robomax-training-notebook.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 85
    },
    "colab_type": "code",
    "id": "H6QHo6Vaie-2",
    "outputId": "77727bea-3e11-48f7-a52a-e3f2386e4761"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Redirecting output to ‘wget-log.10’.\n",
      "\n",
      "Redirecting output to ‘wget-log.11’.\n"
     ]
    }
   ],
   "source": [
    "# Initial project direcctory setup\n",
    "\n",
    "!mkdir -p data\n",
    "!wget -c \"https://rajpurkar.github.io/SQuAD-explorer/dataset/train-v2.0.json\" -O \"data/stanford_train_data.json\"\n",
    "!wget -c \"https://s3.amazonaws.com/dl4j-distribution/GoogleNews-vectors-negative300.bin.gz\"\n",
    "!gzip -df \"GoogleNews-vectors-negative300.bin.gz\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 445
    },
    "colab_type": "code",
    "id": "N_Sl_8YzgxwD",
    "outputId": "14590c95-1e6d-40ec-ec65-b3716be1c0ac"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Collecting tqdm\n",
      "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/79/43/19c9fee28110cd47f73e6bc596394337fe9f3e5825b4de402bbf30b3beb5/tqdm-4.26.0-py2.py3-none-any.whl (43kB)\n",
      "\r",
      "\u001b[K    23% |███████▌                        | 10kB 17.5MB/s eta 0:00:01\r",
      "\u001b[K    46% |███████████████                 | 20kB 2.8MB/s eta 0:00:01\r",
      "\u001b[K    70% |██████████████████████▌         | 30kB 3.2MB/s eta 0:00:01\r",
      "\u001b[K    93% |██████████████████████████████  | 40kB 3.1MB/s eta 0:00:01\r",
      "\u001b[K    100% |████████████████████████████████| 51kB 3.5MB/s \n",
      "\u001b[?25hInstalling collected packages: tqdm\n",
      "Successfully installed tqdm-4.26.0\n",
      "Requirement already satisfied: Whoosh in /usr/local/lib/python3.6/dist-packages (2.7.4)\n",
      "Requirement already satisfied: gensim in /usr/local/lib/python3.6/dist-packages (3.5.0)\n",
      "Requirement already satisfied: six>=1.5.0 in /usr/local/lib/python3.6/dist-packages (from gensim) (1.11.0)\n",
      "Requirement already satisfied: numpy>=1.11.3 in /usr/local/lib/python3.6/dist-packages (from gensim) (1.14.5)\n",
      "Requirement already satisfied: smart-open>=1.2.1 in /usr/local/lib/python3.6/dist-packages (from gensim) (1.6.0)\n",
      "Requirement already satisfied: scipy>=0.18.1 in /usr/local/lib/python3.6/dist-packages (from gensim) (0.19.1)\n",
      "Requirement already satisfied: requests in /usr/local/lib/python3.6/dist-packages (from smart-open>=1.2.1->gensim) (2.18.4)\n",
      "Requirement already satisfied: boto>=2.32 in /usr/local/lib/python3.6/dist-packages (from smart-open>=1.2.1->gensim) (2.49.0)\n",
      "Requirement already satisfied: boto3 in /usr/local/lib/python3.6/dist-packages (from smart-open>=1.2.1->gensim) (1.9.4)\n",
      "Requirement already satisfied: bz2file in /usr/local/lib/python3.6/dist-packages (from smart-open>=1.2.1->gensim) (0.98)\n",
      "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.6/dist-packages (from requests->smart-open>=1.2.1->gensim) (2018.8.24)\n",
      "Requirement already satisfied: urllib3<1.23,>=1.21.1 in /usr/local/lib/python3.6/dist-packages (from requests->smart-open>=1.2.1->gensim) (1.22)\n",
      "Requirement already satisfied: chardet<3.1.0,>=3.0.2 in /usr/local/lib/python3.6/dist-packages (from requests->smart-open>=1.2.1->gensim) (3.0.4)\n",
      "Requirement already satisfied: idna<2.7,>=2.5 in /usr/local/lib/python3.6/dist-packages (from requests->smart-open>=1.2.1->gensim) (2.6)\n",
      "Requirement already satisfied: jmespath<1.0.0,>=0.7.1 in /usr/local/lib/python3.6/dist-packages (from boto3->smart-open>=1.2.1->gensim) (0.9.3)\n",
      "Requirement already satisfied: botocore<1.13.0,>=1.12.4 in /usr/local/lib/python3.6/dist-packages (from boto3->smart-open>=1.2.1->gensim) (1.12.4)\n",
      "Requirement already satisfied: s3transfer<0.2.0,>=0.1.10 in /usr/local/lib/python3.6/dist-packages (from boto3->smart-open>=1.2.1->gensim) (0.1.13)\n",
      "Requirement already satisfied: python-dateutil<3.0.0,>=2.1; python_version >= \"2.7\" in /usr/local/lib/python3.6/dist-packages (from botocore<1.13.0,>=1.12.4->boto3->smart-open>=1.2.1->gensim) (2.5.3)\n",
      "Requirement already satisfied: docutils>=0.10 in /usr/local/lib/python3.6/dist-packages (from botocore<1.13.0,>=1.12.4->boto3->smart-open>=1.2.1->gensim) (0.14)\n"
     ]
    }
   ],
   "source": [
    "# Install the requried libraries \n",
    "\n",
    "!pip install tqdm\n",
    "!pip install Whoosh\n",
    "!pip install gensim"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 153
    },
    "colab_type": "code",
    "id": "fV-NyA9fg0P4",
    "outputId": "91625783-1239-486e-95fd-7abf153ec271"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[nltk_data] Downloading package punkt to /root/nltk_data...\n",
      "[nltk_data]   Package punkt is already up-to-date!\n",
      "[nltk_data] Downloading package averaged_perceptron_tagger to\n",
      "[nltk_data]     /root/nltk_data...\n",
      "[nltk_data]   Package averaged_perceptron_tagger is already up-to-\n",
      "[nltk_data]       date!\n",
      "[nltk_data] Downloading package wordnet to /root/nltk_data...\n",
      "[nltk_data]   Package wordnet is already up-to-date!\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import nltk\n",
    "import json\n",
    "import gensim\n",
    "import unicodedata\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from tqdm import tqdm\n",
    "from whoosh.fields import Schema, ID, KEYWORD, TEXT\n",
    "from whoosh.index import create_in\n",
    "from whoosh.query import Term\n",
    "from whoosh.qparser import QueryParser\n",
    "from whoosh import scoring\n",
    "from whoosh import qparser\n",
    "from IPython.display import display, HTML\n",
    "\n",
    "nltk.download('punkt')\n",
    "nltk.download('averaged_perceptron_tagger')\n",
    "nltk.download('wordnet')\n",
    "tqdm.pandas()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "ZdGifJHrdhax"
   },
   "outputs": [],
   "source": [
    "# Generate dataframe from SQuAD json file\n",
    "\n",
    "squad_data = json.loads(open(\"data/stanford_train_data.json\").read())['data']\n",
    "\n",
    "squad_para_df      = pd.DataFrame(columns=['id', 'title', 'context'])\n",
    "squad_questions_df = pd.DataFrame(columns=['id', 'para_id', 'question'])\n",
    "squad_answers_df   = pd.DataFrame(columns=['question_id', 'answer_start', 'answer_text'])\n",
    "\n",
    "paraIdx = 0\n",
    "for document in tqdm(squad_data):\n",
    "    for paragraph in document['paragraphs']:     \n",
    "        p = unicodedata.normalize('NFD', paragraph['context']).encode('ascii', 'ignore')\n",
    "        squad_para_df = squad_para_df.append({'id': paraIdx, 'title': document['title'], \\\n",
    "                                              'context': p}, ignore_index=True)        \n",
    "        for qas in paragraph['qas']:\n",
    "            q = unicodedata.normalize('NFD', qas['question']).encode('ascii', 'ignore')\n",
    "            squad_questions_df = squad_questions_df.append({'id' : qas['id'], 'para_id': paraIdx, \\\n",
    "                                                            'question': q}, ignore_index=True)\n",
    "            \n",
    "            for answer in qas['answers']:\n",
    "                a = unicodedata.normalize('NFD', answer['text']).encode('ascii', 'ignore')\n",
    "                squad_answers_df = squad_answers_df.append({'question_id': qas['id'], \\\n",
    "                                                            'answer_start': answer['answer_start'], \\\n",
    "                                                            'answer_text': a}, ignore_index=True)\n",
    "        paraIdx += 1\n",
    "\n",
    "squad_para_df.to_csv('data/squad_paragraphs.tsv', sep='\\t')\n",
    "squad_questions_df.to_csv('data/squad_questions.tsv', sep='\\t')\n",
    "squad_answers_df.to_csv('data/squad_answers.tsv', sep='\\t')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "EIt5_9DYdha1",
    "outputId": "cdf91cb8-c020-4cb9-8619-15fbff4d90c1"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 18896/18896 [00:21<00:00, 887.02it/s] \n"
     ]
    }
   ],
   "source": [
    "# Build indexes for the SQuAD dataset\n",
    "\n",
    "!rm -r squad_index\n",
    "\n",
    "SQUAD_INDEX_DIR = \"squad_index\"\n",
    "\n",
    "schema = Schema(content=TEXT, idx=ID(unique=True, stored=True))\n",
    " \n",
    "# Create index dir if it does not exists.\n",
    "if not os.path.exists(SQUAD_INDEX_DIR):\n",
    "    os.mkdir(SQUAD_INDEX_DIR)\n",
    "\n",
    "# Initialize index\n",
    "search_index = create_in(SQUAD_INDEX_DIR, schema)\n",
    "\n",
    "# Fill index with posts from DB\n",
    "writer = search_index.writer()\n",
    "\n",
    "squad_para_df = pd.read_csv('data/squad_paragraphs.tsv', sep='\\t', header=0)\n",
    "\n",
    "for sqIdx, row in tqdm(squad_para_df.iterrows(), total=squad_para_df.shape[0]):\n",
    "    writer.update_document(content=row['context'], idx=str(row['id']))\n",
    "\n",
    "writer.commit()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "iVw-AQPhdhbV",
    "outputId": "e6ecc21d-4d71-40a7-ae78-28990e8bfc35"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['global', 'warming']"
      ]
     },
     "execution_count": 4,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# We identify the main intent of a question through POS tagging wherein which \n",
    "# we currently select the adjectives and the common noun words\n",
    "\n",
    "tags_to_process = ('JJ', 'NN')\n",
    "\n",
    "def fetch_intent_from_question(text):\n",
    "    tokenized = nltk.word_tokenize(text)\n",
    "    tagged = nltk.pos_tag(tokenized)\n",
    "    entities = [word for word, tag in tagged if tag.startswith(tags_to_process)]\n",
    "    \n",
    "    return entities\n",
    "\n",
    "fetch_intent_from_question(\"What is affecting global warming?\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1969
    },
    "colab_type": "code",
    "id": "2d4Jigf1dha4",
    "outputId": "0d7e16c1-a902-4d9d-9046-87875696c744"
   },
   "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>title</th>\n",
       "      <th>context</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>7</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>The library system of the university is divide...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>38</th>\n",
       "      <td>38</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>The University of Notre Dame du Lac (or simply...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>27</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>Since 2005, Notre Dame has been led by John I....</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>The university is the major seat of the Congre...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>6</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>The Joan B. Kroc Institute for International P...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>31</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>The University of Notre Dame has made being a ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>52</th>\n",
       "      <td>52</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>The \"Notre Dame Victory March\" is the fight so...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>17</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>The university is affiliated with the Congrega...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>20</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>One of the main driving forces in the growth o...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>22</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>Holy Cross Father John Francis O\\'Hara was ele...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>9</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>In 2015-2016, Notre Dame ranked 18th overall a...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>47</th>\n",
       "      <td>47</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>On July 1, 2014, the University of Notre Dame ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50</th>\n",
       "      <td>50</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>Football gameday traditions During home games,...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>Architecturally, the school has a Catholic cha...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>29</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>A Science Hall was built in 1883 under the dir...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>19</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>In 1919 Father James Burns became president of...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>48</th>\n",
       "      <td>48</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>The Notre Dame football team has a long histor...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>46</th>\n",
       "      <td>46</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>Notre Dame's conference affiliations for all o...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>8</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>Notre Dame is known for its competitive admiss...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>23</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>The Rev. John J. Cavanaugh, C.S.C. served as p...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>25</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>Hesburgh is also credited with transforming th...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10612</th>\n",
       "      <td>10612</td>\n",
       "      <td>Gothic_architecture</td>\n",
       "      <td>At the Abbey Saint-Denis, Noyon Cathedral, Not...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1469</th>\n",
       "      <td>1469</td>\n",
       "      <td>2008_Summer_Olympics_torch_relay</td>\n",
       "      <td>Reporters Without Borders organised several sy...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49</th>\n",
       "      <td>49</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>George Gipp was the school\\'s legendary footba...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>53</th>\n",
       "      <td>53</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>In the film Knute Rockne, All American, Knute ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>40</th>\n",
       "      <td>40</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>Besides its prominence in sports, Notre Dame i...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39</th>\n",
       "      <td>39</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>Notre Dame rose to national prominence in the ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>21</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>The success of its football team made Notre Da...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14216</th>\n",
       "      <td>14216</td>\n",
       "      <td>Paris</td>\n",
       "      <td>During the 19th century, Paris was the home an...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>14</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>As of 2012[update] research continued in many ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4313</th>\n",
       "      <td>4313</td>\n",
       "      <td>Southampton</td>\n",
       "      <td>Southampton's police service is provided by Ha...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5017</th>\n",
       "      <td>5017</td>\n",
       "      <td>Ministry_of_Defence_(United_Kingdom)</td>\n",
       "      <td>The headquarters of the MoD are in Whitehall a...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8629</th>\n",
       "      <td>8629</td>\n",
       "      <td>Montevideo</td>\n",
       "      <td>Parque Rodo is both a barrio (neighbourhood) o...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14639</th>\n",
       "      <td>14639</td>\n",
       "      <td>Appalachian_Mountains</td>\n",
       "      <td>Mountains of the Long Range in Newfoundland re...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7249</th>\n",
       "      <td>7249</td>\n",
       "      <td>Valencia</td>\n",
       "      <td>In the 15th century the dome was added and the...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5018</th>\n",
       "      <td>5018</td>\n",
       "      <td>Ministry_of_Defence_(United_Kingdom)</td>\n",
       "      <td>Henry VIII's wine cellar at the Palace of Whit...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17250</th>\n",
       "      <td>17250</td>\n",
       "      <td>Georgian_architecture</td>\n",
       "      <td>In towns, which expanded greatly during the pe...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>16</td>\n",
       "      <td>University_of_Notre_Dame</td>\n",
       "      <td>About 80% of undergraduates and 20% of graduat...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8662</th>\n",
       "      <td>8662</td>\n",
       "      <td>Montevideo</td>\n",
       "      <td>Hospital Maciel is one of the oldest hospitals...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2782</th>\n",
       "      <td>2782</td>\n",
       "      <td>Oklahoma_City</td>\n",
       "      <td>The city is home to several colleges and unive...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7469</th>\n",
       "      <td>7469</td>\n",
       "      <td>Order_of_the_British_Empire</td>\n",
       "      <td>Honorary knighthoods are appointed to citizens...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10396</th>\n",
       "      <td>10396</td>\n",
       "      <td>Carnival</td>\n",
       "      <td>Tarragona has one of the region's most complet...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18038</th>\n",
       "      <td>18038</td>\n",
       "      <td>Mosaic</td>\n",
       "      <td>With the building of Christian basilicas in th...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10625</th>\n",
       "      <td>10625</td>\n",
       "      <td>Gothic_architecture</td>\n",
       "      <td>The distinctive characteristic of English cath...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10619</th>\n",
       "      <td>10619</td>\n",
       "      <td>Gothic_architecture</td>\n",
       "      <td>Externally, towers and spires are characterist...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13124</th>\n",
       "      <td>13124</td>\n",
       "      <td>Strasbourg</td>\n",
       "      <td>The German Renaissance has bequeathed the city...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13119</th>\n",
       "      <td>13119</td>\n",
       "      <td>Strasbourg</td>\n",
       "      <td>After the ceasefire following the Fall of Fran...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13703</th>\n",
       "      <td>13703</td>\n",
       "      <td>Annelid</td>\n",
       "      <td>However, leeches and their closest relatives h...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17259</th>\n",
       "      <td>17259</td>\n",
       "      <td>Georgian_architecture</td>\n",
       "      <td>In towns even most better-off people lived in ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12170</th>\n",
       "      <td>12170</td>\n",
       "      <td>Dissolution_of_the_Soviet_Union</td>\n",
       "      <td>Gorbachev refused to make any changes to the s...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2905</th>\n",
       "      <td>2905</td>\n",
       "      <td>Universal_Studios</td>\n",
       "      <td>Despite Laemmle's role as an innovator, he was...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8642</th>\n",
       "      <td>8642</td>\n",
       "      <td>Montevideo</td>\n",
       "      <td>The National History Museum of Montevideo is l...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10508</th>\n",
       "      <td>10508</td>\n",
       "      <td>Yale_University</td>\n",
       "      <td>Alumnus Eero Saarinen, Finnish-American archit...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14597</th>\n",
       "      <td>14597</td>\n",
       "      <td>Neoclassical_architecture</td>\n",
       "      <td>Neoclassicism also influenced city planning; t...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18355</th>\n",
       "      <td>18355</td>\n",
       "      <td>Bras%C3%ADlia</td>\n",
       "      <td>Juscelino Kubitschek, President of Brazil from...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12159</th>\n",
       "      <td>12159</td>\n",
       "      <td>Dissolution_of_the_Soviet_Union</td>\n",
       "      <td>The \"Jeltoqsan\" (Kazakh for \"December\") of 198...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2755</th>\n",
       "      <td>2755</td>\n",
       "      <td>Oklahoma_City</td>\n",
       "      <td>On April 19, 1995, the Alfred P. Murrah Federa...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3995</th>\n",
       "      <td>3995</td>\n",
       "      <td>University_of_Kansas</td>\n",
       "      <td>The University of Kansas School of Architectur...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7476</th>\n",
       "      <td>7476</td>\n",
       "      <td>Order_of_the_British_Empire</td>\n",
       "      <td>Knights Grand Cross and Knights Commander pref...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12635</th>\n",
       "      <td>12635</td>\n",
       "      <td>Cork_(city)</td>\n",
       "      <td>Other notable places include Elizabeth Fort, t...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>100 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          id                                 title  \\\n",
       "7          7              University_of_Notre_Dame   \n",
       "38        38              University_of_Notre_Dame   \n",
       "27        27              University_of_Notre_Dame   \n",
       "2          2              University_of_Notre_Dame   \n",
       "6          6              University_of_Notre_Dame   \n",
       "31        31              University_of_Notre_Dame   \n",
       "52        52              University_of_Notre_Dame   \n",
       "17        17              University_of_Notre_Dame   \n",
       "20        20              University_of_Notre_Dame   \n",
       "22        22              University_of_Notre_Dame   \n",
       "9          9              University_of_Notre_Dame   \n",
       "47        47              University_of_Notre_Dame   \n",
       "50        50              University_of_Notre_Dame   \n",
       "0          0              University_of_Notre_Dame   \n",
       "29        29              University_of_Notre_Dame   \n",
       "19        19              University_of_Notre_Dame   \n",
       "48        48              University_of_Notre_Dame   \n",
       "46        46              University_of_Notre_Dame   \n",
       "8          8              University_of_Notre_Dame   \n",
       "23        23              University_of_Notre_Dame   \n",
       "25        25              University_of_Notre_Dame   \n",
       "10612  10612                   Gothic_architecture   \n",
       "1469    1469      2008_Summer_Olympics_torch_relay   \n",
       "49        49              University_of_Notre_Dame   \n",
       "53        53              University_of_Notre_Dame   \n",
       "40        40              University_of_Notre_Dame   \n",
       "39        39              University_of_Notre_Dame   \n",
       "21        21              University_of_Notre_Dame   \n",
       "14216  14216                                 Paris   \n",
       "14        14              University_of_Notre_Dame   \n",
       "...      ...                                   ...   \n",
       "4313    4313                           Southampton   \n",
       "5017    5017  Ministry_of_Defence_(United_Kingdom)   \n",
       "8629    8629                            Montevideo   \n",
       "14639  14639                 Appalachian_Mountains   \n",
       "7249    7249                              Valencia   \n",
       "5018    5018  Ministry_of_Defence_(United_Kingdom)   \n",
       "17250  17250                 Georgian_architecture   \n",
       "16        16              University_of_Notre_Dame   \n",
       "8662    8662                            Montevideo   \n",
       "2782    2782                         Oklahoma_City   \n",
       "7469    7469           Order_of_the_British_Empire   \n",
       "10396  10396                              Carnival   \n",
       "18038  18038                                Mosaic   \n",
       "10625  10625                   Gothic_architecture   \n",
       "10619  10619                   Gothic_architecture   \n",
       "13124  13124                            Strasbourg   \n",
       "13119  13119                            Strasbourg   \n",
       "13703  13703                               Annelid   \n",
       "17259  17259                 Georgian_architecture   \n",
       "12170  12170       Dissolution_of_the_Soviet_Union   \n",
       "2905    2905                     Universal_Studios   \n",
       "8642    8642                            Montevideo   \n",
       "10508  10508                       Yale_University   \n",
       "14597  14597             Neoclassical_architecture   \n",
       "18355  18355                         Bras%C3%ADlia   \n",
       "12159  12159       Dissolution_of_the_Soviet_Union   \n",
       "2755    2755                         Oklahoma_City   \n",
       "3995    3995                  University_of_Kansas   \n",
       "7476    7476           Order_of_the_British_Empire   \n",
       "12635  12635                           Cork_(city)   \n",
       "\n",
       "                                                 context  \n",
       "7      The library system of the university is divide...  \n",
       "38     The University of Notre Dame du Lac (or simply...  \n",
       "27     Since 2005, Notre Dame has been led by John I....  \n",
       "2      The university is the major seat of the Congre...  \n",
       "6      The Joan B. Kroc Institute for International P...  \n",
       "31     The University of Notre Dame has made being a ...  \n",
       "52     The \"Notre Dame Victory March\" is the fight so...  \n",
       "17     The university is affiliated with the Congrega...  \n",
       "20     One of the main driving forces in the growth o...  \n",
       "22     Holy Cross Father John Francis O\\'Hara was ele...  \n",
       "9      In 2015-2016, Notre Dame ranked 18th overall a...  \n",
       "47     On July 1, 2014, the University of Notre Dame ...  \n",
       "50     Football gameday traditions During home games,...  \n",
       "0      Architecturally, the school has a Catholic cha...  \n",
       "29     A Science Hall was built in 1883 under the dir...  \n",
       "19     In 1919 Father James Burns became president of...  \n",
       "48     The Notre Dame football team has a long histor...  \n",
       "46     Notre Dame's conference affiliations for all o...  \n",
       "8      Notre Dame is known for its competitive admiss...  \n",
       "23     The Rev. John J. Cavanaugh, C.S.C. served as p...  \n",
       "25     Hesburgh is also credited with transforming th...  \n",
       "10612  At the Abbey Saint-Denis, Noyon Cathedral, Not...  \n",
       "1469   Reporters Without Borders organised several sy...  \n",
       "49     George Gipp was the school\\'s legendary footba...  \n",
       "53     In the film Knute Rockne, All American, Knute ...  \n",
       "40     Besides its prominence in sports, Notre Dame i...  \n",
       "39     Notre Dame rose to national prominence in the ...  \n",
       "21     The success of its football team made Notre Da...  \n",
       "14216  During the 19th century, Paris was the home an...  \n",
       "14     As of 2012[update] research continued in many ...  \n",
       "...                                                  ...  \n",
       "4313   Southampton's police service is provided by Ha...  \n",
       "5017   The headquarters of the MoD are in Whitehall a...  \n",
       "8629   Parque Rodo is both a barrio (neighbourhood) o...  \n",
       "14639  Mountains of the Long Range in Newfoundland re...  \n",
       "7249   In the 15th century the dome was added and the...  \n",
       "5018   Henry VIII's wine cellar at the Palace of Whit...  \n",
       "17250  In towns, which expanded greatly during the pe...  \n",
       "16     About 80% of undergraduates and 20% of graduat...  \n",
       "8662   Hospital Maciel is one of the oldest hospitals...  \n",
       "2782   The city is home to several colleges and unive...  \n",
       "7469   Honorary knighthoods are appointed to citizens...  \n",
       "10396  Tarragona has one of the region's most complet...  \n",
       "18038  With the building of Christian basilicas in th...  \n",
       "10625  The distinctive characteristic of English cath...  \n",
       "10619  Externally, towers and spires are characterist...  \n",
       "13124  The German Renaissance has bequeathed the city...  \n",
       "13119  After the ceasefire following the Fall of Fran...  \n",
       "13703  However, leeches and their closest relatives h...  \n",
       "17259  In towns even most better-off people lived in ...  \n",
       "12170  Gorbachev refused to make any changes to the s...  \n",
       "2905   Despite Laemmle's role as an innovator, he was...  \n",
       "8642   The National History Museum of Montevideo is l...  \n",
       "10508  Alumnus Eero Saarinen, Finnish-American archit...  \n",
       "14597  Neoclassicism also influenced city planning; t...  \n",
       "18355  Juscelino Kubitschek, President of Brazil from...  \n",
       "12159  The \"Jeltoqsan\" (Kazakh for \"December\") of 198...  \n",
       "2755   On April 19, 1995, the Alfred P. Murrah Federa...  \n",
       "3995   The University of Kansas School of Architectur...  \n",
       "7476   Knights Grand Cross and Knights Commander pref...  \n",
       "12635  Other notable places include Elizabeth Fort, t...  \n",
       "\n",
       "[100 rows x 3 columns]"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Initialize the Whoosh searcher which looks up the indexes to fetch the documents\n",
    "# related to the provided question\n",
    "\n",
    "searcher = search_index.searcher()\n",
    "og = qparser.OrGroup.factory(0.95)\n",
    "qp = QueryParser(\"content\", schema=search_index.schema, group=og)\n",
    "\n",
    "\n",
    "def find_candidate_answer_set(Q):\n",
    "    intent = fetch_intent_from_question(Q)\n",
    "    query = qp.parse(\" \".join(intent))\n",
    "\n",
    "    result = searcher.search(query, limit=100)\n",
    "    \n",
    "    result_index_list = []\n",
    "    for r in result: result_index_list.append(int(r['idx']))\n",
    "    \n",
    "    return result_index_list\n",
    "\n",
    "Q = \"What is in front of the Notre Dame Main Building?\"\n",
    "squad_para_df = pd.read_csv('data/squad_paragraphs.tsv', sep='\\t', header=0)\n",
    "result_index_list = find_candidate_answer_set(Q)\n",
    "display(squad_para_df.iloc[result_index_list])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 364
    },
    "colab_type": "code",
    "id": "1U-37puIdha8",
    "outputId": "b63753fe-cbae-4968-ef63-94df2b8368ac"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  3%|▎         | 2999/87599 [02:35<1:18:41, 17.92it/s]"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAeoAAAFKCAYAAADScRzUAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1b34aad860>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Generate histogram of the answer index distribution, that is, from the obtained document list\n",
    "# from the searcher, at which index does the sentence containing the answer lies for each question\n",
    "\n",
    "COLLECTOR_LIMIT = 3000\n",
    "\n",
    "collector = []\n",
    "squad_questions_df = pd.read_csv('data/squad_questions.tsv', sep='\\t', header=0)\n",
    "for idx, row in tqdm(squad_questions_df.iterrows(), total=squad_questions_df.shape[0]):\n",
    "    result_index_list = find_candidate_answer_set(row['question'])\n",
    "    if row['para_id'] in result_index_list: collector.append(result_index_list.index(row['para_id']))\n",
    "    else: collector.append(-1)\n",
    "    if len(collector) >= COLLECTOR_LIMIT: break\n",
    "\n",
    "plt.hist(collector, bins=200)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 371
    },
    "colab_type": "code",
    "id": "rnAXkaTwdhbC",
    "outputId": "1e069266-9aca-481f-ba0f-5cf4d8ebfa6a"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f1b347ca630>"
      ]
     },
     "execution_count": 7,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA3YAAAFRCAYAAAA4mz8qAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nhR0REREREZGH48KOiIiIiIjIw/0Pb0FEvR2OraYAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1b3277e0f0>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Build a histogram of the commonly occuring terms in the questions to gain more insights\n",
    "# regarding the type of questions asked\n",
    "\n",
    "from collections import Counter\n",
    "\n",
    "squad_questions_df = pd.read_csv('data/squad_questions.tsv', sep='\\t', header=0)\n",
    "questions = squad_questions_df.question.str.cat(sep=' ')\n",
    "\n",
    "commo_terms_list = Counter(questions.lower().split(' ')).most_common(50)\n",
    "\n",
    "common_terms_df = pd.DataFrame(commo_terms_list, columns=['word', 'frequency'])\n",
    "common_terms_df.plot(kind='bar', x='word', figsize=(15,5))\n",
    "\n",
    "# Commonly occuring question words: what, when, which, how, where, be"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 7225
    },
    "colab_type": "code",
    "id": "trznNpxwdhbI",
    "outputId": "33463a7a-507f-4be6-90e4-904bb3b2ca70"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "100%|██████████| 87599/87599 [00:05<00:00, 15699.02it/s]\u001b[A\n",
      "100%|██████████| 446952/446952 [00:00<00:00, 561163.68it/s]\u001b[A"
     ]
    }
   ],
   "source": [
    "# Formulate dataset for training\n",
    "\n",
    "squad_paragraphs_df = pd.read_csv('data/squad_paragraphs.tsv', sep='\\t', header=0)\n",
    "squad_answers_df = pd.read_csv('data/squad_answers.tsv', sep='\\t', header=0)\n",
    "squad_questions_df = pd.read_csv('data/squad_questions.tsv', sep='\\t', header=0)\n",
    "\n",
    "combi = pd.merge(squad_answers_df, squad_questions_df, left_on='question_id', right_on='id')\n",
    "combi.drop(['question_id', 'id'], axis=1, inplace=True)\n",
    "combi = pd.merge(squad_paragraphs_df, combi, left_on='id', right_on='para_id')\n",
    "combi.drop(['para_id', 'id', 'title'], axis=1, inplace=True)\n",
    "combi['context'] = combi.progress_apply(lambda x: x.context[:x.answer_start] + \" <#ANSWER> \" + x.context[x.answer_start:], axis=1)\n",
    "\n",
    "rows = []\n",
    "combi.progress_apply(lambda row: [rows.append([row['question'], row['answer_text'], sent, \"<#ANSWER>\" in sent]) \n",
    "                         for sent in nltk.sent_tokenize(row['context'])], axis=1)\n",
    "df = pd.DataFrame(rows, columns=['question', 'answer_text', 'sentence', 'target'])\n",
    "\n",
    "feature_df = df\n",
    "feature_df.sentence = feature_df.sentence.progress_apply(lambda x: x.replace('<#ANSWER>', ''))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "PMTONbzYdhbK"
   },
   "outputs": [],
   "source": [
    "# Backup main feature_base \n",
    "\n",
    "feature_df.to_csv('data/question_answers_df.tsv', sep='\\t')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "Ue-fER_8dhbQ"
   },
   "outputs": [],
   "source": [
    "feature_df = pd.read_csv('data/question_answers_df.tsv', sep='\\t', header=0)\n",
    "\n",
    "# To randomly sample a smaller dataset uncomment the following line:\n",
    "# feature_df = feature_df.sample(frac=0.1, random_state=4)\n",
    "\n",
    "feature_df.question.astype('str', inplace=True)\n",
    "feature_df.answer_text.astype('str', inplace=True)\n",
    "feature_df.sentence.astype('str', inplace=True)\n",
    "\n",
    "feature_df.columns.values[0] = 'index_id'\n",
    "feature_df.dropna(axis=0, how='any', inplace=True)\n",
    "\n",
    "print (feature_df.dtypes)\n",
    "display(feature_df.head())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "B7QyrVSmdhbS"
   },
   "outputs": [],
   "source": [
    "# Load Google's pre-trained Word2Vec model.\n",
    "\n",
    "w2v_model = gensim.models.KeyedVectors.load_word2vec_format('GoogleNews-vectors-negative300.bin', binary=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature: Exact Match\n",
    "\n",
    "Calculates the total number of stemmed words which belongs to both the question and the sentence being analyzed."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "rcKtM57gdhbb"
   },
   "outputs": [],
   "source": [
    "from collections import Counter\n",
    "from nltk.stem import PorterStemmer\n",
    "\n",
    "def feature_exact_match(row):\n",
    "    ps = PorterStemmer()\n",
    "    return sum(\n",
    "        (\n",
    "            Counter([ps.stem(x) for x in row['sentence'].lower().split()]) \n",
    "            & \n",
    "            Counter([ps.stem(x) for x in row['question'].lower().split()])\n",
    "        ).values()\n",
    "    )\n",
    "feature_df['exact_match'] = feature_df.progress_apply(feature_exact_match, axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature: Alignment Score\n",
    "\n",
    "We use the Word Mover's Distance (WMD) to calculate how closely a sentence aligns to the given question using Google Word2Vec embedded word vectors. The research paper discussing the approach is present here: http://proceedings.mlr.press/v37/kusnerb15.pdf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "wmcLrushdhbd"
   },
   "outputs": [],
   "source": [
    "import itertools\n",
    "\n",
    "def feature_alignment_score(row):\n",
    "    intents = [x for x in fetch_intent_from_question(row['question']) if x in w2v_model.vocab]\n",
    "    tokens = [x for x in nltk.word_tokenize(row['sentence']) if x in w2v_model.vocab]\n",
    "    total_items = len(intents) * len(tokens)\n",
    "    iter_items = itertools.product(intents, tokens)\n",
    "    alignment_score = 0.0\n",
    "    for a, b in iter_items:\n",
    "        alignment_score += w2v_model.distance(a, b)\n",
    "    if total_items == 0: return float('NaN')\n",
    "    return alignment_score / total_items\n",
    "feature_df['alignment_score'] = feature_df.progress_apply(feature_alignment_score, axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature: Max Alignment Score\n",
    "\n",
    "We modify the above mentioned Word Mover's Distance in a manner to calculate the score of the maximum aligned intent term from the question to any other word in the sentence."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "IoVTQyDRdhbh"
   },
   "outputs": [],
   "source": [
    "def feature_max_alignment(row):\n",
    "    intents = [x for x in fetch_intent_from_question(row['question']) if x in w2v_model.vocab]\n",
    "    tokens =  [x for x in fetch_intent_from_question(row['sentence']) if x in w2v_model.vocab]\n",
    "    if len(intents) * len(tokens) == 0: return float('NaN')\n",
    "    max_alignment_score = 0.0\n",
    "    for intent in intents:\n",
    "        max_alignment_score += np.min([w2v_model.distance(x, intent) for x in tokens])\n",
    "    return max_alignment_score / len(intents)\n",
    "feature_df['max_alignment'] = feature_df.progress_apply(feature_max_alignment, axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature: Number of Intent Words\n",
    "\n",
    "As mentioned previously, we term a word belonging to the Question to be an `intent word` if it's identified to be an Adjective or a Common Noun by a Part-Of-Speech tagger."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "UIQ5oJLXdhbj"
   },
   "outputs": [],
   "source": [
    "def feature_number_of_intent(row):\n",
    "    intents = [x for x in fetch_intent_from_question(row['question']) if x in w2v_model.vocab]\n",
    "    return len(intents)\n",
    "feature_df['number_of_intent'] = feature_df.progress_apply(feature_number_of_intent, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "-XvXKT4pdhbl"
   },
   "outputs": [],
   "source": [
    "from nltk import word_tokenize          \n",
    "from nltk.stem import WordNetLemmatizer \n",
    "\n",
    "class LemmaTokenizer(object):\n",
    "    def __init__(self):\n",
    "        self.wnl = WordNetLemmatizer()\n",
    "    def __call__(self, articles):\n",
    "        return [self.wnl.lemmatize(t) for t in word_tokenize(articles)]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature: Count Vectorizer Match\n",
    "\n",
    "We build count vectors for every question and sentence and calculate the total matching words between each pair after lemmatizing the each of them."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "vvFFDMmUdhbo"
   },
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "\n",
    "ctv = CountVectorizer(tokenizer=LemmaTokenizer(), analyzer='word',token_pattern=r'\\w{1,}', \n",
    "                      ngram_range=(1, 2), stop_words = 'english')\n",
    "ctv.fit(feature_df.sentence.values + feature_df.question.values)\n",
    "\n",
    "sent_ctv = ctv.transform(feature_df.sentence.values)\n",
    "ques_ctv = ctv.transform(feature_df.question.values)\n",
    "\n",
    "ctv_score = []\n",
    "for s, q in tqdm(zip(sent_ctv, ques_ctv), total=sent_ctv.shape[0]):\n",
    "    dot = s.dot(q.T).data\n",
    "    val = dot[0] if len(dot) else 0\n",
    "    ctv_score.append(val)\n",
    "    \n",
    "feature_df['count_vectorizer_match'] = pd.Series(ctv_score).values"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature: Tf-Idf Match Score\n",
    "\n",
    "We calculate the TF-IDF (https://en.wikipedia.org/wiki/Tf%E2%80%93idf) values for the entire vocabulary spanning over question and sentences corpus. We then use these values to vectorize each question and sentence. The Tf-Idf match between any question-sentence pair is then calculated as the dot product between these two vectors."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "lpndpIkMdhbr"
   },
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import TfidfVectorizer\n",
    "\n",
    "tfv = TfidfVectorizer(tokenizer=LemmaTokenizer(), min_df=3,  max_features=None, \n",
    "            strip_accents='unicode', analyzer='word',token_pattern=r'\\w{1,}',\n",
    "            ngram_range=(1, 2), use_idf=1,smooth_idf=1,sublinear_tf=1,\n",
    "            stop_words = 'english')\n",
    "tfv.fit(feature_df.sentence.values + feature_df.question.values)\n",
    "\n",
    "sent_tfv = tfv.transform(feature_df.sentence.values)\n",
    "ques_tfv = tfv.transform(feature_df.question.values)\n",
    "\n",
    "print (sent_tfv.shape)\n",
    "\n",
    "tfv_score = []\n",
    "for s, q in tqdm(zip(sent_tfv, ques_tfv), total=sent_tfv.shape[0]):\n",
    "    dot = s.dot(q.T).data\n",
    "    val = dot[0] if len(dot) else 0.0\n",
    "    tfv_score.append(val)\n",
    "\n",
    "feature_df['tfidf_match'] = pd.Series(tfv_score).values"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature: Question Coverage\n",
    "\n",
    "We define the coverage for a given question-sentence pair as the ratio of the exactly matched words between the question and the sentence over the total number of words in the question."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 5236
    },
    "colab_type": "code",
    "id": "435IE1-Cdhb1",
    "outputId": "4526403a-396f-43fe-9bce-f249bdf99b20"
   },
   "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>index_id</th>\n",
       "      <th>question</th>\n",
       "      <th>answer_text</th>\n",
       "      <th>sentence</th>\n",
       "      <th>target</th>\n",
       "      <th>exact_match</th>\n",
       "      <th>alignment_score</th>\n",
       "      <th>max_alignment</th>\n",
       "      <th>number_of_intent</th>\n",
       "      <th>count_vectorizer_match</th>\n",
       "      <th>tfidf_match</th>\n",
       "      <th>question_words</th>\n",
       "      <th>coverage</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>68873</th>\n",
       "      <td>68873</td>\n",
       "      <td>Where is the theater District located?</td>\n",
       "      <td>south of Boston Common</td>\n",
       "      <td>Other performing-arts organizations located in...</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>0.881072</td>\n",
       "      <td>5.594476e-01</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0.025818</td>\n",
       "      <td>6</td>\n",
       "      <td>0.166667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>195732</th>\n",
       "      <td>195732</td>\n",
       "      <td>Where can the public view the Royal Collection?</td>\n",
       "      <td>at the Queen's Gallery</td>\n",
       "      <td>It is not the monarch's personal property, unl...</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>0.879658</td>\n",
       "      <td>7.210963e-01</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>8</td>\n",
       "      <td>0.125000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>301126</th>\n",
       "      <td>301126</td>\n",
       "      <td>How many players can an NFL team have in the o...</td>\n",
       "      <td>90</td>\n",
       "      <td>Split-squad games, a practice common in baseba...</td>\n",
       "      <td>False</td>\n",
       "      <td>3</td>\n",
       "      <td>0.810726</td>\n",
       "      <td>4.730812e-01</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>0.100368</td>\n",
       "      <td>11</td>\n",
       "      <td>0.272727</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>318729</th>\n",
       "      <td>318729</td>\n",
       "      <td>What document declares supreme power for the p...</td>\n",
       "      <td>the Constitution of Estonia</td>\n",
       "      <td>The people exercise their supreme power of the...</td>\n",
       "      <td>False</td>\n",
       "      <td>3</td>\n",
       "      <td>0.828007</td>\n",
       "      <td>2.086947e-01</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>0.354543</td>\n",
       "      <td>8</td>\n",
       "      <td>0.375000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>193367</th>\n",
       "      <td>193367</td>\n",
       "      <td>What value does Deloitte place on Arsenal?</td>\n",
       "      <td>331.3m</td>\n",
       "      <td>3m (435.5m), ranking Arsenal seventh among wo...</td>\n",
       "      <td>True</td>\n",
       "      <td>0</td>\n",
       "      <td>0.855577</td>\n",
       "      <td>6.227328e-01</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>0.114317</td>\n",
       "      <td>7</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20510</th>\n",
       "      <td>20510</td>\n",
       "      <td>Which city in North America held the most slav...</td>\n",
       "      <td>Charleston, South Carolina</td>\n",
       "      <td>It also became a center of slavery, with 42% o...</td>\n",
       "      <td>True</td>\n",
       "      <td>2</td>\n",
       "      <td>0.865245</td>\n",
       "      <td>3.315323e-01</td>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "      <td>0.081870</td>\n",
       "      <td>12</td>\n",
       "      <td>0.166667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>320616</th>\n",
       "      <td>320616</td>\n",
       "      <td>Compared to the rest of the US, does Alaska ha...</td>\n",
       "      <td>few</td>\n",
       "      <td>The western part of Alaska has no road system ...</td>\n",
       "      <td>False</td>\n",
       "      <td>7</td>\n",
       "      <td>0.850673</td>\n",
       "      <td>4.609409e-01</td>\n",
       "      <td>7</td>\n",
       "      <td>7</td>\n",
       "      <td>0.470676</td>\n",
       "      <td>15</td>\n",
       "      <td>0.466667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>56702</th>\n",
       "      <td>56702</td>\n",
       "      <td>From what passage is cited as being one of the...</td>\n",
       "      <td>Nag Hammadi's Apocalypse of Peter</td>\n",
       "      <td>However, his accusers were excommunicated both...</td>\n",
       "      <td>False</td>\n",
       "      <td>2</td>\n",
       "      <td>0.908814</td>\n",
       "      <td>6.398091e-01</td>\n",
       "      <td>6</td>\n",
       "      <td>1</td>\n",
       "      <td>0.078640</td>\n",
       "      <td>18</td>\n",
       "      <td>0.111111</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>343785</th>\n",
       "      <td>343785</td>\n",
       "      <td>What religion was at the center of education i...</td>\n",
       "      <td>Islam</td>\n",
       "      <td>Wiesner-Hanks also says that Protestants also ...</td>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "      <td>0.884671</td>\n",
       "      <td>7.056823e-01</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>12</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>129620</th>\n",
       "      <td>129620</td>\n",
       "      <td>What type of carrier is the Charles de Gaulle?</td>\n",
       "      <td>CATOBAR</td>\n",
       "      <td>1 CATOBAR carrier: Charles de Gaulle is a 42,0...</td>\n",
       "      <td>False</td>\n",
       "      <td>5</td>\n",
       "      <td>0.853132</td>\n",
       "      <td>2.051662e-01</td>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>0.384170</td>\n",
       "      <td>9</td>\n",
       "      <td>0.555556</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>203405</th>\n",
       "      <td>203405</td>\n",
       "      <td>The Ottomans controlled mainland Greece in wha...</td>\n",
       "      <td>1460</td>\n",
       "      <td>With the Turkish conquest, many Byzantine Gree...</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>0.889290</td>\n",
       "      <td>6.251095e-01</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>8</td>\n",
       "      <td>0.125000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>306098</th>\n",
       "      <td>306098</td>\n",
       "      <td>Who was nominated to the presidency of the Off...</td>\n",
       "      <td>Sirri Amer</td>\n",
       "      <td>They selected Muhammad Naguib, a popular gener...</td>\n",
       "      <td>False</td>\n",
       "      <td>4</td>\n",
       "      <td>0.943691</td>\n",
       "      <td>8.358789e-01</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0.006483</td>\n",
       "      <td>10</td>\n",
       "      <td>0.400000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>346900</th>\n",
       "      <td>346900</td>\n",
       "      <td>Under a Muslim Nawab there was a princely stat...</td>\n",
       "      <td>Tonk</td>\n",
       "      <td>They also resisted Mughal incursions into Indi...</td>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "      <td>0.888309</td>\n",
       "      <td>6.904991e-01</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>13</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>200448</th>\n",
       "      <td>200448</td>\n",
       "      <td>Why did Kim's war plan need revising?</td>\n",
       "      <td>South Korean agents had learned about the plans</td>\n",
       "      <td>On 7 June 1950, Kim Il-sung called for a Korea...</td>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "      <td>0.907051</td>\n",
       "      <td>5.310434e-01</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0.071265</td>\n",
       "      <td>7</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>216455</th>\n",
       "      <td>216455</td>\n",
       "      <td>What free service did Marshall's provide custo...</td>\n",
       "      <td>personal shopper</td>\n",
       "      <td>It grew to become a major chain before convert...</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>0.920871</td>\n",
       "      <td>8.219873e-01</td>\n",
       "      <td>7</td>\n",
       "      <td>2</td>\n",
       "      <td>0.008248</td>\n",
       "      <td>13</td>\n",
       "      <td>0.076923</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>148679</th>\n",
       "      <td>148679</td>\n",
       "      <td>What gun did the Royal Navy start using?</td>\n",
       "      <td>4.5-inch (114 mm) gun in a twin turret</td>\n",
       "      <td>At the same time the Royal Navy adopted a new ...</td>\n",
       "      <td>True</td>\n",
       "      <td>4</td>\n",
       "      <td>0.880607</td>\n",
       "      <td>1.647606e-01</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>0.183972</td>\n",
       "      <td>8</td>\n",
       "      <td>0.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>440677</th>\n",
       "      <td>440677</td>\n",
       "      <td>In which period did North and South america join?</td>\n",
       "      <td>Neogene</td>\n",
       "      <td>Supercontinent Laurasia had not yet separated ...</td>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "      <td>0.902033</td>\n",
       "      <td>8.490692e-01</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>9</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>364615</th>\n",
       "      <td>364615</td>\n",
       "      <td>How does Texas handle people who don't speak E...</td>\n",
       "      <td>. The Government of Texas, through Section 205...</td>\n",
       "      <td>The Government of Texas, through Section 2054....</td>\n",
       "      <td>False</td>\n",
       "      <td>2</td>\n",
       "      <td>0.888766</td>\n",
       "      <td>2.415099e-01</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>0.097705</td>\n",
       "      <td>13</td>\n",
       "      <td>0.153846</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>305686</th>\n",
       "      <td>305686</td>\n",
       "      <td>What organization did Nasser create with Syria?</td>\n",
       "      <td>United Arab Republic</td>\n",
       "      <td>In 1962, Nasser began a series of major social...</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>0.879276</td>\n",
       "      <td>5.257678e-01</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>0.065970</td>\n",
       "      <td>7</td>\n",
       "      <td>0.142857</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>173707</th>\n",
       "      <td>173707</td>\n",
       "      <td>What tax collectors did the Umayyads often lea...</td>\n",
       "      <td>Roman</td>\n",
       "      <td>These new converts continues to pay the same t...</td>\n",
       "      <td>False</td>\n",
       "      <td>2</td>\n",
       "      <td>0.904721</td>\n",
       "      <td>7.470397e-01</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>0.090524</td>\n",
       "      <td>14</td>\n",
       "      <td>0.142857</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>324241</th>\n",
       "      <td>324241</td>\n",
       "      <td>What connects the parts inside a female insect...</td>\n",
       "      <td>ducts</td>\n",
       "      <td>Accessory glands or glandular parts of the ovi...</td>\n",
       "      <td>False</td>\n",
       "      <td>3</td>\n",
       "      <td>0.857441</td>\n",
       "      <td>5.160477e-01</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>10</td>\n",
       "      <td>0.300000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>335020</th>\n",
       "      <td>335020</td>\n",
       "      <td>Who is a professor of civil and environmental ...</td>\n",
       "      <td>Mark Z. Jacobson</td>\n",
       "      <td>Jacobson says that energy costs with a wind, s...</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>0.906014</td>\n",
       "      <td>8.001624e-01</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>12</td>\n",
       "      <td>0.083333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>94375</th>\n",
       "      <td>94375</td>\n",
       "      <td>The forebrain during development is known as w...</td>\n",
       "      <td>prosencephalon</td>\n",
       "      <td>In many classes of vertebrates, such as fish a...</td>\n",
       "      <td>False</td>\n",
       "      <td>3</td>\n",
       "      <td>0.870792</td>\n",
       "      <td>4.084119e-01</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0.191546</td>\n",
       "      <td>8</td>\n",
       "      <td>0.375000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>288263</th>\n",
       "      <td>288263</td>\n",
       "      <td>What school thought that language was supposed...</td>\n",
       "      <td>Vedantins</td>\n",
       "      <td>In contrast,   Vedantins extended the scope an...</td>\n",
       "      <td>True</td>\n",
       "      <td>3</td>\n",
       "      <td>0.884770</td>\n",
       "      <td>4.200257e-01</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>0.100897</td>\n",
       "      <td>14</td>\n",
       "      <td>0.214286</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>81607</th>\n",
       "      <td>81607</td>\n",
       "      <td>The sahwm was an early family member of what i...</td>\n",
       "      <td>the oboe</td>\n",
       "      <td>Stringed instruments such as the viol had emer...</td>\n",
       "      <td>False</td>\n",
       "      <td>2</td>\n",
       "      <td>0.878302</td>\n",
       "      <td>6.933014e-01</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>0.127450</td>\n",
       "      <td>10</td>\n",
       "      <td>0.200000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>365833</th>\n",
       "      <td>365833</td>\n",
       "      <td>What year was the Polaris Missile Facility Atl...</td>\n",
       "      <td>1996</td>\n",
       "      <td>At the   1996 closure of the station\\'s Polari...</td>\n",
       "      <td>True</td>\n",
       "      <td>5</td>\n",
       "      <td>0.881376</td>\n",
       "      <td>2.571229e-01</td>\n",
       "      <td>6</td>\n",
       "      <td>9</td>\n",
       "      <td>0.336317</td>\n",
       "      <td>10</td>\n",
       "      <td>0.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>317255</th>\n",
       "      <td>317255</td>\n",
       "      <td>When was the first ground attack in the post-9...</td>\n",
       "      <td>21 March 2003</td>\n",
       "      <td>The first ground attack came at the Battle of ...</td>\n",
       "      <td>True</td>\n",
       "      <td>6</td>\n",
       "      <td>0.858254</td>\n",
       "      <td>2.558539e-01</td>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "      <td>0.168082</td>\n",
       "      <td>11</td>\n",
       "      <td>0.545455</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>255946</th>\n",
       "      <td>255946</td>\n",
       "      <td>What kind of nutritional value do the processe...</td>\n",
       "      <td>reduced</td>\n",
       "      <td>Nonetheless, processed foods tend to have an i...</td>\n",
       "      <td>False</td>\n",
       "      <td>7</td>\n",
       "      <td>0.844933</td>\n",
       "      <td>4.375494e-01</td>\n",
       "      <td>7</td>\n",
       "      <td>6</td>\n",
       "      <td>0.277159</td>\n",
       "      <td>16</td>\n",
       "      <td>0.437500</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>367845</th>\n",
       "      <td>367845</td>\n",
       "      <td>Muffling the sound of batter made what easier ...</td>\n",
       "      <td>sleep</td>\n",
       "      <td>Although many civilians had used them as such ...</td>\n",
       "      <td>False</td>\n",
       "      <td>3</td>\n",
       "      <td>0.936749</td>\n",
       "      <td>6.979640e-01</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>0.047241</td>\n",
       "      <td>12</td>\n",
       "      <td>0.250000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>268458</th>\n",
       "      <td>268458</td>\n",
       "      <td>What does a perfective verb have added in pref...</td>\n",
       "      <td>an added prefix</td>\n",
       "      <td>In prefix pairs, the perfective verb has   an ...</td>\n",
       "      <td>True</td>\n",
       "      <td>5</td>\n",
       "      <td>0.883137</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>3</td>\n",
       "      <td>8</td>\n",
       "      <td>0.475677</td>\n",
       "      <td>10</td>\n",
       "      <td>0.500000</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>250727</th>\n",
       "      <td>250727</td>\n",
       "      <td>What was the Swiss term for Switzerland during...</td>\n",
       "      <td>Switzer</td>\n",
       "      <td>The English adjective Swiss is a loan from Fre...</td>\n",
       "      <td>False</td>\n",
       "      <td>4</td>\n",
       "      <td>0.859583</td>\n",
       "      <td>3.169239e-01</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>0.162114</td>\n",
       "      <td>13</td>\n",
       "      <td>0.307692</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>193139</th>\n",
       "      <td>193139</td>\n",
       "      <td>Since what year has the away kit been changed ...</td>\n",
       "      <td>2009</td>\n",
       "      <td>When Nike took over from Adidas as Arsenal's k...</td>\n",
       "      <td>False</td>\n",
       "      <td>3</td>\n",
       "      <td>0.879752</td>\n",
       "      <td>6.002202e-01</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>0.188231</td>\n",
       "      <td>11</td>\n",
       "      <td>0.272727</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>146463</th>\n",
       "      <td>146463</td>\n",
       "      <td>Which MacBook Pro port can transfer data at sp...</td>\n",
       "      <td>Thunderbolt</td>\n",
       "      <td>Starting with the then-new iMac G5, released i...</td>\n",
       "      <td>False</td>\n",
       "      <td>2</td>\n",
       "      <td>0.909808</td>\n",
       "      <td>6.633747e-01</td>\n",
       "      <td>7</td>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>15</td>\n",
       "      <td>0.133333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>293273</th>\n",
       "      <td>293273</td>\n",
       "      <td>What is Ireland Act 1914?</td>\n",
       "      <td>which granted home rule to Ireland as a consti...</td>\n",
       "      <td>This body was suspended in 1972 and Northern I...</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>0.886877</td>\n",
       "      <td>3.625741e-01</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0.077654</td>\n",
       "      <td>5</td>\n",
       "      <td>0.200000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>327504</th>\n",
       "      <td>327504</td>\n",
       "      <td>What is the largest island in Paris?</td>\n",
       "      <td>Ile de la Cite</td>\n",
       "      <td>Montmartre gained its name from the martyrdom ...</td>\n",
       "      <td>False</td>\n",
       "      <td>2</td>\n",
       "      <td>0.877316</td>\n",
       "      <td>5.874769e-01</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0.075387</td>\n",
       "      <td>7</td>\n",
       "      <td>0.285714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>232822</th>\n",
       "      <td>232822</td>\n",
       "      <td>Who runs the school system in Myanmar ?</td>\n",
       "      <td>the Ministry of Education</td>\n",
       "      <td>The education system is based on the United Ki...</td>\n",
       "      <td>False</td>\n",
       "      <td>3</td>\n",
       "      <td>0.872168</td>\n",
       "      <td>1.648207e-01</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0.155153</td>\n",
       "      <td>8</td>\n",
       "      <td>0.375000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92542</th>\n",
       "      <td>92542</td>\n",
       "      <td>What sort of variants did the Human Genome Pro...</td>\n",
       "      <td>alternative variants</td>\n",
       "      <td>Subsequently, the sequencing in the Human Geno...</td>\n",
       "      <td>True</td>\n",
       "      <td>12</td>\n",
       "      <td>0.870705</td>\n",
       "      <td>9.538773e-02</td>\n",
       "      <td>7</td>\n",
       "      <td>8</td>\n",
       "      <td>0.411323</td>\n",
       "      <td>17</td>\n",
       "      <td>0.705882</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>332343</th>\n",
       "      <td>332343</td>\n",
       "      <td>What were gunki?</td>\n",
       "      <td>13th and 14th century warrior writings</td>\n",
       "      <td>It is that exactly that will be the great fame...</td>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "      <td>-1.000000</td>\n",
       "      <td>-1.000000e+00</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>3</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>233882</th>\n",
       "      <td>233882</td>\n",
       "      <td>What was the difference that made American cot...</td>\n",
       "      <td>longer, stronger fibers</td>\n",
       "      <td>By the 1840s, India was no longer capable of s...</td>\n",
       "      <td>False</td>\n",
       "      <td>3</td>\n",
       "      <td>0.869010</td>\n",
       "      <td>3.289867e-01</td>\n",
       "      <td>5</td>\n",
       "      <td>6</td>\n",
       "      <td>0.089205</td>\n",
       "      <td>12</td>\n",
       "      <td>0.250000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>251911</th>\n",
       "      <td>251911</td>\n",
       "      <td>What percentage of recyclable materials are be...</td>\n",
       "      <td>66% to 96%</td>\n",
       "      <td>The country is heavily active in recycling and...</td>\n",
       "      <td>True</td>\n",
       "      <td>6</td>\n",
       "      <td>0.866035</td>\n",
       "      <td>2.816685e-01</td>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>0.345680</td>\n",
       "      <td>10</td>\n",
       "      <td>0.600000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>47767</th>\n",
       "      <td>47767</td>\n",
       "      <td>Soldiers from which country were brought to th...</td>\n",
       "      <td>Cuban</td>\n",
       "      <td>Nevertheless, Massamba-Debat was unable to rec...</td>\n",
       "      <td>False</td>\n",
       "      <td>3</td>\n",
       "      <td>0.902506</td>\n",
       "      <td>5.264569e-01</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0.034243</td>\n",
       "      <td>9</td>\n",
       "      <td>0.333333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>352825</th>\n",
       "      <td>352825</td>\n",
       "      <td>At which point in a game is the time the offen...</td>\n",
       "      <td>last three minutes of a half</td>\n",
       "      <td>Instead of the penalty being 5 yards with the ...</td>\n",
       "      <td>False</td>\n",
       "      <td>3</td>\n",
       "      <td>0.840422</td>\n",
       "      <td>6.845175e-01</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>19</td>\n",
       "      <td>0.157895</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>156711</th>\n",
       "      <td>156711</td>\n",
       "      <td>Who was the target audience of natural history?</td>\n",
       "      <td>French polite society</td>\n",
       "      <td>The idea of taste (le gout) was a social indic...</td>\n",
       "      <td>False</td>\n",
       "      <td>3</td>\n",
       "      <td>0.900061</td>\n",
       "      <td>7.600788e-01</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>0.004412</td>\n",
       "      <td>8</td>\n",
       "      <td>0.375000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>133566</th>\n",
       "      <td>133566</td>\n",
       "      <td>What foreign organization encouraged Villa's g...</td>\n",
       "      <td>The German Secret Service</td>\n",
       "      <td>By early 1917 President Venustiano Carranza ha...</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>0.921094</td>\n",
       "      <td>6.709331e-01</td>\n",
       "      <td>6</td>\n",
       "      <td>1</td>\n",
       "      <td>0.099685</td>\n",
       "      <td>10</td>\n",
       "      <td>0.100000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>341277</th>\n",
       "      <td>341277</td>\n",
       "      <td>When was the earliest known Asian farm site fo...</td>\n",
       "      <td>2012</td>\n",
       "      <td>In   2012, news was released about a new farmi...</td>\n",
       "      <td>True</td>\n",
       "      <td>8</td>\n",
       "      <td>0.866599</td>\n",
       "      <td>1.959512e-01</td>\n",
       "      <td>7</td>\n",
       "      <td>7</td>\n",
       "      <td>0.273193</td>\n",
       "      <td>12</td>\n",
       "      <td>0.666667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>403224</th>\n",
       "      <td>403224</td>\n",
       "      <td>What is the policy in the Muslim world on depi...</td>\n",
       "      <td>prohibited</td>\n",
       "      <td>Images of Mohammed are likewise   prohibited.</td>\n",
       "      <td>True</td>\n",
       "      <td>0</td>\n",
       "      <td>0.874294</td>\n",
       "      <td>6.225225e-01</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>0.291195</td>\n",
       "      <td>11</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>94695</th>\n",
       "      <td>94695</td>\n",
       "      <td>The energy used for metabolism of the brain in...</td>\n",
       "      <td>2025%</td>\n",
       "      <td>The energy consumption of the brain does not v...</td>\n",
       "      <td>False</td>\n",
       "      <td>6</td>\n",
       "      <td>0.852285</td>\n",
       "      <td>3.799238e-01</td>\n",
       "      <td>5</td>\n",
       "      <td>4</td>\n",
       "      <td>0.167887</td>\n",
       "      <td>13</td>\n",
       "      <td>0.461538</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>175726</th>\n",
       "      <td>175726</td>\n",
       "      <td>What did Victoria think of Albert at 17 years ...</td>\n",
       "      <td>He possesses every quality that could be desir...</td>\n",
       "      <td>The parties did not undertake a formal engagem...</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>0.890218</td>\n",
       "      <td>7.854847e-01</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>0.007962</td>\n",
       "      <td>11</td>\n",
       "      <td>0.090909</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>183375</th>\n",
       "      <td>183375</td>\n",
       "      <td>What are Seleucid military colonies called?</td>\n",
       "      <td>katoikiai</td>\n",
       "      <td>After being defeated by Ptolemy IV's forces at...</td>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "      <td>0.863692</td>\n",
       "      <td>5.025759e-01</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>6</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26046</th>\n",
       "      <td>26046</td>\n",
       "      <td>Who first utilized solar architecture?</td>\n",
       "      <td>Greeks and Chinese</td>\n",
       "      <td>Advanced solar architecture and urban planning...</td>\n",
       "      <td>True</td>\n",
       "      <td>3</td>\n",
       "      <td>0.872083</td>\n",
       "      <td>2.516206e-01</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>0.304794</td>\n",
       "      <td>5</td>\n",
       "      <td>0.600000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>140962</th>\n",
       "      <td>140962</td>\n",
       "      <td>Since what year has tourism been ongoing to An...</td>\n",
       "      <td>1957</td>\n",
       "      <td>Not all vessels associated with Antarctic tour...</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>0.862064</td>\n",
       "      <td>3.232502e-01</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0.109296</td>\n",
       "      <td>9</td>\n",
       "      <td>0.111111</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>305869</th>\n",
       "      <td>305869</td>\n",
       "      <td>How long did Nasser attend law school?</td>\n",
       "      <td>one semester</td>\n",
       "      <td>In 1937, Nasser applied to the Royal Military ...</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>0.898796</td>\n",
       "      <td>5.497170e-01</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>0.042988</td>\n",
       "      <td>7</td>\n",
       "      <td>0.142857</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>347055</th>\n",
       "      <td>347055</td>\n",
       "      <td>What is the name of the region which is home t...</td>\n",
       "      <td>Vagad region</td>\n",
       "      <td>The Hadoti region lies to the southeast, on th...</td>\n",
       "      <td>False</td>\n",
       "      <td>4</td>\n",
       "      <td>0.862094</td>\n",
       "      <td>5.692305e-01</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>0.151472</td>\n",
       "      <td>12</td>\n",
       "      <td>0.333333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>107310</th>\n",
       "      <td>107310</td>\n",
       "      <td>What state-owned organization operates New Del...</td>\n",
       "      <td>Delhi Transport Corporation (DTC)</td>\n",
       "      <td>New Delhi has one of India's largest bus trans...</td>\n",
       "      <td>False</td>\n",
       "      <td>3</td>\n",
       "      <td>0.813312</td>\n",
       "      <td>1.928254e-01</td>\n",
       "      <td>6</td>\n",
       "      <td>7</td>\n",
       "      <td>0.378811</td>\n",
       "      <td>9</td>\n",
       "      <td>0.333333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>360934</th>\n",
       "      <td>360934</td>\n",
       "      <td>Which film did Kahn first work with Spielberg on?</td>\n",
       "      <td>Close Encounters</td>\n",
       "      <td>An example of this is his production relations...</td>\n",
       "      <td>False</td>\n",
       "      <td>2</td>\n",
       "      <td>0.887770</td>\n",
       "      <td>6.211694e-01</td>\n",
       "      <td>6</td>\n",
       "      <td>1</td>\n",
       "      <td>0.040909</td>\n",
       "      <td>9</td>\n",
       "      <td>0.222222</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>208767</th>\n",
       "      <td>208767</td>\n",
       "      <td>When did the battle for Mexico City take place</td>\n",
       "      <td>September 8 to September 15, 1847</td>\n",
       "      <td>The American invasion into the Federal Distric...</td>\n",
       "      <td>False</td>\n",
       "      <td>3</td>\n",
       "      <td>0.895585</td>\n",
       "      <td>5.548246e-01</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>0.031428</td>\n",
       "      <td>9</td>\n",
       "      <td>0.333333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20481</th>\n",
       "      <td>20481</td>\n",
       "      <td>What was the name of the island the English tr...</td>\n",
       "      <td>Run</td>\n",
       "      <td>Several intertribal wars among the Native Amer...</td>\n",
       "      <td>False</td>\n",
       "      <td>5</td>\n",
       "      <td>0.925379</td>\n",
       "      <td>7.976864e-01</td>\n",
       "      <td>7</td>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>18</td>\n",
       "      <td>0.277778</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>208415</th>\n",
       "      <td>208415</td>\n",
       "      <td>What option does the Recommended Exposure Inde...</td>\n",
       "      <td>to specify a camera models EI choices arbitrarily</td>\n",
       "      <td>The Recommended Exposure Index (REI) technique...</td>\n",
       "      <td>True</td>\n",
       "      <td>4</td>\n",
       "      <td>0.868986</td>\n",
       "      <td>1.993931e-01</td>\n",
       "      <td>5</td>\n",
       "      <td>6</td>\n",
       "      <td>0.351701</td>\n",
       "      <td>9</td>\n",
       "      <td>0.444444</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>377995</th>\n",
       "      <td>377995</td>\n",
       "      <td>What does the Age of Rationalism succeed?</td>\n",
       "      <td>Renaissance philosophy</td>\n",
       "      <td>Traditionally, the European intellectual trans...</td>\n",
       "      <td>False</td>\n",
       "      <td>3</td>\n",
       "      <td>0.863494</td>\n",
       "      <td>6.030717e-01</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0.058145</td>\n",
       "      <td>7</td>\n",
       "      <td>0.428571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72703</th>\n",
       "      <td>72703</td>\n",
       "      <td>Who was the coach of the Pittsburgh Gladiators...</td>\n",
       "      <td>Joe Haering</td>\n",
       "      <td>On Saturday, July 23, 1989, much of America le...</td>\n",
       "      <td>True</td>\n",
       "      <td>4</td>\n",
       "      <td>0.874320</td>\n",
       "      <td>1.986821e-08</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>0.313277</td>\n",
       "      <td>10</td>\n",
       "      <td>0.400000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>44685 rows × 13 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        index_id                                           question  \\\n",
       "68873      68873             Where is the theater District located?   \n",
       "195732    195732    Where can the public view the Royal Collection?   \n",
       "301126    301126  How many players can an NFL team have in the o...   \n",
       "318729    318729  What document declares supreme power for the p...   \n",
       "193367    193367         What value does Deloitte place on Arsenal?   \n",
       "20510      20510  Which city in North America held the most slav...   \n",
       "320616    320616  Compared to the rest of the US, does Alaska ha...   \n",
       "56702      56702  From what passage is cited as being one of the...   \n",
       "343785    343785  What religion was at the center of education i...   \n",
       "129620    129620     What type of carrier is the Charles de Gaulle?   \n",
       "203405    203405  The Ottomans controlled mainland Greece in wha...   \n",
       "306098    306098  Who was nominated to the presidency of the Off...   \n",
       "346900    346900  Under a Muslim Nawab there was a princely stat...   \n",
       "200448    200448              Why did Kim's war plan need revising?   \n",
       "216455    216455  What free service did Marshall's provide custo...   \n",
       "148679    148679           What gun did the Royal Navy start using?   \n",
       "440677    440677  In which period did North and South america join?   \n",
       "364615    364615  How does Texas handle people who don't speak E...   \n",
       "305686    305686    What organization did Nasser create with Syria?   \n",
       "173707    173707  What tax collectors did the Umayyads often lea...   \n",
       "324241    324241  What connects the parts inside a female insect...   \n",
       "335020    335020  Who is a professor of civil and environmental ...   \n",
       "94375      94375  The forebrain during development is known as w...   \n",
       "288263    288263  What school thought that language was supposed...   \n",
       "81607      81607  The sahwm was an early family member of what i...   \n",
       "365833    365833  What year was the Polaris Missile Facility Atl...   \n",
       "317255    317255  When was the first ground attack in the post-9...   \n",
       "255946    255946  What kind of nutritional value do the processe...   \n",
       "367845    367845  Muffling the sound of batter made what easier ...   \n",
       "268458    268458  What does a perfective verb have added in pref...   \n",
       "...          ...                                                ...   \n",
       "250727    250727  What was the Swiss term for Switzerland during...   \n",
       "193139    193139  Since what year has the away kit been changed ...   \n",
       "146463    146463  Which MacBook Pro port can transfer data at sp...   \n",
       "293273    293273                          What is Ireland Act 1914?   \n",
       "327504    327504               What is the largest island in Paris?   \n",
       "232822    232822            Who runs the school system in Myanmar ?   \n",
       "92542      92542  What sort of variants did the Human Genome Pro...   \n",
       "332343    332343                                   What were gunki?   \n",
       "233882    233882  What was the difference that made American cot...   \n",
       "251911    251911  What percentage of recyclable materials are be...   \n",
       "47767      47767  Soldiers from which country were brought to th...   \n",
       "352825    352825  At which point in a game is the time the offen...   \n",
       "156711    156711    Who was the target audience of natural history?   \n",
       "133566    133566  What foreign organization encouraged Villa's g...   \n",
       "341277    341277  When was the earliest known Asian farm site fo...   \n",
       "403224    403224  What is the policy in the Muslim world on depi...   \n",
       "94695      94695  The energy used for metabolism of the brain in...   \n",
       "175726    175726  What did Victoria think of Albert at 17 years ...   \n",
       "183375    183375        What are Seleucid military colonies called?   \n",
       "26046      26046             Who first utilized solar architecture?   \n",
       "140962    140962  Since what year has tourism been ongoing to An...   \n",
       "305869    305869             How long did Nasser attend law school?   \n",
       "347055    347055  What is the name of the region which is home t...   \n",
       "107310    107310  What state-owned organization operates New Del...   \n",
       "360934    360934  Which film did Kahn first work with Spielberg on?   \n",
       "208767    208767     When did the battle for Mexico City take place   \n",
       "20481      20481  What was the name of the island the English tr...   \n",
       "208415    208415  What option does the Recommended Exposure Inde...   \n",
       "377995    377995         What does the Age of Rationalism succeed?    \n",
       "72703      72703  Who was the coach of the Pittsburgh Gladiators...   \n",
       "\n",
       "                                              answer_text  \\\n",
       "68873                              south of Boston Common   \n",
       "195732                             at the Queen's Gallery   \n",
       "301126                                                 90   \n",
       "318729                        the Constitution of Estonia   \n",
       "193367                                             331.3m   \n",
       "20510                          Charleston, South Carolina   \n",
       "320616                                                few   \n",
       "56702                   Nag Hammadi's Apocalypse of Peter   \n",
       "343785                                              Islam   \n",
       "129620                                            CATOBAR   \n",
       "203405                                               1460   \n",
       "306098                                         Sirri Amer   \n",
       "346900                                               Tonk   \n",
       "200448    South Korean agents had learned about the plans   \n",
       "216455                                   personal shopper   \n",
       "148679             4.5-inch (114 mm) gun in a twin turret   \n",
       "440677                                            Neogene   \n",
       "364615  . The Government of Texas, through Section 205...   \n",
       "305686                               United Arab Republic   \n",
       "173707                                              Roman   \n",
       "324241                                              ducts   \n",
       "335020                                   Mark Z. Jacobson   \n",
       "94375                                      prosencephalon   \n",
       "288263                                          Vedantins   \n",
       "81607                                            the oboe   \n",
       "365833                                               1996   \n",
       "317255                                      21 March 2003   \n",
       "255946                                            reduced   \n",
       "367845                                              sleep   \n",
       "268458                                    an added prefix   \n",
       "...                                                   ...   \n",
       "250727                                            Switzer   \n",
       "193139                                               2009   \n",
       "146463                                        Thunderbolt   \n",
       "293273  which granted home rule to Ireland as a consti...   \n",
       "327504                                     Ile de la Cite   \n",
       "232822                          the Ministry of Education   \n",
       "92542                                alternative variants   \n",
       "332343             13th and 14th century warrior writings   \n",
       "233882                            longer, stronger fibers   \n",
       "251911                                         66% to 96%   \n",
       "47767                                               Cuban   \n",
       "352825                       last three minutes of a half   \n",
       "156711                              French polite society   \n",
       "133566                          The German Secret Service   \n",
       "341277                                               2012   \n",
       "403224                                         prohibited   \n",
       "94695                                               2025%   \n",
       "175726  He possesses every quality that could be desir...   \n",
       "183375                                          katoikiai   \n",
       "26046                                  Greeks and Chinese   \n",
       "140962                                               1957   \n",
       "305869                                       one semester   \n",
       "347055                                       Vagad region   \n",
       "107310                  Delhi Transport Corporation (DTC)   \n",
       "360934                                   Close Encounters   \n",
       "208767                  September 8 to September 15, 1847   \n",
       "20481                                                 Run   \n",
       "208415  to specify a camera models EI choices arbitrarily   \n",
       "377995                             Renaissance philosophy   \n",
       "72703                                         Joe Haering   \n",
       "\n",
       "                                                 sentence  target  \\\n",
       "68873   Other performing-arts organizations located in...   False   \n",
       "195732  It is not the monarch's personal property, unl...   False   \n",
       "301126  Split-squad games, a practice common in baseba...   False   \n",
       "318729  The people exercise their supreme power of the...   False   \n",
       "193367   3m (435.5m), ranking Arsenal seventh among wo...    True   \n",
       "20510   It also became a center of slavery, with 42% o...    True   \n",
       "320616  The western part of Alaska has no road system ...   False   \n",
       "56702   However, his accusers were excommunicated both...   False   \n",
       "343785  Wiesner-Hanks also says that Protestants also ...   False   \n",
       "129620  1 CATOBAR carrier: Charles de Gaulle is a 42,0...   False   \n",
       "203405  With the Turkish conquest, many Byzantine Gree...   False   \n",
       "306098  They selected Muhammad Naguib, a popular gener...   False   \n",
       "346900  They also resisted Mughal incursions into Indi...   False   \n",
       "200448  On 7 June 1950, Kim Il-sung called for a Korea...   False   \n",
       "216455  It grew to become a major chain before convert...   False   \n",
       "148679  At the same time the Royal Navy adopted a new ...    True   \n",
       "440677  Supercontinent Laurasia had not yet separated ...   False   \n",
       "364615  The Government of Texas, through Section 2054....   False   \n",
       "305686  In 1962, Nasser began a series of major social...   False   \n",
       "173707  These new converts continues to pay the same t...   False   \n",
       "324241  Accessory glands or glandular parts of the ovi...   False   \n",
       "335020  Jacobson says that energy costs with a wind, s...   False   \n",
       "94375   In many classes of vertebrates, such as fish a...   False   \n",
       "288263  In contrast,   Vedantins extended the scope an...    True   \n",
       "81607   Stringed instruments such as the viol had emer...   False   \n",
       "365833  At the   1996 closure of the station\\'s Polari...    True   \n",
       "317255  The first ground attack came at the Battle of ...    True   \n",
       "255946  Nonetheless, processed foods tend to have an i...   False   \n",
       "367845  Although many civilians had used them as such ...   False   \n",
       "268458  In prefix pairs, the perfective verb has   an ...    True   \n",
       "...                                                   ...     ...   \n",
       "250727  The English adjective Swiss is a loan from Fre...   False   \n",
       "193139  When Nike took over from Adidas as Arsenal's k...   False   \n",
       "146463  Starting with the then-new iMac G5, released i...   False   \n",
       "293273  This body was suspended in 1972 and Northern I...   False   \n",
       "327504  Montmartre gained its name from the martyrdom ...   False   \n",
       "232822  The education system is based on the United Ki...   False   \n",
       "92542   Subsequently, the sequencing in the Human Geno...    True   \n",
       "332343  It is that exactly that will be the great fame...   False   \n",
       "233882  By the 1840s, India was no longer capable of s...   False   \n",
       "251911  The country is heavily active in recycling and...    True   \n",
       "47767   Nevertheless, Massamba-Debat was unable to rec...   False   \n",
       "352825  Instead of the penalty being 5 yards with the ...   False   \n",
       "156711  The idea of taste (le gout) was a social indic...   False   \n",
       "133566  By early 1917 President Venustiano Carranza ha...   False   \n",
       "341277  In   2012, news was released about a new farmi...    True   \n",
       "403224      Images of Mohammed are likewise   prohibited.    True   \n",
       "94695   The energy consumption of the brain does not v...   False   \n",
       "175726  The parties did not undertake a formal engagem...   False   \n",
       "183375  After being defeated by Ptolemy IV's forces at...   False   \n",
       "26046   Advanced solar architecture and urban planning...    True   \n",
       "140962  Not all vessels associated with Antarctic tour...   False   \n",
       "305869  In 1937, Nasser applied to the Royal Military ...   False   \n",
       "347055  The Hadoti region lies to the southeast, on th...   False   \n",
       "107310  New Delhi has one of India's largest bus trans...   False   \n",
       "360934  An example of this is his production relations...   False   \n",
       "208767  The American invasion into the Federal Distric...   False   \n",
       "20481   Several intertribal wars among the Native Amer...   False   \n",
       "208415  The Recommended Exposure Index (REI) technique...    True   \n",
       "377995  Traditionally, the European intellectual trans...   False   \n",
       "72703   On Saturday, July 23, 1989, much of America le...    True   \n",
       "\n",
       "        exact_match  alignment_score  max_alignment  number_of_intent  \\\n",
       "68873             1         0.881072   5.594476e-01                 2   \n",
       "195732            1         0.879658   7.210963e-01                 3   \n",
       "301126            3         0.810726   4.730812e-01                 4   \n",
       "318729            3         0.828007   2.086947e-01                 4   \n",
       "193367            0         0.855577   6.227328e-01                 4   \n",
       "20510             2         0.865245   3.315323e-01                 5   \n",
       "320616            7         0.850673   4.609409e-01                 7   \n",
       "56702             2         0.908814   6.398091e-01                 6   \n",
       "343785            0         0.884671   7.056823e-01                 5   \n",
       "129620            5         0.853132   2.051662e-01                 4   \n",
       "203405            1         0.889290   6.251095e-01                 4   \n",
       "306098            4         0.943691   8.358789e-01                 3   \n",
       "346900            0         0.888309   6.904991e-01                 5   \n",
       "200448            0         0.907051   5.310434e-01                 3   \n",
       "216455            1         0.920871   8.219873e-01                 7   \n",
       "148679            4         0.880607   1.647606e-01                 4   \n",
       "440677            0         0.902033   8.490692e-01                 4   \n",
       "364615            2         0.888766   2.415099e-01                 4   \n",
       "305686            1         0.879276   5.257678e-01                 4   \n",
       "173707            2         0.904721   7.470397e-01                 5   \n",
       "324241            3         0.857441   5.160477e-01                 5   \n",
       "335020            1         0.906014   8.001624e-01                 6   \n",
       "94375             3         0.870792   4.084119e-01                 2   \n",
       "288263            3         0.884770   4.200257e-01                 2   \n",
       "81607             2         0.878302   6.933014e-01                 4   \n",
       "365833            5         0.881376   2.571229e-01                 6   \n",
       "317255            6         0.858254   2.558539e-01                 5   \n",
       "255946            7         0.844933   4.375494e-01                 7   \n",
       "367845            3         0.936749   6.979640e-01                 4   \n",
       "268458            5         0.883137   0.000000e+00                 3   \n",
       "...             ...              ...            ...               ...   \n",
       "250727            4         0.859583   3.169239e-01                 4   \n",
       "193139            3         0.879752   6.002202e-01                 3   \n",
       "146463            2         0.909808   6.633747e-01                 7   \n",
       "293273            1         0.886877   3.625741e-01                 2   \n",
       "327504            2         0.877316   5.874769e-01                 3   \n",
       "232822            3         0.872168   1.648207e-01                 3   \n",
       "92542            12         0.870705   9.538773e-02                 7   \n",
       "332343            0        -1.000000  -1.000000e+00                 0   \n",
       "233882            3         0.869010   3.289867e-01                 5   \n",
       "251911            6         0.866035   2.816685e-01                 4   \n",
       "47767             3         0.902506   5.264569e-01                 3   \n",
       "352825            3         0.840422   6.845175e-01                 5   \n",
       "156711            3         0.900061   7.600788e-01                 4   \n",
       "133566            1         0.921094   6.709331e-01                 6   \n",
       "341277            8         0.866599   1.959512e-01                 7   \n",
       "403224            0         0.874294   6.225225e-01                 4   \n",
       "94695             6         0.852285   3.799238e-01                 5   \n",
       "175726            1         0.890218   7.854847e-01                 5   \n",
       "183375            0         0.863692   5.025759e-01                 3   \n",
       "26046             3         0.872083   2.516206e-01                 3   \n",
       "140962            1         0.862064   3.232502e-01                 3   \n",
       "305869            1         0.898796   5.497170e-01                 4   \n",
       "347055            4         0.862094   5.692305e-01                 4   \n",
       "107310            3         0.813312   1.928254e-01                 6   \n",
       "360934            2         0.887770   6.211694e-01                 6   \n",
       "208767            3         0.895585   5.548246e-01                 4   \n",
       "20481             5         0.925379   7.976864e-01                 7   \n",
       "208415            4         0.868986   1.993931e-01                 5   \n",
       "377995            3         0.863494   6.030717e-01                 2   \n",
       "72703             4         0.874320   1.986821e-08                 3   \n",
       "\n",
       "        count_vectorizer_match  tfidf_match  question_words  coverage  \n",
       "68873                        1     0.025818               6  0.166667  \n",
       "195732                       0     0.000000               8  0.125000  \n",
       "301126                       2     0.100368              11  0.272727  \n",
       "318729                       4     0.354543               8  0.375000  \n",
       "193367                       1     0.114317               7  0.000000  \n",
       "20510                        2     0.081870              12  0.166667  \n",
       "320616                       7     0.470676              15  0.466667  \n",
       "56702                        1     0.078640              18  0.111111  \n",
       "343785                       0     0.000000              12  0.000000  \n",
       "129620                       5     0.384170               9  0.555556  \n",
       "203405                       0     0.000000               8  0.125000  \n",
       "306098                       1     0.006483              10  0.400000  \n",
       "346900                       0     0.000000              13  0.000000  \n",
       "200448                       1     0.071265               7  0.000000  \n",
       "216455                       2     0.008248              13  0.076923  \n",
       "148679                       4     0.183972               8  0.500000  \n",
       "440677                       0     0.000000               9  0.000000  \n",
       "364615                       4     0.097705              13  0.153846  \n",
       "305686                       1     0.065970               7  0.142857  \n",
       "173707                       1     0.090524              14  0.142857  \n",
       "324241                       0     0.000000              10  0.300000  \n",
       "335020                       0     0.000000              12  0.083333  \n",
       "94375                        1     0.191546               8  0.375000  \n",
       "288263                       2     0.100897              14  0.214286  \n",
       "81607                        2     0.127450              10  0.200000  \n",
       "365833                       9     0.336317              10  0.500000  \n",
       "317255                       3     0.168082              11  0.545455  \n",
       "255946                       6     0.277159              16  0.437500  \n",
       "367845                       1     0.047241              12  0.250000  \n",
       "268458                       8     0.475677              10  0.500000  \n",
       "...                        ...          ...             ...       ...  \n",
       "250727                       3     0.162114              13  0.307692  \n",
       "193139                       3     0.188231              11  0.272727  \n",
       "146463                       0     0.000000              15  0.133333  \n",
       "293273                       1     0.077654               5  0.200000  \n",
       "327504                       1     0.075387               7  0.285714  \n",
       "232822                       1     0.155153               8  0.375000  \n",
       "92542                        8     0.411323              17  0.705882  \n",
       "332343                       0     0.000000               3  0.000000  \n",
       "233882                       6     0.089205              12  0.250000  \n",
       "251911                       5     0.345680              10  0.600000  \n",
       "47767                        1     0.034243               9  0.333333  \n",
       "352825                       0     0.000000              19  0.157895  \n",
       "156711                       1     0.004412               8  0.375000  \n",
       "133566                       1     0.099685              10  0.100000  \n",
       "341277                       7     0.273193              12  0.666667  \n",
       "403224                       1     0.291195              11  0.000000  \n",
       "94695                        4     0.167887              13  0.461538  \n",
       "175726                       1     0.007962              11  0.090909  \n",
       "183375                       0     0.000000               6  0.000000  \n",
       "26046                        3     0.304794               5  0.600000  \n",
       "140962                       1     0.109296               9  0.111111  \n",
       "305869                       1     0.042988               7  0.142857  \n",
       "347055                       1     0.151472              12  0.333333  \n",
       "107310                       7     0.378811               9  0.333333  \n",
       "360934                       1     0.040909               9  0.222222  \n",
       "208767                       1     0.031428               9  0.333333  \n",
       "20481                        0     0.000000              18  0.277778  \n",
       "208415                       6     0.351701               9  0.444444  \n",
       "377995                       1     0.058145               7  0.428571  \n",
       "72703                        5     0.313277              10  0.400000  \n",
       "\n",
       "[44685 rows x 13 columns]"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "feature_df['question_words'] = feature_df.question.apply(lambda x: len(x.split()))\n",
    "feature_df['coverage'] = feature_df.exact_match / feature_df.question_words\n",
    "\n",
    "# Handle missing/null data\n",
    "\n",
    "feature_df.alignment_score.fillna(-1.0, inplace=True)\n",
    "feature_df.max_alignment.fillna(-1.0, inplace=True)\n",
    "display(feature_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1607
    },
    "colab_type": "code",
    "id": "eX8wASKTdhb6",
    "outputId": "c5beab26-4f78-4a22-dff5-fd49365f1f3d"
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA6gAAAE+CAYAAACA4T1fAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9f06645d68>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA6gAAAE+CAYAAACA4T1fAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4yLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvNQv5yAAAIABJREFUeJzt3X2UlnWdP/D3DMM0EYM5OGNaag9m\nlKJEJAqRASJI2aKCApm10rqu4MbGhpM916ZTSpEtbWuKsppGYqdFQ+C04LYuhNqcRazdpdQt8gFm\nFAQZiAHn90e/5jgyAgozc8G8XudwzlwP931/vnh/5vbN93vdV0lLS0tLAAAAoIuVdnUBAAAAkAio\nAAAAFISACgAAQCEIqAAAABSCgAoAAEAhCKgAAAAUgoAKQLfyne98J5/97GeTJB/72Mfyq1/9qosr\nat+OHTvyk5/8pKvLAIBOJaAC0G3NmzcvJ554YleX0a5f//rXAioA3Y6ACsAh684778zZZ5+ds846\nKx/5yEfyxBNPtDk+YsSIPPTQQ0mS733vezn99NNz/vnn5wc/+EFGjBiR5E8zrl/5ylcyderUjBw5\nMuPHj8+GDRuSJB/96Edzww035MILL8xpp52WH/zgB/nud7+bMWPGZOzYsVm3bl2S5Omnn85ll12W\n0aNHZ/To0fn3f//3JMkf/vCHvO9978u//Mu/5JxzzsmwYcOyaNGiNDY2Ztq0afmv//qvTJ48eY9j\nXL9+fT72sY9l7NixOfPMM/Otb30rSdLS0pJrrrkmI0aMyOjRo3PjjTcmSV544YV861vfypgxYzJm\nzJjU1tamqampdTzf+ta3cvbZZ6e+vj6bN2/Opz/96YwePTojR47MXXfddSD+swDAyxJQATgkPfPM\nM/nKV76Sm2++OUuXLs2xxx6b7373u+2e+5vf/CY33nhj/vVf/zW33357Fi9e3Ob44sWLc9VVV+Vn\nP/tZ+vbt2yaoPfjgg/nBD36Qa665Jtdee23e8IY3ZPHixTn++ONbz7vyyivTr1+/LFmyJDfccENm\nzpyZjRs3Jkk2btyY0tLS3H333bnqqqsye/bsHHHEEfnUpz6VAQMG5Pbbb9/jOG+55Za8973vzaJF\ni3L33Xdn3bp12bBhQxYuXJiHH344S5YsyV133ZXbbrstDz/8cO699978/Oc/z49//OP89Kc/zebN\nm3PLLbe0Pt8jjzySn/70pxk4cGDq6upSWlqae++9N3feeWe+853vZO3ata/mPwcA7BMBFYBDUt++\nffPLX/4yb3jDG5IkgwYNap3RfKkHH3wwp556ampqavKa17wm559/fpvjgwYNyhvf+MaUlJTkne98\nZ5566qnWY8OHD09ZWVlOOOGEbNu2LaNHj06SnHDCCdmwYUOampqyatWqfPzjH0+SHHfccXnPe97T\nOou6c+fOnHfeeUmSE088MU8++eQrHuf999+fhx56KOXl5fnmN7+Zmpqa/PznP8/o0aPTs2fP9O7d\nO4sWLUr//v1z3333Zdy4cenVq1d69OiR8847L//5n//Z+nxnnHFGSkv/9L8Hy5cvz8UXX5zS0tJU\nVVVl1KhRWbp06SuqDwBeibKuLgAAOsKuXbty/fXXZ9myZdm1a1e2bt2at7zlLe2eu3nz5hx22GGt\n20ceeWSb45WVla0/9+jRI7t27Wrdft3rXte6/8XbpaWleeGFF7Jly5a0tLRk4sSJrY9pamrKaaed\n1vq4Xr16tXnMK/Hxj388L7zwQr785S9nw4YN+chHPpIrrrgiGzduTJ8+fVrP+/NrPPvss23Gethh\nh+WZZ55ps/1nW7ZsyfTp01vH9sc//jFjxox5RfUBwCshoAJwSFq0aFGWLVuW2267LVVVVfnRj36U\nu+++u91ze/fu3XodZpLWa0wPhL59+6ZHjx656667WsPrn/3hD3/Y7+cvKyvLpZdemksvvTSPP/54\n/uqv/irvec97cvjhh7cuI06SxsbGVFRU5IgjjsimTZta92/atClHHHFEu89dU1OTOXPm5IQTTtjv\nOgFgX1jiC8Ah6Zlnnskb3/jGVFVVZePGjbn33nuzdevWds89+eSTs2rVqjz77LMH/PYuZWVlOeOM\nM/LDH/4wSbJt27Z85jOfabNM+OUe9/zzz6elpWWP533hC19oXaJ77LHH5ogjjkhJSUlGjBiRn/70\np9mxY0eampoyefLkrF27Nh/4wAeycOHCbNu2LTt37syCBQtyxhlntPvcI0aMaK17586dufrqqwt7\nWx4ADg0CKgCHpA996EPZtGlTRo0alRkzZmT69Ol5+umn2w2pJ598cs4999yce+65ufjiizN8+PAD\nWsuXvvSlPPjggxkzZkzOPffcHHPMMTnqqKP2+Jj3vOc92bBhQ4YNG9ZmSfFLTZw4sfVbeceOHZt3\nv/vdOf300zN27Ni8733vy1lnnZVzzz0348ePz8CBAzNmzJi8//3vz3nnnZcPfehDecMb3pCLL764\n3eeePn16tmzZktGjR+eDH/xgXnjhhbzjHe/Yr78LANiTkpa9/dMsAHQDLS0tKSkpSZLcd999mT17\ntvuQAkAnM4MKQLf37LPP5rTTTssTTzyRlpaW3HvvvRkwYEBXlwUA3Y4ZVABIcscdd2Tu3LkpKSnJ\nW9/61nzta19L3759u7qsPP/88xk/fny7x3r37p0FCxZ0ckUA0HEEVAAAAArBEl8AAAAKQUAFAACg\nEMq6uoD2NDRs6eoS9ujww3tl48amvZ8I3Yi+gLb0BLSlJ6Ct7t4T1dWV7e43g/oqlJX16OoSoHD0\nBbSlJ6AtPQFt6Yn2CagAAAAUgoAKAABAIQioAAAAFIKACgAAQCEIqAAAABSCgAoAAEAhCKgAAAAU\nQllXFwAAAEDneeqpJ3PxxRPzjnf0a9339re/I5/85Izdzp027dJ86lMz89a3Ht8ptQmoAAAAXeSS\numUH9Pnm1o7Yp/OOPfa4/OM/3nBAX/tAEFABAAC6uZ07d+ZrX/tSGho2ZNu2bbnkkkszdOiw1uNr\n1/5PZs36enr27Jny8vJ8+cvXpEeP0lx99ZezZcuW7Nq1K9OnfzrHH//2/apDQAUAAOjmtmzZnFNP\nPS1nn/2hPPHEH/L5z9e2CaiLFt2dc88dnzFjPphf/vLBPPvsM1m+/GcZPHhIzjlnXB5//LF8+9vX\nZfbs7+5XHXsNqKtWrconP/nJvP3tf0rCJ5xwQj7xiU9k5syZ2bVrV6qrq3PttdemvLw8CxcuzLx5\n81JaWpoLLrggEyZMSHNzc2pra/Pkk0+mR48eueaaa3LMMcfsV9EAAADteemS2X1d8trd/P73v8u0\naZe2bg8cOCibNm3MwoU/TklJaTZvfq7N+e973xm57rq6rFv3+4wcOSrHHffmrFnzcDZt2pglSxYl\nSf74x+37Xdc+zaCeeuqpuf7661u3P/OZz2Ty5Mk5++yz881vfjMLFizIuHHjMmfOnCxYsCA9e/bM\n+PHjM2rUqCxfvjx9+vTJrFmzcv/992fWrFmZPXv2fhcOAADAq/PSa1Dvvfee/P73v8ucOTdm8+bN\n+cQnPtrm/EGDTs2NN/5LVqz4j/zDP3wp06ZNT8+eZfm7v/t0Tjrp5ANW16u6zcyqVasycuTIJMnw\n4cOzcuXKrF69Ov37909lZWUqKioycODA1NfXZ+XKlRk1alSSZMiQIamvrz9gxQMAALD/Nm3alKOO\nOjqlpaX5939flubm5jbH77prfjZvfi5nnXV2Lrxwctau/Z+8610n5ec/vy9J8vjjj+WHP7xtv+vY\npxnU3/72t7nsssvy3HPPZdq0adm2bVvKy8uTJH379k1DQ0MaGxtTVVXV+piqqqrd9peWlqakpCQ7\nduxofTwAAABd6wMfGJHa2k/l179+JB/84IdTU1OTm2/+fuvxN77xmHz+87Xp3bt3evbsmauu+mIq\nKiryta99KZdf/om88MILmT797/e7jr0G1De/+c2ZNm1azj777Kxbty4XX3xxdu3a1Xq8paWl3ce9\n0v0vdvjhvVJW1mOv53Wl6urKri4BCkdfQFt6AtrSE3SEc2b86x6PF/l9V11dmbtn/UUXvO47snDh\nT3bbt2jRT1u3P/KRC3Z73DnnjN5t3w03fO+A1rbXgHrkkUdm7NixSZJjjz02RxxxRNasWZPt27en\noqIi69evT01NTWpqatLY2Nj6uA0bNmTAgAGpqalJQ0ND+vXrl+bm5rS0tOx19nTjxqb9HFbHqq6u\nTEPDlq4uAwpFX0BbegLa0hN0laK+77p7T7zcPxzs9RrUhQsX5qabbkqSNDQ05Jlnnsl5552XJUuW\nJEmWLl2aYcOG5ZRTTsmaNWuyefPmbN26NfX19Rk0aFCGDh2axYsXJ0mWL1+ewYMHH6gxAQAAcAjZ\n6wzqiBEj8vd///f5t3/7tzQ3N+dLX/pS3vnOd+bKK6/M/Pnzc/TRR2fcuHHp2bNnZsyYkSlTpqSk\npCRTp05NZWVlxo4dmxUrVmTSpEkpLy9PXV1dZ4wLAACAg0xJy75cFNrJij7V3d2n46E9+gLa0hPQ\nlp6go7z0vqcvVdT7oHb3nnjVS3wBAACgMwioAAAAFMI+3QcVAACAg993vvOt/O///neeffaZbN++\nPUcf/cb06XNYrr762q4uLYmACgAA0GWmLpt5QJ9vzohv7PH4FVf8XZJk0aK789hjj2batOkH9PX3\nlyW+AAAA3Vh9/UOZOXN6pk27NP/zP/+dD35wZOuxz31uZurrH0pT09Z87nMz88lP/k2mTbs0v/3t\nbzqkFjOoAAAA3dyjj/42d9zx45SXl7d7/Ec/uiODBw/JOeeMy+OPP5Zvf/u6zJ793QNeh4AKAADQ\nzR1//NtfNpwmyZo1D2fTpo1ZsmRRkuSPf9zeIXUIqAAAAN1cz549292/c+fO/3+8LH/3d5/OSSed\n3KF1CKgAAMAh65K6ZW2259aO6KJKDh4lJSXZvv1PM6Rr1/5vkuRd7zopP//5fTnppJPz+OOPZdWq\nFZk48aID/toCKgAAAK3GjRufSy/9WN785rfmHe94Z5Jk/PgL87WvfSmXX/6JvPDCC5k+/e875LVL\nWlpaWjrkmfdDQ8OWri5hj6qrKwtfI3Q2fQFt6QloS0/QUV46Q7o3RZlB7e49UV1d2e5+t5kBAACg\nEARUAAAACkFABQAAoBAEVAAAAApBQAUAAKAQBFQAAAAKQUAFAACgEARUAAAACkFABQAAoBAEVAAA\nAApBQAUAAKAQBFQAAAAKQUAFAACgEARUAAAACkFABQAAoBAEVAAAAApBQAUAAKAQBFQAAAAKQUAF\nAACgEARUAAAACkFABQAAoBAEVAAAAApBQAUAAKAQBFQAAAAKQUAFAACgEARUAAAACkFABQAAoBAE\nVAAAAAqhrKsLAAAA2FeX1C3r6hLoQGZQAQAAKAQBFQAAgELYp4C6ffv2nHnmmfnxj3+cp556Kh/9\n6EczefLkfPKTn8yOHTuSJAsXLsz555+fCRMm5M4770ySNDc3Z8aMGZk0aVIuuuiirFu3ruNGAgAA\nwEFtnwLqP/3TP+Wwww5Lklx//fWZPHlybr/99hx33HFZsGBBmpqaMmfOnNxyyy259dZbM2/evGza\ntCn33HNP+vTpkzvuuCOXXXZZZs2a1aGDAQAA4OC114D66KOP5re//W0+8IEPJElWrVqVkSNHJkmG\nDx+elStXZvXq1enfv38qKytTUVGRgQMHpr6+PitXrsyoUaOSJEOGDEl9fX3HjQQAAICD2l4D6te/\n/vXU1ta2bm/bti3l5eVJkr59+6ahoSGNjY2pqqpqPaeqqmq3/aWlpSkpKWldEgwAAAAvtsfbzPzk\nJz/JgAEDcswxx7R7vKWl5YDsf6nDD++VsrIe+3RuV6muruzqEqBw9AW0pSegLT1BERTpfVikWopi\njwH1vvvuy7p163Lffffl6aefTnl5eXr16pXt27enoqIi69evT01NTWpqatLY2Nj6uA0bNmTAgAGp\nqalJQ0ND+vXrl+bm5rS0tLTOvu7Jxo1N+z+yDlRdXZmGhi1dXQYUir6AtvQEtKUnKIqivA+7e0+8\nXDjf4xLf2bNn56677sqPfvSjTJgwIZdffnmGDBmSJUuWJEmWLl2aYcOG5ZRTTsmaNWuyefPmbN26\nNfX19Rk0aFCGDh2axYsXJ0mWL1+ewYMHH+BhAQAAcKjY4wxqe6644opceeWVmT9/fo4++uiMGzcu\nPXv2zIwZMzJlypSUlJRk6tSpqayszNixY7NixYpMmjQp5eXlqaur64gxAAAAcAgoadnXC0M7UdGn\nurv7dDy0R19AW3oC2tITHCiX1C3br8fPrR1xgCrZP929J17VEl8AAADoLAIqAAAAhSCgAgAAUAgC\nKgAAAIUgoAIAAFAIAioAAACFIKACAABQCAIqAAAAhSCgAgAAUAgCKgAAAIUgoAIAAFAIAioAAACF\nIKACAABQCAIqAAAAhSCgAgAAUAgCKgAAAIUgoAIAAFAIAioAAACFIKACAABQCAIqAAAAhSCgAgAA\nUAgCKgAAAIVQ1tUFAAAAdJZL6pa12Z5bO6KLKqE9ZlABAAAoBAEVAACAQhBQAQAAKAQBFQAAgEIQ\nUAEAACgEARUAAIBCEFABAAAoBAEVAACAQhBQAQAAKAQBFQAAgEIQUAEAACgEARUAAIBCEFABAAAo\nBAEVAACAQhBQAQAAKAQBFQAAgEIQUAEAACgEARUAAIBCEFABAAAohLK9nbBt27bU1tbmmWeeyR//\n+Mdcfvnl6devX2bOnJldu3aluro61157bcrLy7Nw4cLMmzcvpaWlueCCCzJhwoQ0NzentrY2Tz75\nZHr06JFrrrkmxxxzTGeMDQAAgIPIXmdQly9fnpNOOim33XZbZs+enbq6ulx//fWZPHlybr/99hx3\n3HFZsGBBmpqaMmfOnNxyyy259dZbM2/evGzatCn33HNP+vTpkzvuuCOXXXZZZs2a1RnjAgAA4CCz\n1xnUsWPHtv781FNP5cgjj8yqVavy5S9/OUkyfPjwzJ07N295y1vSv3//VFZWJkkGDhyY+vr6rFy5\nMuPGjUuSDBkyJFdddVVHjAMAADgEXVK3rKtLoBPtNaD+2cSJE/P000/ne9/7Xv7yL/8y5eXlSZK+\nffumoaEhjY2Nqaqqaj2/qqpqt/2lpaUpKSnJjh07Wh8PAAAAySsIqD/84Q/z3//93/n0pz+dlpaW\n1v0v/vnFXun+Fzv88F4pK+uxr6V1ierqyq4uAQpHX0BbegLa0hMUUVe+L/XE7vYaUB955JH07ds3\nRx11VN75zndm165ded3rXpft27enoqIi69evT01NTWpqatLY2Nj6uA0bNmTAgAGpqalJQ0ND+vXr\nl+bm5rS0tOx19nTjxqb9H1kHqq6uTEPDlq4uAwpFX0BbegLa0hMUVVe9L7t7T7xcON/rlyQ99NBD\nmTt3bpKksbExTU1NGTJkSJYsWZIkWbp0aYYNG5ZTTjkla9asyebNm7N169bU19dn0KBBGTp0aBYv\nXpzkT1+4NHjw4AM1JgAAAA4he51BnThxYj772c9m8uTJ2b59e77whS/kpJNOypVXXpn58+fn6KOP\nzrhx49KzZ8/MmDEjU6ZMSUlJSaZOnZrKysqMHTs2K1asyKRJk1JeXp66urrOGBcAAAAHmZKWfbko\ntJMVfaq7u0/HQ3v0BbSlJ6AtPcGr1dHf4ju3dkSHPv/L6e498aqX+AIAAEBnEFABAAAoBAEVAACA\nQhBQAQAAKAQBFQAAgEIQUAEAACgEARUAAIBCEFABAAAoBAEVAACAQhBQAQAAKAQBFQAAgEIQUAEA\nACgEARUAAIBCEFABAAAoBAEVAACAQhBQAQAAKAQBFQAAgEIQUAEAACgEARUAAIBCEFABAAAoBAEV\nAACAQhBQAQAAKAQBFQAAgEIQUAEAACgEARUAAIBCEFABAAAoBAEVAACAQhBQAQAAKAQBFQAAgEIQ\nUAEAACgEARUAAIBCEFABAAAoBAEVAACAQhBQAQAAKAQBFQAAgEIQUAEAACgEARUAAIBCKOvqAgAA\nALrKJXXL2mzPrR3RRZWQmEEFAACgIARUAAAACsESXwAAoDBeuuSW7sUMKgAAAIWwTzOo3/jGN/LL\nX/4yO3fuzF//9V+nf//+mTlzZnbt2pXq6upce+21KS8vz8KFCzNv3ryUlpbmggsuyIQJE9Lc3Jza\n2to8+eST6dGjR6655pocc8wxHT0uAAAADjJ7Dai/+MUv8pvf/Cbz58/Pxo0bc+655+b000/P5MmT\nc/bZZ+eb3/xmFixYkHHjxmXOnDlZsGBBevbsmfHjx2fUqFFZvnx5+vTpk1mzZuX+++/PrFmzMnv2\n7M4YGwAAAAeRvS7xfe9735tvf/vbSZI+ffpk27ZtWbVqVUaOHJkkGT58eFauXJnVq1enf//+qays\nTEVFRQYOHJj6+vqsXLkyo0aNSpIMGTIk9fX1HTgcAAAADlZ7Dag9evRIr169kiQLFizI+9///mzb\nti3l5eVJkr59+6ahoSGNjY2pqqpqfVxVVdVu+0tLS1NSUpIdO3Z0xFgAAAA4iO3zt/j+7Gc/y4IF\nCzJ37tycddZZrftbWlraPf+V7n+xww/vlbKyHvtaWpeorq7s6hKgcPQFtKUnoC09wcGgM9+nemJ3\n+xRQ/+M//iPf+973cuONN6aysjK9evXK9u3bU1FRkfXr16empiY1NTVpbGxsfcyGDRsyYMCA1NTU\npKGhIf369Utzc3NaWlpaZ19fzsaNTfs3qg5WXV2ZhoYtXV0GFIq+gLb0BLSlJzhYdNb7tLv3xMuF\n870u8d2yZUu+8Y1v5J//+Z/z+te/PsmfriVdsmRJkmTp0qUZNmxYTjnllKxZsyabN2/O1q1bU19f\nn0GDBmXo0KFZvHhxkmT58uUZPHjwgRoTAAAAh5C9zqAuWrQoGzduzPTp01v31dXV5XOf+1zmz5+f\no48+OuPGjUvPnj0zY8aMTJkyJSUlJZk6dWoqKyszduzYrFixIpMmTUp5eXnq6uo6dEAAAAAcnEpa\n9uWi0E5W9Knu7j4dD+3RF9CWnoC29AT76pK6ZV36+nNrR3TK63T3nnjVS3wBAACgMwioAAAAFIKA\nCgAAQCEIqAAAABSCgAoAAEAhCKgAAAAUgoAKAABAIQioAAAAFIKACgAAQCEIqAAAABSCgAoAAEAh\nCKgAAAAUgoAKAABAIQioAAAAFIKACgAAQCEIqAAAABSCgAoAAEAhCKgAAAAUgoAKAABAIQioAAAA\nFIKACgAAQCEIqAAAABSCgAoAAEAhCKgAAAAUgoAKAABAIQioAAAAFIKACgAAQCEIqAAAABSCgAoA\nAEAhCKgAAAAUQllXFwAAAHRfl9Qt6+oSKBAzqAAAABSCgAoAAEAhCKgAAAAUgoAKAABAIQioAAAA\nFIKACgAAQCEIqAAAABSC+6ACAAD8fy+9L+vc2hFdVEn3JKACAACF99pTF7fZ3vbAmC6qhI5kiS8A\nAACFIKACAABQCPu0xHft2rW5/PLL8/GPfzwXXXRRnnrqqcycOTO7du1KdXV1rr322pSXl2fhwoWZ\nN29eSktLc8EFF2TChAlpbm5ObW1tnnzyyfTo0SPXXHNNjjnmmI4eFwAAcAiz5PfQtNcZ1Kampnz1\nq1/N6aef3rrv+uuvz+TJk3P77bfnuOOOy4IFC9LU1JQ5c+bklltuya233pp58+Zl06ZNueeee9Kn\nT5/ccccdueyyyzJr1qwOHRAAAHDwe+2pi9v8oXvYa0AtLy/P97///dTU1LTuW7VqVUaOHJkkGT58\neFauXJnVq1enf//+qaysTEVFRQYOHJj6+vqsXLkyo0aNSpIMGTIk9fX1HTQUAAAADmZ7DahlZWWp\nqKhos2/btm0pLy9PkvTt2zcNDQ1pbGxMVVVV6zlVVVW77S8tLU1JSUl27NhxIMcAAADAIWC/bzPT\n0tJyQPa/2OGH90pZWY/9qqujVVdXdnUJUDj6AtrSE9CWnuBg1JHvWz2xu1cVUHv16pXt27enoqIi\n69evT01NTWpqatLY2Nh6zoYNGzJgwIDU1NSkoaEh/fr1S3Nzc1paWlpnX1/Oxo1Nr6asTlNdXZmG\nhi1dXQYUir6AtvQEtKUnOFh11Pu2u/fEy4XzV3WbmSFDhmTJkiVJkqVLl2bYsGE55ZRTsmbNmmze\nvDlbt25NfX19Bg0alKFDh2bx4j9d1Lx8+fIMHjz4VQ4BAACAQ9leZ1AfeeSRfP3rX88TTzyRsrKy\nLFmyJNddd11qa2szf/78HH300Rk3blx69uyZGTNmZMqUKSkpKcnUqVNTWVmZsWPHZsWKFZk0aVLK\ny8tTV1fXGeMCAADgIFPSsi8XhXayok91d/fpeGiPvoC29AS0pSd4OZfULUuy+31NX6mOug/q3NoR\nHfK83b0nDugSXwAAADjQBFQAAAAKQUAFAACgEARUAAAACuFV3QcVAADg1fjzlyIdaC/9kqWO+tIk\nOpYZVAAAAApBQAUAAKAQBFQAAAAKwTWoAABAp3vpNaOQmEEFAACgIARUAAAACkFABQAAoBAEVAAA\nAApBQAUAAKAQBFQAAAAKQUAFAACgENwHFQAA6HBTl81Mkrz21C4u5BW6pG5Zm+25tSO6qJLuwQwq\nAAAAhWAGFQAAOOS89tTFbba3PTCmiyrhlTCDCgAAQCEIqAAAABSCJb4AAECH+fOXDB1sX45E1zCD\nCgAAQCEIqAAAABSCJb6vwjkz/rXNtnshAQAA7D8zqAAAABSCgAoAAEAhWOILAAB0iKnLZvr2Xl4R\nM6gAAAAUgoAKAABAIQioAAAAFIJrUAEAgANi6rKZXV3Cy3rtqYtbf972wJgurIQ9MYMKAABAIZhB\nBQAA2EeX1C1rsz23dkQXVXJoMoMKAABAIQioAAAAFIIlvgAAwKtS5C9F4uBkBhUAAIBCMIMKAAB0\nKy++5UzitjNFIqACAAD7xJLe3flW3wPLEl8AAAAKwQwqAADQLjOmdLZOCahXX311Vq9enZKSklx1\n1VU5+eSTO+NlO8xL16wnpvEBADj4CaR0tQ4PqA888EB+97vfZf78+Xn00Udz1VVXZf78+R39sgAA\nAPvkQH5p0kuvSU1cl/pKdHixNSkiAAAF4ElEQVRAXblyZc4888wkydve9rY899xzef7559O7d++O\nfmkAAOBFzJBSdB0eUBsbG3PiiSe2bldVVaWhoUFABQCAV+iSumXtXG7Ggba3v+NXOsPa3qzqnnTn\nGddO/5KklpaWvZ5TXV3ZCZW8ej+68J+6ugQopKL3LnQ2PQFt6Yn9d/esv0jyF11dBhd2dQGHrg6/\nzUxNTU0aGxtbtzds2JDq6uqOflkAAAAOMh0eUIcOHZolS5YkSX71q1+lpqbG8l4AAAB20+FLfAcO\nHJgTTzwxEydOTElJSb74xS929EsCAABwECpp2ZeLQgEAAKCDdfgSXwAAANgXAioAAACFIKDuowce\neCCnn356li9f3u7xhQsX5vzzz8+ECRNy5513dnJ10Lmam5szY8aMTJo0KRdddFHWrVu32zknnnhi\nPvrRj7b+2bVrVxdUCh3v6quvzoUXXpiJEyfm4YcfbnNsxYoVGT9+fC688MLMmTOniyqEzrenvhgx\nYkQmT57c+vmwfv36LqoSOs/atWtz5pln5rbbbtvtmM+Ktjr9PqgHo9///ve5+eabM3DgwHaPNzU1\nZc6cOVmwYEF69uyZ8ePHZ9SoUXn961/fyZVC57jnnnvSp0+fzJo1K/fff39mzZqV2bNntzmnd+/e\nufXWW7uoQugcDzzwQH73u99l/vz5efTRR3PVVVdl/vz5rcf/4R/+ITfddFOOPPLIXHTRRRk9enSO\nP/74LqwYOt7e+iJJvv/97+d1r3tdF1UInaupqSlf/epXc/rpp7d73GdFW2ZQ90F1dXX+8R//MZWV\n7d9cevXq1enfv38qKytTUVGRgQMHpr6+vpOrhM6zcuXKjBo1KkkyZMgQ73e6rZUrV+bMM89Mkrzt\nbW/Lc889l+effz5Jsm7duhx22GE56qijUlpamjPOOCMrV67synKhU+ypL6A7Ki8vz/e///3U1NTs\ndsxnxe4E1H3w2te+Nj169HjZ442NjamqqmrdrqqqSkNDQ2eUBl3ixe/50tLSlJSUZMeOHW3O2bFj\nR2bMmJGJEyfm5ptv7ooyocM1Njbm8MMPb91+8e//hoYGnw10S3vqiz/74he/mEmTJuW6666LG0pw\nqCsrK0tFRUW7x3xW7M4S35e48847d7uG9IorrsiwYcP2+Tn8ouVQ0l5PrF69us12e+/5mTNn5sMf\n/nBKSkpy0UUXZdCgQenfv3+H1gpdze9/2N1L++Jv//ZvM2zYsBx22GGZOnVqlixZkjFjxnRRdUDR\nCKgvMWHChEyYMOEVPaampiaNjY2t2xs2bMiAAQMOdGnQJdrridra2jQ0NKRfv35pbm5OS0tLysvL\n25wzadKk1p9PO+20rF27VkDlkNPe7//q6up2j61fv77d5V1wqNlTXyTJuHHjWn9+//vfn7Vr1wqo\ndFs+K3Znie8BcMopp2TNmjXZvHlztm7dmvr6+gwaNKiry4IOM3To0CxevDhJsnz58gwePLjN8cce\neywzZsxIS0tLdu7cmfr6+rz97W/vilKhQw0dOjRLlixJkvzqV79KTU1NevfunSR505velOeffz5/\n+MMfsnPnzixfvjxDhw7tynKhU+ypL7Zs2ZIpU6a0Xhby4IMP+nygW/NZsbuSFuuR9uq+++7LTTfd\nlMceeyxVVVWprq7O3Llzc8MNN+S9731v3v3ud2fx4sW56aabWpczfvjDH+7qsqHD7Nq1K5/73Ofy\nf//3fykvL09dXV2OOuqoNj1x7bXX5he/+EVKS0szYsSI/M3f/E1Xlw0d4rrrrstDDz2UkpKSfPGL\nX8yvf/3rVFZWZtSoUXnwwQdz3XXXJUnOOuusTJkypYurhc6xp76YN29efvKTn+Q1r3lN3vWud+Xz\nn/98SkpKurpk6DCPPPJIvv71r+eJJ55IWVlZjjzyyIwYMSJvetObfFa0Q0AFAACgECzxBQAAoBAE\nVAAAAApBQAUAAKAQBFQAAAAKQUAFAACgEARUAAAACkFABQAAoBAEVAAAAArh/wHqSNnp6Y73TgAA\nAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9f083f3198>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA64AAAE+CAYAAACN/00YAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9f10350f28>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA64AAAE+CAYAAACN/00YAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9f0803a6d8>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA60AAAE+CAYAAABmyPYbAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAMi9AKAAAAADAsQisAAAAAwLD+P94e7Vv1AZl2AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9f10b29240>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "feature_to_test = 'count_vectorizer_match'\n",
    "\n",
    "def plot_feature_hist(feature_to_test, xticks=[], bins=200):\n",
    "    plt.figure(figsize=(16,5))\n",
    "    plt.hist(feature_df[feature_df.target == 0][feature_to_test], bins=bins, label='False')\n",
    "    plt.hist(feature_df[feature_df.target == 1][feature_to_test], bins=bins, label='True')\n",
    "    plt.legend(loc='upper right')\n",
    "    plt.title(feature_to_test)\n",
    "    if len(xticks): plt.xticks(xticks)\n",
    "    plt.show()\n",
    "\n",
    "plot_feature_hist('max_alignment')\n",
    "plot_feature_hist('alignment_score')\n",
    "plot_feature_hist('count_vectorizer_match', list(range(0, 30,1)))\n",
    "plot_feature_hist('tfidf_match')\n",
    "plot_feature_hist('coverage')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "collapsed": true,
    "id": "aToQX-NZdhb9"
   },
   "outputs": [],
   "source": [
    "# save features as csv\n",
    "\n",
    "feature_df.to_csv('data/feature_df_all.tsv', sep='\\t')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 534
    },
    "colab_type": "code",
    "id": "KnTTbPCPdhb_",
    "outputId": "ea4185b1-d4b9-45ec-b321-3cb9377263d7"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AuC:  0.8985756363492077\n",
      "Accuracy score:  0.865955542294495\n",
      "[[9700 1090]\n",
      " [ 707 1909]]\n",
      "             precision    recall  f1-score   support\n",
      "\n",
      "          0       0.93      0.90      0.92     10790\n",
      "          1       0.64      0.73      0.68      2616\n",
      "\n",
      "avg / total       0.87      0.87      0.87     13406\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAj4AAAFKCAYAAAAdaVXmAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n8AEAAIZB8AEAAIbx/wE/sIN91bYkuwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9f0a007160>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn import metrics\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "features = feature_df[['question_words', 'exact_match', 'coverage', 'number_of_intent', \\\n",
    "                       'alignment_score', 'max_alignment', 'count_vectorizer_match', 'tfidf_match']]\n",
    "target = feature_df['target'].map({True: 1, False: 0})\n",
    "\n",
    "features_train, features_test, target_train, target_test = train_test_split(features, target, test_size=0.3, random_state=6)\n",
    " \n",
    "model = RandomForestClassifier(n_estimators=10, max_depth=8, random_state=0, class_weight={0: 1, 1: 2.5})\n",
    "model = model.fit(features_train, target_train)\n",
    "\n",
    "(pd.Series(model.feature_importances_, index=features.columns).plot(kind='barh')) \n",
    "\n",
    "predicted = model.predict(features_test)\n",
    "probs = model.predict_proba(features_test)\n",
    "# print (probs)\n",
    "print (\"AuC: \", metrics.roc_auc_score(target_test, probs[:, 1]))\n",
    " \n",
    "print (\"Accuracy score: \", metrics.accuracy_score(target_test, predicted))\n",
    "print (metrics.confusion_matrix(target_test, predicted))\n",
    "print (metrics.classification_report(target_test, predicted))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 376
    },
    "colab_type": "code",
    "id": "fxTpSlZtdhcD",
    "outputId": "064a9d80-1032-4f58-b9a6-6fd62a9d9bcb"
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfYAAAFnCAYAAABU0WtaAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/wAMAw20VoNEHgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f9f03b41780>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import roc_curve, auc\n",
    "import matplotlib.pyplot as plt\n",
    "import random\n",
    "\n",
    "false_positive_rate, true_positive_rate, thresholds = roc_curve(target_test, probs[:, 1])\n",
    "roc_auc = auc(false_positive_rate, true_positive_rate)\n",
    "\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.plot(false_positive_rate, true_positive_rate, 'b', label='AUC = %0.2f'% roc_auc)\n",
    "plt.legend(loc='lower right')\n",
    "plt.plot([0,1],[0,1],'r--')\n",
    "plt.xlim([-0.1,1.2])\n",
    "plt.ylim([-0.1,1.2])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "2GGRujbWdhcF",
    "outputId": "60b584c4-8c2f-4729-ec85-574d632be175"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['random_forest_model_sample.sav']"
      ]
     },
     "execution_count": 85,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.externals import joblib\n",
    "\n",
    "filename = 'model/random_forest_model_sample.sav'\n",
    "joblib.dump(model, filename)"
   ]
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "github-qa-rf.ipynb",
   "provenance": [],
   "version": "0.3.2"
  },
  "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": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}