAB8t3Mxdb8zn+auGcO6AfWfYFpX6eW7qap7/ai0vX5fGqb1aVhurc47v1+3mxa8z+N+K7QAkxjWoSJQCzjGwXWPuOKVbRaVw195i5m7Yw4guzUhpGFPd4w8QCDgyduazKCubqSu2M3XFdgpKyv5bdG6ewPDOTZmzfjdrK/2MvIprEMXJPVswuEMTynPWNo0b0q9NCh2aNqqz8XVKzkREIlBhiZ8TH/+Kzs0SeOvWEZ4rRiu35nHxP76hU/MExt84nKb7JTJ7i308+ulyJszayLDOTXnxx2nV/jJdtzOfc575mqLSAMd3bcYVx7Xn3P6tQ1pF+c17i3l3XhbT7j+FVinxFJT4OPnxabRr0pB7Tu/Bza/OpXVKPFuyi2ieGMtzVw1haMcmnp8fCDgu+9d3LMrKZlD7xqR1asqobs0Z3rlpxeco9vlZviWPXXuL2VNQSvPEWE4OJj6h8vnSrdz+2jxSk+J45srB+AOOW8enExNt3HN6D/q1SaZnqyQaxZZVetbu2MtTX6xi8qItFc9Ijm/ARYPb8ouze5FYqSLkDzjOeGo6sdFRfHL3aLL2FPLApLIKWHmic/Hgtjx5iBXQ7XlFfJ+xm1kZu1i3M5+oYHI2e91u4hpEcd0Jndi4u5ApS7ZQ6nfEx0Rx/oA2jB3WnsHtm1Sb/GzPK+K5/63h/fmbyAuurdcsIZaz+rXizD4tydpTyGdLtzJ3/R4Gd2jM2f1acXzXZsRGe+tq3ZxTyJQlW/l0yZaKyRyVNYyJrpgUUVnn5gm8+9PjPf6EDo+SMxGRCPTC9LU8+ukK3r51JMM6H9rg+K9WbOfWCem0So7n3z9Jo0fLJHz+AFNXbOfhj5axOaeQG0/ozP1n9fQ0pmhzdiHvpmfxTnoWG3cX0KVFAj8/oyfn9GtV68pDYYmfYX/+kjP6tNwncXhz9kYemLSYKINerZJ57abhZO4p4I7X57E5u4hrR3TkZ6d3r6hAVeeN2Rt5cNJi/nbZQC4d2q5W8dbWwsxs7npjPll7CoiOMjo0bcQr1w+rdibnyq15rNu5lz6tU2jftOFBE/X35mdx71sLuWZEB96fvxkzuHRoO5onxpGaFMf5A9uEbAxZxo69/HXKSqYs3UpyfAMuGdqOk3um8tnSrXwwfxP5JX5aJsdxVt9W9G6djHPgcMGvkLm7gPHfbaDUH+CCQW0Y0aUZA9ql0D01KSyTJsoT1IBzbNhVwNLNOazcurfKpWJaJMVx92ndQxrD/pSciYhEmNyiUkY/9hWDOzTmleuHHdYz5m/cwy3j0yks8XNW31Z8tXI7u/NL6Noigb9eOvCQKk/lAgHHF8u38bfPVrJ6+16GdW7Ky9cdt08F51CVJxRv3DyCkV1/GNvl8we48B/fADD+huEVXZk5haU8NmUFb87eSFJ8DA+c04srhx2863Hn3mJOe2I6vVol8eYt3iuQ4ZRXVMojHy0ju7CUv1068ICxfofL5w9w2pPT2bCrgLSOTXh67CDaNQnv8h2ZuwtokRS3T9K3t9jHl8u2MWXJVqat2k5RadXrvV0wsA33ndGDTs0TwhpjfaTkTEQkwjz5+UqenbqGyXeNol/blMN+zubsQm6bkM6a7Xs5rXdLzu3fmlN6tSCuQe2qJ/6A4605mfzugyUM69SU/1x/3GFXZK4cN4vNOYVMu//kAxKnYp+fmKioKqtzK7bm8tCHS5mVsZs3bxnBiIMM2r/v7QV8tHBzWMfM1SfzN5aNqbtyWIcjNoC/OkWlfvYUlGAYZmAABnENog95bNrRRMmZiEgECAQc01fvYPx3G/hq5XZ+1K81z189JCTP9vkDYflFXV71Or13Kv+8Zug+s/682LArn5Men8b9Z/bgzlMPvRupsMTPmU9PJyY6ik/vGX1A0jkxPYv731nInad04/6zeh7y80XCpbrkrO5TahGRY1x2QQkvfp3BKU9M4/r/zGFRVg53ndKNv1zSP2TvEa4KykWD2/HHMX35cvl2bhufTl5R6SHdPzE9iyiDSw5zHFjD2Gj+dGF/Mnbk88K0jH3O/eebddz/zkJO6NaMO0/tdljPF6kL9W/hDxGRY4Rzjpe/Wc/jn62gqDTAcZ2a8PMze3J231bENoicfztfO7ITAA99tIwxz3/DuGvT6JZa9eKxmbsLGD9rQ3AwfBQzV+/gxB4taJ3S8LDf/6QeLbhgYBue/2oNx3VqQmnA8fWqHbw0cx1n9W3Js1cOrnU3rsiRpG5NEZE64A84/jh5Ga98u57Te6dy3xk991mdPxLNytjFHa/No9gX4J/XDGF09xYV59ZsL1sS4tMlW4gyo2OzRgRc2fijP13Yj+O7Na/Ve2/PK+L0J6aTW+SraLtsaDv+cnH/ejHuSmR/GnMmIlKP5Bf7uPetBXy+bBs3j+7Mg+f0Pmo2mt6cXcgNr8whY2c+/7pmKKf0SmXayu3c+fp8zOCq4R247vhOtaqUHcyirGzWbN9LuyaNaN+0YVjeQyRUlJyJiNQTy7fkcsfr81i3M58/nNeH607oXNchhVx2QQnXvjSbFVtzueK49rz+/UZ6tkrmpZ+k0aaxEiYRqD4505gzEZEw2Z5XxAPvLmbdznz6tEmmdXI842dtILlhDK/dNJzju9auK6++atwolgk3Dee6/8xmwqyNnNYrlWevHFwv9zcUqY/0J0VEJAxmZezirjfmk1dUyujuLViYmc3HewoZ3b05T10x6ICNrI82KQ1jmHDjcL5Zs5PTercM+YrvIkczJWciEvGcc2zJKaJVcnydj90q8QX4x7Q1PPu/1XRqnsCEG4fTs1XZwqeFJf4q9/E7WiXENeDMvq3qOgyRiKPkTEQims8f4P53FvL+gs00T4zjtF6pXH5cO4Z2PPhelIGAC0sSt2RTDve/s5AVW/O4aHBb/nhhv322NTqWEjMROXxKzkQkYhX7/Nz1+nw+X7aNn4zsyK78Ej5ZvIX35m/ik3tGH7DWlnOOp79czSvfrue1m4bXalskgFJ/gL9OWcH8jdlk7ilgW24xqUlxvPjjNM7o07JWzxaRY5en5MzM0oDRQBugEFgCfOGc2xPG2ERE9hEIOB6ZvIxlW3Jp3DCGbblFLMzK4ZExfflxcCHUHXnFnP7kdB6ctIi3bhlZUSEr9vl58N3FTJq/iQZRxm/fX8Kknx5fqwraXz5ZwcvfrGNYp6aM7t6Czs0TuGZ4x5BtYi0ix6ZqV+Yzs+vNbB7wINAQWAlsB0YBX5rZf82sQ/jDFJFjzd5iH5MXbcbnD1S0Pf3lKl75dj3FvgAbdhWQXVjKE5cNrEjMAFokxfHbc3szZ/0eXpu9EYB1O/O59qXZTJq/ifvP7MFfLx3Agsxs3p6bWXGfP+D2eS8o2/dxQnA1+/19sGATL3+zjuuO78Tbt43kb5cN5I5TuikxE5Faq6ly1gg4wTlXWNVJMxsEdAc2hjowETk6rN2xl3HTM7j79O60PYQ1rh76cCkT07MY3rkpz101mIWZOTw7dQ2Xp7XjsUsGYHbwitelQ9vx4cLNPPbpCtZsy+P12RuJjY7imbGDGDOoLc453pyTyaNTVnB6n5b8b/k2Hv9sJcW+AGf0bsnoHs35cvl2Pl28hYCDKINTe7VkzKA2NE+Mo9jn54F3F5PWsQm//lHvUPyYREQqaBFaEQkL5xwT07P4w4dLKSjx8+ORHXlkTD9P987buIeL//Eto7o1Z+6G3STHx1BY4qdziwTevnUk8TE1D6zP3F3AmU99TZHPzxVp7bnvzB6kJsVXnF+5NY8fPTuDRjHR5BX7SOvYhI7NEvhy+TZyCktJimvA1SM6cm7/1kxZuoU3Z2eyK7+k4v4WSXF8fNcoUpPjq3p7EZFq1XqHADP7K/AnysabTQEGAPc65ybUcN/ZwDNANPBv59yj+51/CjgleNgISHXONQ6e8wOLg+c2OucuqO69lJyJ1L1in5+FmTksyspm5pqdTFu5gxFdmtIotgFz1u3m+9+cRqPY6gv2gYDjwn98w9acIqbefzKZuwv46YR0cot8fHTXqEOqvi3Kyia2QRS9WlW9Z+XTX67i/fmbuO/Mnpw/oDVmRqk/wOJNOXRLTSQ5/ocuymKfn9Xb9pJbWEpOYSmDOjTW9kAicthCkZwtcM4NMrOLgPOA+4CvnXMDq7knGlgFnAFkAXOAK51zyw5y/V3AYOfcDcHjvc65xKqurYqSM5Ejq9QfoEGUVXQvzly9kwcmLSJrT9koiNYp8VwzoiO3ndSVeRv3cNkL3/HYJf254rgDh6luzy2iNOBokxLPW3MyeWDSYp66YiAXDW4HQFGpn6JSP40bxR65DygiEkah2L6p/LpzgXeccznVjfcIGgascc5lBIN4ExgDVJmcAVcCf/AYj4jUoUnzsnhg0mJSk+IY3b0FxaV+Js3fRJfmCfzz6iEM7dRkny7EtI5N6J6ayOvfbzwgOdu4q4DznptBbpGPRrHRBJwjrWMTLhzUtuKa+JhoT12ZIiJHA6/J2WQzW0FZt+ZPzawFUFTDPW2BzErHWcDwqi40s45AZ2BqpeZ4M5sL+IBHnXPvV3HfLcAtAB06aNKoyJHw/vxN/PydhQzp0ISmCbFMXriZglI/t5/clbtP615lEmVmXDW8Aw9/tIwlm3Iq1hcrKvVz++vpOOCh8/uwYXcBWXsK+eVZPasd8C8icjTzlJw55x4IjjvLcc75zSyfsipYqIwFJjrn/JXaOjrnNplZF2CqmS12zq3dL65xwDgo69YMYTwiUoUPFmzivrcXMKJzM16+7jgaxkbj8wfIL/GT0rD6JSQuHtyOx6as4PXZG/m/i/oD8OePl7NkUy7jrh2qbX5ERIIOZYeAXkAnM6t8z6vVXL8JaF/puF2wrSpjgTsqNzjnNgW/ZpjZNGAwsPbAW0XkSMgpKOWXExeR1rEpL12XVrEVUYPoKFIaVrtkIgApjWI4b0AbJs7NYvW2PFIaxvDl8u3cPLqzEjMRkUq87hAwHugKLADKq1uO6pOzOUB3M+tMWVI2Friqimf3ApoA31VqawIUOOeKzaw5cALwVy+xikh4fLRoM8W+AL87r0+NMy4P5v4zewJly1ys3r6XH/VvxS/P7hXKMEVEIp7Xv2HTgD7uEBZFc875zOxO4DPKltJ42Tm31MweAeY65z4MXjoWeHO/Z/cG/mVmAcp2MXj0YLM8ReTImJieRc+WSfRrW/WyFF60Sonnb5cddJK3iIjgPTlbArQCthzKw51znwCf7Nf2+/2OH6rivm+B/ofyXiISPmu257EgM5vfnttbA/VFRMLMa3LWHFhmZrOB4vLGmhaGFZGjwzvpWURHGWMqLW8hIiLh4TU5eyicQYhI/VLiC7BzbzFtGjfE5w/w3rxNnNKzBS2S4uo6NBGRo57XpTSmm1lL4Lhg02zn3PbwhSUidWXDrnxuf20eSzfnMqxTU4Z3acr2vGIuHdqurkMTETkm1Dz/HTCzy4HZwGXA5cD3ZnZpOAMTkSNvypItnPfsTLL2FPLTk7uyKbuQ56auoUmjGE7t1bKuwxMROSZ47db8DXBcebUsuEPAl8DEcAUmIuFTVFq2Ik7l1fzTN+zhtgnzGNi+Mc9fNZh2TRpx3xk9+GzpVpo0iiW2gad/y4mISC15Tc6i9uvG3IXHqpuI1D83vDKHEl+Ad24bWTH78vXvN5IY14A3bh5esY5ZTHQU5w1oU5ehiogcc7wmZ1PM7DPgjeDxFey3RIaIRIbFWTl8u3YXAHPW72FY56bkFZXyyeItXDi47WEvMCsiIqHhdULAL8zsEspW6gcY55x7L3xhiUi4/OebdSTERhPTIIoXZ2QwrHNTPlq4hcJSP1cc177mB4iISFh5/ieyc+5d4N0wxiIiYbY9r4iPFm3mqmEdSG4Yw9+/WsO6nfm8NTeTni2TGNgupa5DFBE55lU7bszMZga/5plZbqVXnpnlHpkQRSRUXpu1kVK/4yfHd+LakR2JiYriN+8tZmFmNpcf116r/4uI1APVVs6cc6OCX5OOTDgiEi7FPj+vfb+BU3q2oEuLRAAuHNyGt+dmERNtXDRYq/+LiNQHXtc5G++lTUTqrw8XbGbn3hKuP6FzRdtNo7sAcEafljRNiK2r0EREpBKvY876Vj4wswbA0NCHIyLhUOzz8/SXq+nTOpnR3ZtXtPdomcQL1wyhX1uNNRMRqS+qTc7M7EHg10DDSmPMDCgBxoU5NhEJkVe/3cCm7EIeu2TAAePKzu7Xuo6iEhGRqlTbremc+0twvNnjzrnk4CvJOdfMOffgEYpRRGohu6CE56au5qQeLRhVqWomIiL1k9d1zh40syZAdyC+UvvX4QpMRELj71PXkFfs44FzetV1KCIi4oGn5MzMbgLuAdoBC4ARwHfAqeELTURqa1bGLl79bgOXDmlH79bJdR2OiIh44HV/zHuA44ANzrlTgMFAdtiiEpFa8fkDPPH5Sq56cRZtGsdz/1k96zokERHxyOtszSLnXJGZYWZxzrkVZqa/7UXqofxiH9f9ZzZz1u/hsqHteOiCviTEab9MEZFI4fVv7Cwzawy8D3xhZnuADeELS0QOh88f4K435jNvYzZPXzGIC7WwrIhIxPE6IeCi4LcPmdlXQAowJWxRiRwBeUWlJMXH1HUYIeOc46GPljJ1xXb+76L+SsxERCKU1x0CRphZEoBzbjowjbJxZyIRJbeolDdmb+TSf35L/4c+543ZG4/I+67elsf479azbmd+WJ6fX+zjsSkrmTBrI7ee1IWrhncIy/uIiEj4ee3W/CcwpNLx3iraROq1Yp+fC56byfpdBXRLTaRXqyT+75PlnNYrldTk+JofAGzKLmRSehY3je5Cw9hoT/cUlfq5dXw6GcHErFOzRtx5ancuHdrusD9LubyiUl6euZ7/fLuO7IJSLhnSjl+dpSUzREQimdfkzJxzrvzAORcIbuEkEjEmzdvE+l0FPHflYM4b0Jr1uwo46+mveWTyMv5+1RCKSv38cfIyNmcX8syVg0ner8szr6iU6/8zm1Xb9rJq+16eHTsIM6OwxM9v3ltMiT/AmEFtOalHC2Ib/FCU/vvUNWTszOfJyweyt9jHu/M28YuJC2maEMOpvVoe1mdxzvHZ0q38/oOlbM8r5vTeLbnz1G4Mat+4Vj8jERGpe14TrAwzu5uyahnA7UBGeEISCT2fP8A/p61lYLsUzhvQGjOjc/ME7jylG09+sYoTum3kjdkbWZSVQ3SU8eOXZvPqjcMqEjR/wPGzNxewdkc+Ywa14YMFm+nZMpFrR3bipv/OYe6GPTRuGMPkRVto3CiGn57UlRtGdWbtjr28MH0tFw9py8VDyipllw5tx+X/+o67Xp/PpNtPoGerpEP6LIuzcnh26mq+WLaN3q2TGffjNCVlIiJHEatUEDv4RWapwLOULTrrgP8BP3PObQ9veN6lpaW5uXPn1nUYUk/4A47N2YW0b9oIgPfnb+Jnby3gX9cO5ay+rSquK/b5+dEzM1i7I5/EuAY8dcUgnHPc8fo8+rRJ4ZEL+lLqD/Dhws28+t0G/jimL9eM6MjP3lrABws206FpI7bkFPL0FYM5s29LZqzewYRZG5m6YjvdUxNpEB3FttwivrzvJJomxFa879acIi74+0xioqN496fH0yql5m7VSfOyePmbdSzZlEvDmGh+dnp3bhjVmZhor8sViohIfWFm6c65tCrPeUnOIoGSMynnDzjufmM+Hy/ewrUjOvKrc3px0fPfYAZT7jmRqKh9N/5emJnNP6at4Zdn96Jri0QAvli2jdtfS6fU/8Ofj6uHd+DPF/UHysaRXTFuFqu25vHCtUM5qUeLfZ75v+XbeOijpWTuLjzokhaLsrK54l+zaBBtPHhOb8Ye1/6A2Mq99v0GfvPeEnq2TOLqER24cHDbA7pdRUQkchx2cmZmv3TO/dXMnqOsYrYP59zdoQuzdpScCZSNxXrg3cW8NTeTE3u0YMbqHTRpFMvu/BKeGTuIMYO8Ly+xalse63bm0yg2muT4GAa0S8Hsh+SpqNRPbmHpQScTFJX6Wbo5lyEdGu9zX2UZO/by6/cWMytjN8d1asJvz+3DwP26KL/P2MXV//6eE7o15+XrjiP6IAmciIhEjtokZ+c55yab2U+qOu+c+2+IYqw1JWfinOPPHy/n3zPXcfep3bjvzJ7MXreb+99ZSGyDKKbcM5oG9bAL0DnHO+lZPPrpCnbnl3BOv1bcNLoLLRLjKCz1c+WLs2jcKIb3bj+BlIaqlomIHA1qk5yNd85da2b3OOeeCVuEIaDkTKYs2cJtE+Zx3fGd+MP5fSqqVaX+AKX+AI1i6/cE47yiUv49Yx3/npFBfom/oj0pvgHv33FCRZeriIhEvtokZ8uA04FPgZOBffpTnHO7Qxdm7Sg5O7YVlvg5/cnpJMU3YPJdo+plhcyrXXuLmbN+D3uLfeQX+xjZtRk9Wh7ajE4REanfqkvOaiolvEDZzMwuQDr7Jmcu2C5yxJT6A1XOTnz+qzVsyi7k7VtHRnRiBtAsMY6z+7Wq+UIRETkqVftbzDn3rHOuN/Cyc66Lc65zpZcSMzminvh8Jf3+8Bl/+WQ5OQWlFe3rd+Yz7usMLhrclmGdm9ZhhCIiIrVXbeXMzJKdc7nAb8zsgN969albU45uHyzYxHNT19CzZRLjZmTwxuyNnNGnFc45lmzOISbaePAcbVskIiKRr6ZuzdeB8zIxOsQAACAASURBVCjr0nSoW1PqwMLMbH45cRHDOjdlwo3DWbtjL098voqZa3YQEx1FXIMo/u/i/p73xxQREanPtAit1Gvbcn9YSf/DO0fts8q+iIhIpKpuQoCnkdNmdoKZJQS/v8bMnjSzDh7uO9vMVprZGjN7oIrzT5nZguBrlZll73c+2cyyzOzvXuKUo0tRqZ9bxqeTV+Tj3z9JU2ImIiLHBK/T2v4JFJjZQODnwFpgfHU3mFk08DxwDtAHuNLM+lS+xjl3r3NukHNuEPAcMGm/x/wR+NpjjFJPbc8r4o3ZG7n9tXSmrfS2HatzjgcnLWZhZjZPXTGIXq2SwxyliIhI/eB1VU6fc86Z2Rjg7865l8zsxhruGQascc5lAJjZm8AYYNlBrr8S+EP5gZkNBVoCU4Aqy35SPxWV+nlnbiaLN+WweFMuK7bm4hzERkcxY/VOPrl7dMWG5JXtyCtmzvrdlPoDLMjM5r35m7j/zB77bFQuIiJytPOanOWZ2YPANcCJZhYF1LSPTFsgs9JxFjC8qgvNrCPQGZgaPI4Cngi+3+kHewMzuwW4BaBDhxp7WeUIefyzlbw0cx1NE2Lp1zaFc/r14Iw+LUmIbcC5z87g7jfn8/atI/dZryy3qJQLn/+GTdmFFW1jBrXhjlO61cVHEBERqTNek7MrgKuAG51zW4PjzR4PYRxjgYnOufI9a24HPnHOZR1sw2gA59w4YByUTQgIYTxymJxzTFmylVN7pfLST9IO2PD70UsGcMfr83ji81U8UGnpi4c/XMaWnELGXTuUrqmJxEZH0a5Jw4NuGC4iInK08lw5A55xzvnNrAfQC3ijhns2Ae0rHbcLtlVlLHBHpeORwGgzux1IBGLNbK9z7oBJBVK/LNuSy6bsQu4+rVuVidW5A1rzzdoOvDB9LQHnuPu07sxcvYN352Vx16ndOFNdmCIicozzmpx9TVmy1AT4HJhDWTXt6mrumQN0N7POlCVlYymrvu3DzHoBTYDvytucc1dXOn8dkKbE7MhbtjmXP3y4hMcuGUAXj5tuf750G2ZwWu+WB73m9+f1IRBwjPs6gw8WbKLYF6B/2xTuPq17qEIXERGJWF5na5pzrgC4GPiHc+4yoF91NzjnfMCdwGfAcuBt59xSM3vEzC6odOlY4E13tCy4dpRwzvHwR0uZs34Pj0w+2ByOA32xbBtDOzSheWLcQa+Jj4nm0UsGMOn242mRFEeJL8BTVwyscs9MERGRY43XypmZ2UjKKmXlszRr/E3qnPsE+GS/tt/vd/xQDc94BXjFY5wSIl+t3M7363YzqH1jpq3cwdQV2zi118GrYQBZewpYtiXX8zZKQzo04cM7RrG3xEdyfE3zS0RERI4NXksV9wAPAu8Fq19dgK/CF5bUJX/A8dinK+nUrBGv3zycLi0S+OPk5ZT4AtXe98WybQCHNG4sKsqUmImIiFTiKTlzzn3tnLvAOfdY8DjDOXd3eEOTuvLuvCxWbsvjF2f1olFsA35/Xh/W7cznlW/XVXvfF8u20S01kc7NE45QpCIiIkcfT92aZtYC+CXQF6jYXdo5d2qY4pIjzDnH8i15fL5sK69+t4GB7Rvzo/5lFbCTe6Zyaq9UnvvfGq4d0YmGsdEAFPv8XPbCd5T4AvRqlcT363Zzy4ld6vJjiIiIRDyv3ZqvASsoWyj2YWA9ZbMx5ShQUOJj7LhZ/OjZGTzzv9V0bp7Aoxf332cpjJtHdyGv2Mfny7ZWtE1dvp1FWTkkxDVg9rrdRJtxwcA2dfERREREjhpeJwQ0C27ZdI9zbjow3cyUnEWgwhI/76Rncna/VqQmxVPs83Pr+HTmrN/Nb8/tzZhBbWmRdOBMy+Gdm9K2cUMmzdvEmEFtAZiYnkXL5DjevnUk0VFGIOCIitKisSIiIrXhNTkrDX7dYmbnApuBpuEJScJpYnomv/9gKY9+uoJbTuzCqm15zFi9k79eOoDL09of9L6oKGPMoDa8MH0t2/OKAJi2age3nNiF6GBCpsRMRESk9rwmZ38ysxTg58BzQDJwb9iikrCZsXonrZLjGdKxMU9/uRqA357bu9rErNzFQ9ryj2lr+XDBZpwrm9V5yZB24Q5ZRETkmOIpOXPOTQ5+mwOcEr5wJJx8/gDfrd3FuQNa8+glA5i3cQ+bsws5b4C3cWLdUpMY0C6FSfM24Q84BndoTLdUbzsHiIiIiDfVJmdm9hxw0JX7tZxGZFmYlUNesY9R3ZsDZYvADunQ5JCecdHgtjz8UdmOAX+6sNpNIkREROQw1FQ5m3tEopAjYubqnZjBCV2bH/Yzzh/Yhj99vJzoKON8jxU3ERER8a7a5Mw5998jFYiE38w1O+jXJoUmCbGH/YzmiXFcO6IjcTFRpDTSyv4iIiKh5mmdMzP7wswaVzpuYmafhS8sCbW9xT7mb8yu6NKsjYcu6MuD5/QOQVQiIiKyP6+L0LZwzmWXHzjn9gCp4QlJwmHW2l34Ao7R3WqfnImIiEj4eE3O/GbWofzAzDpSzUQBqX9mrtlJfEwUQzsd2gQAERERObK8rnP2G2CmmU0HDBgN3BK2qCTkZqzewbDOzYhrEF3XoYiIiEg1vK5zNsXMhgAjgk0/c87tDF9YEkrrd+azdkc+Vw7rUPPFIiIiUqe8Vs4IJmOTa7xQ6p2J6VlEGZ4XmxUREZG643XMmdQzzjk+WriZtTv2VnudP+B4d14Wo7u3oFVK/BGKTkRERA6X58qZ1C8zVu/krjfmYwY/6teaO07pRp82yQdc9+3anWzJKeI352rpCxERkUjgdZ2z8V7a5MgZ93UGqUlx3H5yV75etYPznpvB16t2HHDdxPQsUhrGcHrvlnUQpYiIiBwqr92afSsfmFk0MDT04YgXSzblMHPNTq4/oTO/OKsXM391Kt1Tk7j3rQVsyy2quC6nsJQpS7ZywcA2xMdolqaIiEgkqDY5M7MHzSwPGGBmucFXHrAd+OCIRCgHeHFGBgmx0Vw1vGz2ZUqjGJ6/ejAFJX7ufmM+Pn8AgI8XbaHYF+CytHZ1Ga6IiIgcgpr21vwL8Bcz+4tz7sEjFJNUI2tPAZMXbeH64zuR0vCHvS27pSbxpwv78fN3FnLL+HRK/QHmb8ymR8tE+rdNqcOIRURE5FB47dacbWYVv+HNrLGZXRimmKQaL89cjwE3jOp8wLlLhrbjymEdmLl6J7vzSzinXyuevHwQZnbkAxUREZHDYs7VvAuTmS1wzg3ar22+c25w2CI7RGlpaW7u3Ll1HUbIlPoDZOzIp2erpIq2lVvzOP+5mZw/sA1PXD7woPcGAo6oKCVkIiIi9ZWZpTvn0qo657VyVtV1WoYjTEp8AX46YR5nPf01L36dAZQlaz9/ZwGJ8Q148Ee9qr1fiZmIiEjk8ppgzTWzJ4Hng8d3AOnhCenYVuILcPtr8/hy+Tb6tknmz58sJzG+Adtzi1myKZcXrhlC88S4ug5TREREwsRrcnYX8DvgLcABX1CWoEkIBQKOO18vS8weGdOXscd14OZX5/Lr9xYTbcaYQW04u1/rug5TREREwsjrxuf5wANmlhD8XsLgy+Xb+HzZNn79o178eGQnAF64ZijXvzKbzN2FPHxB3+ofICIiIhHPU3JmZscD/wYSgQ5mNhC41Tl3eziDO9a88u162qTEc8MJP8zEbBgbzes3jaDEH9BCsiIiIscArxMCngLOAnYBOOcWAieGK6hj0YqtuXy7dhfXjuxEg+h9/7NERZkSMxERkWOE1+QM51zmfk3+EMdyTPvvt+uJj4niymHt6zoUERERqUNeJwRkBrs2nZnFAPcAy8MX1rFlT34Jk+Zt4uIhbWncKLauwxEREZE65LVydhtlszPbApuAQWi2Zsi8OSeTYl+A644/cNV/ERERObbUWDkzs2jgGefc1UcgnmPSW3M2MrJLs312AxAREZFjU42VM+ecH+hoZupvC4Ode4tZv6uA03qn1nUoIiIiUg94HXOWAXxjZh8CFeucOeeerO4mMzsbeAaIBv7tnHt0v/NPAacEDxsBqc65xmbWEXiPsuQxBnjOOfeCx1gjysLMbAAGtm9cx5GIiIhIfeA1OVsbfEUBnvregt2hzwNnAFnAHDP70Dm3rPwa59y9la6/CyjfSH0LMNI5V2xmicCS4L2bPcYbMRZmZhMdZfRtk1zXoYiIiEg94HXMWY/DGHM2DFjjnMsIPudNYAyw7CDXXwn8AcA5V1KpPY5DWPIj0izIyqFHyyQaxWofeREREQnvmLO2QOW10bKCbQcIdmN2BqZWamtvZouCz3isqqqZmd1iZnPNbO6OHTsOMby655xjYWY2g9qn1HUoIiIiUk+EdczZIRgLTAwmguXPzgQGmFkb4H0zm+ic21b5JufcOGAcQFpamgtRLEfMhl0F5BSWMrCdxpuJiIhIGa/dhWuByfww5qz8VZ1NQOXl7tsF26oyFnijqhPBitkSYLTHWCPGwixNBhAREZF9eaqcOeceBggOzsc5t9fDbXOA7mbWmbKkbCxw1f4XmVkvoAnwXaW2dsAu51yhmTUBRlG2v+dRZUFmNg1joumemljXoYiIiEg94alyZmb9zGw+sBRYambpZta3unuccz7gTuAzyrZ6ets5t9TMHjGzCypdOhZ40zlXuVuyN/C9mS0EpgN/c84t9v6xIsOCzGz6t005YKNzEREROXZ5HXM2DrjPOfcVgJmdDLwIHF/dTc65T4BP9mv7/X7HD1Vx3xfAAI+xRaQSX4Clm3P5yciOdR2KiIiI1CNeSzYJ5YkZgHNuGpAQloiOESu35lHiC2i8mYiIiOzD82xNM/sdMD54fA1lMzjlMC0onwygmZoiIiJSidfK2Q1AC2AS8C7QPNgmh+m7tTtpnhhLuyYN6zoUERERqUe8ztbcA9wd5liOGRt25TNlyVZuHt0FM6vrcERERKQe8Tpb8wsza1zpuImZfRa+sI5uL0xfS4PoKG4c1bmuQxEREZF6xmu3ZnPnXHb5QbCSlhqekI5uW3OKmJiexeVp7UhNjq/rcERERKSe8ZqcBcysQ/lBcC/MiNsuqT54cUYGAQe3nti1rkMRERGResjrbM3fADPNbDpglG2ldEvYojpK7c4v4fXvNzJmUBvaN21U1+GIiIhIPeR1QsAUMxsCjAg2/cw5tzN8YR2d3pmbSWGpn5+epKqZiIiIVM1r5YxgMjY5jLEc9aYs3Uq/tsl0b1nTnvEiIiJyrNKmjkfI1pwi5m/M5uy+reo6FBEREanHqk3OzExrPYTI58u2AnB2PyVnIiIicnA1Vc4mApjZ/45ALEe1KUu20rVFAt1S1aUpIiIiB1fTmLMoM/s10MPM7tv/pHPuyfCEdXTZnV/C9+t2c9tJXeo6FBEREannaqqcjQX8lCVxSVW8xIMvl2/DH3Cc0691XYciIiIi9Vy1lTPn3ErgMTNb5Jz79AjFdNT5bMlW2jZuSN82yXUdioiIiNRzXmdrfmtmT5rZ3ODrCTNLCWtkR4kdecXMWL2Ts/u10ibnIiIiUiOvydnLQB5wefCVC/wnXEEdLfKKSrn+ldlERcHlae3rOhwRERGJAF4Xoe3qnLuk0vHDZrYgHAEdLYpK/dzyajortuTx4o/T6NlKQ/RERESkZl4rZ4VmNqr8wMxOAArDE9LR4VfvLuK7jF387bKBnNIrta7DERERkQjhtXJ2G/BqpXFme4CfhCekyLclp5APFmzmtpO6cuHgtnUdjoiIiEQQrxufLwQGmlly8Dg3rFFFuM+WlO0GcFlauzqORERERCKN543PQUmZV58u2UqPlol0bZFY16GIiIhIhNHG5yG2I6+Y2et3c7YWnBUREZHDoOQsxD5fthXn4BxtcC4iIiKHwVO3ppnFA7cDowAHzAT+6ZwrCmNsEWnKkq10bp5ALy2dISIiIofBa+XsVaAv8Bzwd6APMD5cQUWq7IISvlu7S7sBiIiIyGHzOiGgn3OuT6Xjr8xsWTgCimRfLNuGL+DUpSkiIiKHzWvlbJ6ZjSg/MLPhwNzwhBS5Zq7ZScvkOPq31bajIiIicniqrZyZ2WLKxpjFULb5+cbgcUdgRfjDiyzZBaW0So5Xl6aIiIgctpq6Nc87IlEcJQpL/DSKPaSl40RERET2UW0m4ZzbUP69mUUDLWu651iWX+KjdUp8XYchIiIiEczrUhp3AX8AtgGBYLMDBoQprohUoMqZiIiI1JLXTOIeoKdzblc4g4l0+cU+EuKi6zoMERERiWBeZ2tmAjnhDORoUFDip2GMKmciIiJy+LxmEhnANDP7GCgub3TOPRmWqCKQc478ElXOREREpHa8Vs42Al8AsUBSpVe1zOxsM1tpZmvM7IEqzj9lZguCr1Vmlh1sH2Rm35nZUjNbZGZXeP9IdaPYF8A5NOZMREREasVTJuGce/hQHxyc3fk8cAaQBcwxsw+dcxU7Czjn7q10/V3A4OBhAfBj59xqM2sDpJvZZ8657EON40jJL/YBqHImIiIitVJt5czMXjSz/gc5l2BmN5jZ1Qe5fRiwxjmX4ZwrAd4ExlTzdlcCbwA451Y551YHv98MbAdaVP9R6lZBiR9Q5UxERERqp6ZM4nngd8EEbQmwA4gHugPJwMvAawe5ty1lEwnKZQHDq7rQzDoCnYGpVZwbRll36toaYq1T+SXBylmsKmciIiJy+GpahHYBcLmZJQJpQGugEFjunFsZwjjGAhOdc/7KjWbWGhgP/MQ5F9j/JjO7BbgFoEOHDiEM59DlF5eF3lDJmYiIiNSC1zFne4Fph/jsTUD7Ssftgm1VGQvcUbnBzJKBj4HfOOdmHSSuccA4gLS0NHeI8YVUQXnlLE7dmiIiInL4vM7WPBxzgO5m1tnMYilLwD7c/yIz6wU0Ab6r1BYLvAe86pybGMYYQ+aHMWeqnImIiMjhC1ty5pzzAXcCnwHLgbedc0vN7BEzu6DSpWOBN51zlStflwMnAtdVWmpjULhiDYWKypkmBIiIiEgtHFImYWaNnHMFXq93zn0CfLJf2+/3O36oivsmABMOJba6Vj7mrJGW0hAREZFa8FQ5M7PjzWwZsCJ4PNDM/hHWyCJMeeVMS2mIiIhIbXjt1nwKOAvYBeCcW0hZt6MEVczWjFHlTERERA6f5zFnzrnM/Zr8VV54jCoo8dEwJproKKvrUERERCSCee2DyzSz4wFnZjHAPZQN8pegghK/tm4SERGRWvNaObuNsnXI2lK2Vtkg9luX7FhXUOLXeDMRERGptRqzieAG5s845w62h6ZQtvG51jgTERGR2qqxchbcUqljcGFYOYiyypmSMxEREakdr/1wGcA3ZvYhkF/e6Jx7MixRRaD8Eh+J2rpJREREaslrNrE2+IoCksIXTuQqKPaTmhRX12GIiIhIhPO68fnDAGaWGDzeG86gIlFBqU9bN4mIiEited0hoJ+ZzQeWAkvNLN3M+oY3tMhSUOzX1k0iIiJSa16X0hgH3Oec6+ic6wj8HHgxfGFFnvwSn5bSEBERkVrzmpwlOOe+Kj9wzk0DEsISUQTyBxxFpQHN1hQREZFa8zxb08x+B4wPHl9D2QxO4YdNzzXmTERERGrLa+XsBqAFMAl4F2gebBOgsKRsm1GNORMREZHa8jpbcw9wd5hjiVj5weRMlTMRERGpLa+zNb8ws8aVjpuY2WfhCyuy5BeXdWs21JgzERERqSWv3ZrNnXPZ5QfBSlpqeEKKPAWqnImIiEiIeE3OAmbWofzAzDoCLjwhRZ784IQAjTkTERGR2vJa6vkNMNPMpgMGjAZuCVtUEaagWJUzERERCQ2vEwKmmNkQYESw6WfOuZ3hCyuylC+loXXOREREpLa8Tgg4ASh0zk0GGgO/DnZtCpXGnMWpciYiIiK143XM2T+BAjMbCNwHrAVeDVtUESZflTMREREJEa/Jmc8554AxwPPOueeBpPCFFVkKiv1EGcQ18PrjFBEREama1364PDN7kLJtm040syggJnxhRZb8Eh8JsQ0ws7oORURERCKc11LPFUAxcKNzbivQDng8bFFFmIJiv5bREBERkZDwOltzK/BkpeONaMxZhYJSv5bREBERkZDQIKkQKCj2aesmERERCQklZyFQPuZMREREpLaUnIVAQYnGnImIiEhoeF6E1sy+MLNVZpZhZuvMLCPcwUWK/GJVzkRERCQ0vGYULwH3AumAP3zhRKaCEr8WoBUREZGQ8Jqc5TjnPg1rJBGsoMSvrZtEREQkJLxmFF+Z2ePAJMrWOwPAOTcvLFFFmIISzdYUERGR0PCanA0Pfk2r1OaAU0MbTuQp8QUo9TsSlJyJiIhICHhdhPaUcAcSqQoqNj1Xt6aIiIjUntfZmilm9qSZzQ2+njCzFA/3nW1mK81sjZk9UMX5p8xsQfC1ysyyK52bYmbZZjb50D7SkZVfUjY/IkFLaYiIiEgIeF3n7GUgD7g8+MoF/lPdDWYWDTwPnAP0Aa40sz6Vr3HO3eucG+ScGwQ8R9mYtnKPA9d6jK/OFBSrciYiIiKh4zU56+qc+4NzLiP4ehjoUsM9w4A1wetLgDeBMdVcfyXwRvmBc+5/lCWE9VpBsHKmpTREREQkFLwmZ4VmNqr8wMxOAApruKctkFnpOCvYdgAz6wh0BqZ6jKfeyNeYMxEREQkhrxnFT4H/BseZGbAbuC6EcYwFJjrnDmmBWzO7BbgFoEOHDiEMx7uCYo05ExERkdDxOltzATDQzJKDx7kebtsEtK903C7YVpWxwB1eYtkvrnHAOIC0tDR3qPeHgipnIiIiEkrVZhRmdo1zboKZ3bdfOwDOuSeruX0O0N3MOlOWlI0FrqriPXoBTYDvDi30+qFAszVFREQkhGoq9yQEvyZVca7aSpVzzmdmdwKfAdHAy865pWb2CDDXOfdh8NKxwJvOuX2eZ2YzgF5AopllATc65z6rId4jLl+zNUVERCSEqs0onHP/Cn77pXPum8rngpMCquWc+wT4ZL+23+93/NBB7h1d0/Prg9yisuQsUXtrioiISAh4na35nMe2Y05eUSmJcQ2IjrK6DkVERESOAjWNORsJHA+02G/cWTJlXZXHvNxCH8nxqpqJiIhIaNSUVcQCicHrKo87ywUuDVdQkSSvqJSk+Ji6DkNERESOEjWNOZsOTDezV5xzG45QTBElr8hHkipnIiIiEiJes4oCM3sc6AvElzc6504NS1QRJK+4lBaJcXUdhoiIiBwlvE4IeA1YQdkWSw8D6ylbx+yYV1Y5U7emiIiIhIbX5KyZc+4loNQ5N905dwNwzFfNAHILS0luqG5NERERCQ2vWUVp8OsWMzsX2Aw0DU9IkcM5p8qZiIiIhJTX5OxPwU3Pf07Z+mbJwL1hiypCFJUG8AWcJgSIiIhIyHjd+Hxy8Nsc4JTwhRNZ8orKCoqqnImIiEio1LQI7XNUs4emc+7ukEcUQcq3btIitCIiIhIqNU0ImAukU7Z8xhBgdfA1iLIFao9pucHKWbIqZyIiIhIiNS1C+18AM/spMMo55wsevwDMCH949VtesHKmMWciIiISKl6X0mhC2SSAconBtmOaxpyJiIhIqHkt+TwKzDezrwADTgQeCldQkUKVMxEREQk1r7M1/2NmnwLDg02/cs5tDV9YkeGHypmSMxEREQmNars1zaxX8OsQoA2QGXy1CbYd03ILfZhBQqySMxEREQmNmrKKnwM3A09Ucc5xjG/hlFdUSlJcA6KirK5DERERkaNETbM1bw5+1cKzVdDWTSIiIhJqNS1Ce3F1551zk0IbTmTJLfJpvJmIiIiEVE2ZxfnVnHPAMZ2c5RWVagFaERERCamaujWvP1KBRKLcIh9tUuLrOgwRERE5injukzOzc4G+lG3lBIBz7pFwBBUp8opKSW6VVNdhiIiIyFHE0w4Bwe2argDuomwR2suAjmGMKyLkacyZiIiIhJjX7ZuOd879GNjjnHsYGAn0CF9Y9Z9zjr3FSs5EREQktLwmZ4XBrwVm1gYoBVqHJ6TIUFDixx9wWkpDREREQspr2WeymTUGHgfmUTZT88WwRRUBcrV1k4iIiISB1701/xj89l0zmwzEO+dywhdW/Ve+6bmW0hAREZFQ8johYJGZ/drMujrnio/1xAy06bmIiIiEh9cxZ+cDPuBtM5tjZvebWYcwxlXv5QYrZxpzJiIi8v/t3XuUnVV5x/Hvj0kChFtCiVkCiYmaqBRphBi1oA0WYhCbeCmQSAUvFekSEF2gWC/QqFVA1EqpEiEtlptRLmYBSrBEoUrMjZCQQCCGUJNSQJkxCZPLTPL0j3cfODlzzsw5WeedczLn91lr1px3n3e/5zl79pl5Zu/3fbfVU1XJWUQ8HRFXRMRxwAeBY4Cnco2syb08remRMzMzM6ufWm5C+yqye52dAewEPptXUHuDTVsL05oeOTMzM7P6qSo5k/RbYDAwFzgtItblGtVeYPNL05oeOTMzM7P6qTazOCsi1uQayV5m87Yu2vYRQ4e0NToUMzMzG0CqPefMiVmJwtJNkhodipmZmQ0g1V6taSU2b+vylKaZmZnVXa7JmaSpktZIWivpkjLPf1vS8vT1hKSOoufOlvRk+jo7zzj3xOZt3Ry0ry8GMDMzs/qq9ia0p0k6KD3+oqTbJR3bR5024BrgFOAoYKako4r3iYhPR8SEiJgAXA3cnuoeClwKvAWYBFwqaXhtby1fmzxyZmZmZjmoduTsSxGxWdIJwEnA9cD3+qgzCVgbEesiYgdwKzC9l/1nArekx+8C7ouIFyKiHbgPmFplrP1i87ZuDt7fI2dmZmZWX9UmZzvT91OB2RFxNzCkjzpHAL8v2t6QynpI91AbC9xfa91GKVwQYGZmZlZP1SZnGyVdS3YD2nsk7VtD3WrMAH4SETv73LOIpHMkLZG05Pnnn69jOH3btK3Li56bmZlZ3VWbYJ0O3Au8KyI6gEOBi/uosxEYRB0YoQAAEJVJREFUVbR9ZCorZwYvT2lWXTciZkfExIiYOGLEiD7CqZ9du4It2z1yZmZmZvVXbXL2SuDuiHhS0mTgNGBRH3UWA+MkjZU0hCwBm1e6k6TXA8OBh4qK7wWmSBqeLgSYksqawpYd3UR4dQAzMzOrv2qTs9uAnZJeC8wmG9W6ubcKEdENnEeWVD0GzI2IVZJmSZpWtOsM4NaIiKK6LwBfIUvwFgOzUllTeHnRc09rmpmZWX1VO/SzKyK6Jb0fuDoirpb0cF+VIuIe4J6Ssi+XbF9Woe4cYE6V8fWrzdu86LmZmZnlo9qRsy5JM4GzgLtSWctmJu0vZsnZsKEt2wRmZmaWk2qTs48AbwO+FhFPSRoL/Gd+YTW3P23dATg5MzMzs/qrduHz1cBFwEpJRwMbIuLyXCNrYu2d2cjZ8KF93erNzMzMrDZVnXOWrtC8AVgPCBgl6eyIeCC/0JpXe2c2cubkzMzMzOqt2gsCrgKmRMQaAEnjye5LdlxegTWzjs4uhgzah/0G57puvJmZmbWgarOLwYXEDCAinqClLwjYwfChg5HU6FDMzMxsgKl25GyppOuAG9P2mcCSfEJqfu2dXZ7SNDMzs1xUm5ydC3wSuCBtPwj8Wy4R7QU6Onf4Sk0zMzPLRZ/JmaQ24JGIeD3wrfxDan7tnTsYP/KgRodhZmZmA1Cf55xFxE5gjaTR/RDPXqGjs4thntY0MzOzHFQ7rTkcWCVpEfBioTAiplWuMjBFBB1buxjuaU0zMzPLQbXJ2ZdyjWIvsnl7Nzt3hS8IMDMzs1z0mpxJei0wMiJ+VVJ+AvBMnoE1qw6vq2lmZmY56uucs+8Am8qU/yk913K8OoCZmZnlqa/kbGRErCwtTGVjcomoyb2UnB3gkTMzMzOrv76Ss2G9PLd/PQPZW3R0FqY1PXJmZmZm9ddXcrZE0sdLCyX9PbA0n5CaW2HkbNj+HjkzMzOz+uvras0LgTskncnLydhEYAjwvjwDa1btaeTsECdnZmZmloNek7OIeBb4S0knAken4rsj4v7cI2tSHZ07OHi/QQxqq3bNeDMzM7PqVXWfs4hYACzIOZa9QntnF8MP8PlmZmZmlg8P/9QoW/TcyZmZmZnlw8lZjTo6vXSTmZmZ5cfJWY3aO3f4BrRmZmaWGydnNero7PLSTWZmZpYbJ2c12NG9iy3buz1yZmZmZrlxclaDjq2FdTU9cmZmZmb5cHJWAy/dZGZmZnlzclaD9hcLI2dOzszMzCwfTs5q0P7SyJmnNc3MzCwfTs5q0FFY9NzJmZmZmeXEyVkNOrZmI2ee1jQzM7O8ODmrQXvnDoa07cPQIW2NDsXMzMwGKCdnNeh4MbsBraRGh2JmZmYDlJOzGnjpJjMzM8ubk7MaeOkmMzMzy5uTsxp45MzMzMzylmtyJmmqpDWS1kq6pMI+p0taLWmVpJuLyi+X9Gj6OiPPOKvV3tnF8AM8cmZmZmb5GZTXgSW1AdcAJwMbgMWS5kXE6qJ9xgGfB46PiHZJr0jlpwLHAhOAfYFfSvpZRGzKK95qHPeqYfz54Yc0MgQzMzMb4HJLzoBJwNqIWAcg6VZgOrC6aJ+PA9dERDtARDyXyo8CHoiIbqBb0gpgKjA3x3j7dO2HJjby5c3MzKwF5DmteQTw+6LtDams2HhgvKRfS1ooaWoqfwSYKmmopMOAE4FROcZqZmZm1hTyHDmr9vXHAZOBI4EHJL0xIuZLejPwG+B54CFgZ2llSecA5wCMHj26v2I2MzMzy02eI2cb2X2068hUVmwDMC8iuiLiKeAJsmSNiPhaREyIiJMBped2ExGzI2JiREwcMWJELm/CzMzMrD/lmZwtBsZJGitpCDADmFeyz51ko2ak6cvxwDpJbZL+LJUfAxwDzM8xVjMzM7OmkNu0ZkR0SzoPuBdoA+ZExCpJs4AlETEvPTdF0mqyacuLI+KPkvYDHkzLJG0C/i5dHGBmZmY2oCkiGh1DXUycODGWLFnS6DDMzMzM+iRpaUSUvQ2EVwgwMzMzayJOzszMzMyaiJMzMzMzsybi5MzMzMysiTg5MzMzM2siTs7MzMzMmoiTMzMzM7MmMmDucybpeeDpnA5/GPCHnI69t3Kb9OQ26clt0pPbpCe3SU9uk54GWpu8KiLKrj05YJKzPElaUulGca3KbdKT26Qnt0lPbpOe3CY9uU16aqU28bSmmZmZWRNxcmZmZmbWRJycVWd2owNoQm6TntwmPblNenKb9OQ26clt0lPLtInPOTMzMzNrIh45MzMzM2siTs76IGmqpDWS1kq6pNHx9DdJoyQtkLRa0ipJn0rll0naKGl5+np3o2PtT5LWS1qZ3vuSVHaopPskPZm+D290nP1F0uuK+sJySZskXdhq/UTSHEnPSXq0qKxsv1Dmu+l3ywpJxzYu8vxUaJMrJT2e3vcdkoal8jGSthb1l+83LvL8VGiTip8VSZ9P/WSNpHc1Jup8VWiTHxW1x3pJy1P5gO8nntbshaQ24AngZGADsBiYGRGrGxpYP5L0SuCVEbFM0kHAUuC9wOnAloj4ZkMDbBBJ64GJEfGHorIrgBci4hspkR8eEZ9rVIyNkj43G4G3AB+hhfqJpHcAW4AfRsTRqaxsv0h/fM8H3k3WVv8SEW9pVOx5qdAmU4D7I6Jb0uUAqU3GAHcV9huoKrTJZZT5rEg6CrgFmAQcDvwCGB8RO/s16JyVa5OS568C/hQRs1qhn3jkrHeTgLURsS4idgC3AtMbHFO/iohnImJZerwZeAw4orFRNa3pwA3p8Q1kSWwr+mvgdxGR102hm1ZEPAC8UFJcqV9MJ/tDFBGxEBiW/hkaUMq1SUTMj4jutLkQOLLfA2ugCv2kkunArRGxPSKeAtaS/W0aUHprE0kiGxC4pV+DaiAnZ707Avh90fYGWjgxSf+tvAn4bSo6L01LzGmlKbwkgPmSlko6J5WNjIhn0uP/A0Y2JrSGm8Huv0RbuZ9A5X7h3y+ZjwI/K9oeK+lhSb+S9PZGBdUg5T4r7ifwduDZiHiyqGxA9xMnZ1YVSQcCtwEXRsQm4HvAa4AJwDPAVQ0MrxFOiIhjgVOAT6Yh+ZdEdr5Ay50zIGkIMA34cSpq9X6ym1btF5VI+gLQDdyUip4BRkfEm4DPADdLOrhR8fUzf1Yqm8nu//AN+H7i5Kx3G4FRRdtHprKWImkwWWJ2U0TcDhARz0bEzojYBfyAATjM3puI2Ji+PwfcQfb+ny1MS6XvzzUuwoY5BVgWEc+C+0lSqV+09O8XSR8G3gOcmZJW0tTdH9PjpcDvgPENC7If9fJZafV+Mgh4P/CjQlkr9BMnZ71bDIyTNDaNCMwA5jU4pn6V5vqvBx6LiG8VlRefG/M+4NHSugOVpAPSxRFIOgCYQvb+5wFnp93OBn7amAgbarf/cFu5nxSp1C/mAWelqzbfSnay8zPlDjDQSJoKfBaYFhGdReUj0gUlSHo1MA5Y15go+1cvn5V5wAxJ+0oaS9Ymi/o7vgY6CXg8IjYUClqhnwxqdADNLF1JdB5wL9AGzImIVQ0Oq78dD3wIWFm4jBn4R2CmpAlkUzTrgU80JryGGAnckeWtDAJujoifS1oMzJX0MeBpshNYW0ZKVE9m975wRSv1E0m3AJOBwyRtAC4FvkH5fnEP2ZWaa4FOsitbB5wKbfJ5YF/gvvQ5WhgR5wLvAGZJ6gJ2AedGRLUnzu81KrTJ5HKflYhYJWkusJpsCviTA+1KTSjfJhFxPT3PYYUW6Ce+lYaZmZlZE/G0ppmZmVkTcXJmZmZm1kScnJmZmZk1ESdnZmZmZk3EyZmZmZlZE3FyZmYASIq0uHBh+6K0GHM9jv0fkv62Hsfq43VOk/SYpAVlnrtS0ipJV+7BcSekhcqbkqTJku7aw7oXShraX69nZn1zcmZmBduB90s6rNGBFEt3CK/Wx4CPR8SJZZ47BzgmIi7egzAmkN2TrGrp5rJ7w+/YC4GakjMzy9fe8IvDzPpHNzAb+HTpE6UjX5K2pO+T08LDP5W0TtI3JJ0paZGklZJeU3SYkyQtkfSEpPek+m1pRGtxWvD5E0XHfVDSPLKbb5bGMzMd/1FJl6eyLwMnANeXjo6l4xwILJV0RrrD+G3pdRdLOj7tN0nSQ8oWVP6NpNel1UFmAWdIWp7qXybpoqLjPyppTPpaI+mHZHd4HyXp4qL3909p/wMk3S3pkVT3jDLv8QJJq1O9W4vqzUnt+7Ck6WXqld0ntfU30+utkHS+pAuAw4EFhdFGSVNSGyyT9GNl6+oiaaqkxyUtI1tOx8xy4hUCzKzYNcAKSVfUUOcvgDcAL5AtoXJdREyS9CngfLKRGYAxZOsFvoYsGXgtcBbZskVvlrQv8GtJ89P+xwJHR8RTxS8m6XDgcuA4oB2YL+m9ETFL0juBiyJiSXGdiJgmaUtETEjHuBn4dkT8t6TRZKuAvAF4HHh7Wh3kJOCfI+IDKfGbGBHnpfqX9dIe44CzI2KhpClpexIgYJ6kdwAjgP+NiFPT8Q4pc5xLgLERsV3SsFT2BeD+iPhoKlsk6Rcl9Srtc1b6GUxI7+/QiHhB0meAEyPiD2nU9IvASRHxoqTPAZ9J/eEHwDvJVjT4EWaWGydnZvaSiNiURn0uALZWWW1xYU1ISb8DCsnVSqB4enFuWtT5SUnrgNeTrUt6TNGo3CFkycwOYFFpYpa8GfhlRDyfXvMmsuVc7qwyXsjW6ztK2dJBAAenEaJDgBskjSNbRmdwDccseDoiFqbHU9LXw2n7QLL39yBwVRr1uysiHixznBXATZLu5OX3NgWYVjRqtx8wuqRepX1OAr4fEd0AFZa7eStwFFmSDDAEeIjsZ/VURDwJIOlGsmliM8uBkzMzK/UdYBnw70Vl3aTTIJSdRzWk6LntRY93FW3vYvffMaVrxQXZaNL5EXFv8ROSJgMv7ln4VdkHeGtEbCt53X8FFkTE+ySNAX5Zof5L7ZHsV/S4OG4BX4+Ia0sPIOlYsvPYvirpvyJiVskup5IlnX8DfEHSG9PxPhARa0qONbLkNcvtU+Gt7B4WcF9EzCypO6GaymZWHz7nzMx2k0ZU5pKdXF+wnmwaEWAaezaidJqkfZSdh/ZqYA3ZdOI/SBoMIGm8sgXUe7MI+CtJh0lqA2YCv6oxlvlkU66k1y0kH4cAG9PjDxftvxk4qGh7Pdm0ayHJGlvhde4FPlp03tYRkl6RpmY7I+JG4MrCsYri2QcYFRELgM+luA5MxztfKdOS9KYKr1lun/uATyhdYCHp0DLvbSFwfJpyLpy/Np5suneMXj6HcLfkzczqy8mZmZVzFVB81eYPyBKiR4C3sWejWv9Dllj9DDg3jVpdR3bC/zJJjwLX0seIfppCvQRYADwCLI2In9YYywXAxHRi/Grg3FR+BfB1SQ+XxLGAbBp0eTp5/zbgUEmrgPOAJyrEOh+4GXhI0krgJ2SJ0BvJzgVbDlwKfLWkahtwY6rzMPDdiOgAvkKWGK9Ir/2VMi9baZ/ryH4GK9LP8YOpfDbwc0kL0lTxh4FbJK0gTWmmn9U5wN3pgoDnyr1fM6sPRZTONJiZmZlZo3jkzMzMzKyJODkzMzMzayJOzszMzMyaiJMzMzMzsybi5MzMzMysiTg5MzMzM2siTs7MzMzMmoiTMzMzM7Mm8v+j69rTgVrUwgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10,6))\n",
    "plt.xlabel(\"Number of features selected\")\n",
    "plt.ylabel(\"Cross validation score (nb of correct classifications)\")\n",
    "plt.plot(range(1, len(rfecv.grid_scores_) + 1), rfecv.grid_scores_)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best parameter:  {'C': 1.0, 'penalty': 'l2'}\n",
      "F-1 score: 0.773\n"
     ]
    }
   ],
   "source": [
    "\"\"\"\n",
    "Grid Search CV - Hyperparamater Tuning\n",
    "\"\"\"\n",
    "\n",
    "grid={\"C\":np.logspace(-2,2,5), \"penalty\":[\"l1\",\"l2\"]}\n",
    "lr_cv = GridSearchCV(LogisticRegression(solver='liblinear', random_state=20), grid, cv=cv, scoring = 'f1')\n",
    "\n",
    "pipeline_grid = Pipeline([('scale',scaler), ('gridsearch', lr_cv),])\n",
    "\n",
    "pipeline_grid.fit(X_train2, y_train)\n",
    "\n",
    "print(\"Best parameter: \", lr_cv.best_params_)\n",
    "print(\"F-1 score: %.3f\" %lr_cv.best_score_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "y_test2 = pipeline_grid.predict(X_test2)\n",
    "submit2 = submission.copy()\n",
    "submit2.target = y_test2\n",
    "submit2.to_csv('submit_lr2.csv',index=False)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}