0_EDA_and_Topic_Modeling_with_LDA.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Exploratory Data Analysis and Topic Modeling\n",
    "\n",
    "This section will cover:\n",
    "\n",
    "* initial exploratory analysis\n",
    "* grammar and spelling correction \n",
    "* feature extraction with SpaCy \n",
    "* topic modeling with latent Dirichlet allocation (LDA).\n",
    "* automatic scoring with LDA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import re\n",
    "from datetime import datetime\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "import spacy\n",
    "from spacy.lang.en.stop_words import STOP_WORDS\n",
    "from string import punctuation\n",
    "\n",
    "import language_check\n",
    "\n",
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "from sklearn.decomposition import LatentDirichletAllocation\n",
    "from sklearn.externals import joblib\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "import pyLDAvis\n",
    "from pyLDAvis.sklearn import prepare\n",
    "\n",
    "plt.style.use('seaborn-colorblind')\n",
    "\n",
    "# Setup Pandas\n",
    "pd.set_option('display.width', 500)\n",
    "pd.set_option('display.max_columns', 100)\n",
    "pd.set_option('display.notebook_repr_html', True)\n",
    "pd.set_option('display.max_colwidth', 100)\n",
    "\n",
    "pyLDAvis.enable_notebook()\n",
    "# plt.rcParams['figure.figsize'] = [8, 5]\n",
    "plt.rcParams['figure.dpi']= 100\n",
    "\n",
    "import warnings\n",
    "warnings.simplefilter(\"ignore\", DeprecationWarning)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>essay_id</th>\n",
       "      <th>topic</th>\n",
       "      <th>essay</th>\n",
       "      <th>rater1_domain1</th>\n",
       "      <th>rater2_domain1</th>\n",
       "      <th>rater3_domain1</th>\n",
       "      <th>target_score</th>\n",
       "      <th>rater1_domain2</th>\n",
       "      <th>rater2_domain2</th>\n",
       "      <th>topic2_target</th>\n",
       "      <th>rater1_trait1</th>\n",
       "      <th>rater1_trait2</th>\n",
       "      <th>rater1_trait3</th>\n",
       "      <th>rater1_trait4</th>\n",
       "      <th>rater1_trait5</th>\n",
       "      <th>rater1_trait6</th>\n",
       "      <th>rater2_trait1</th>\n",
       "      <th>rater2_trait2</th>\n",
       "      <th>rater2_trait3</th>\n",
       "      <th>rater2_trait4</th>\n",
       "      <th>rater2_trait5</th>\n",
       "      <th>rater2_trait6</th>\n",
       "      <th>rater3_trait1</th>\n",
       "      <th>rater3_trait2</th>\n",
       "      <th>rater3_trait3</th>\n",
       "      <th>rater3_trait4</th>\n",
       "      <th>rater3_trait5</th>\n",
       "      <th>rater3_trait6</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>5405</th>\n",
       "      <td>8959</td>\n",
       "      <td>4</td>\n",
       "      <td>In this short story, Saeng vows to take the test again in the spring because she knows that she ...</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>3</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      essay_id  topic                                                                                                essay  rater1_domain1  rater2_domain1  rater3_domain1  target_score  rater1_domain2  rater2_domain2  topic2_target  rater1_trait1  rater1_trait2  rater1_trait3  rater1_trait4  rater1_trait5  rater1_trait6  rater2_trait1  rater2_trait2  rater2_trait3  rater2_trait4  rater2_trait5  rater2_trait6  rater3_trait1  rater3_trait2  rater3_trait3  rater3_trait4  rater3_trait5  \\\n",
       "5405      8959      4  In this short story, Saeng vows to take the test again in the spring because she knows that she ...               3               2             NaN             3             NaN             NaN            NaN            NaN            NaN            NaN            NaN            NaN            NaN            NaN            NaN            NaN            NaN            NaN            NaN            NaN            NaN            NaN            NaN            NaN   \n",
       "\n",
       "      rater3_trait6  \n",
       "5405            NaN  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "training_set  = pd.read_csv('training_set_rel3.tsv', sep='\\t', encoding = \"ISO-8859-1\")\\\n",
    "            .rename(columns={'essay_set': 'topic', 'domain1_score': 'target_score', 'domain2_score': 'topic2_target'})\n",
    "training_set.sample()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 12976 entries, 0 to 12975\n",
      "Data columns (total 28 columns):\n",
      "essay_id          12976 non-null int64\n",
      "topic             12976 non-null int64\n",
      "essay             12976 non-null object\n",
      "rater1_domain1    12976 non-null int64\n",
      "rater2_domain1    12976 non-null int64\n",
      "rater3_domain1    128 non-null float64\n",
      "target_score      12976 non-null int64\n",
      "rater1_domain2    1800 non-null float64\n",
      "rater2_domain2    1800 non-null float64\n",
      "topic2_target     1800 non-null float64\n",
      "rater1_trait1     2292 non-null float64\n",
      "rater1_trait2     2292 non-null float64\n",
      "rater1_trait3     2292 non-null float64\n",
      "rater1_trait4     2292 non-null float64\n",
      "rater1_trait5     723 non-null float64\n",
      "rater1_trait6     723 non-null float64\n",
      "rater2_trait1     2292 non-null float64\n",
      "rater2_trait2     2292 non-null float64\n",
      "rater2_trait3     2292 non-null float64\n",
      "rater2_trait4     2292 non-null float64\n",
      "rater2_trait5     723 non-null float64\n",
      "rater2_trait6     723 non-null float64\n",
      "rater3_trait1     128 non-null float64\n",
      "rater3_trait2     128 non-null float64\n",
      "rater3_trait3     128 non-null float64\n",
      "rater3_trait4     128 non-null float64\n",
      "rater3_trait5     128 non-null float64\n",
      "rater3_trait6     128 non-null float64\n",
      "dtypes: float64(22), int64(5), object(1)\n",
      "memory usage: 2.8+ MB\n"
     ]
    }
   ],
   "source": [
    "training_set.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "training_set.groupby('topic').agg('count').plot.bar(y='essay', rot=0, legend=False)\n",
    "plt.title('Essay count by topic #')\n",
    "plt.ylabel('Count')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Since topic 8 has the fewest essays and the most distinct scores, it might prove to be the most challenging topic to model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Count characters and words for each essay\n",
    "training_set['word_count'] = training_set['essay'].str.strip().str.split().str.len()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe8AAAEVCAYAAADJgG2yAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAH5tJREFUeJzt3X+8bHVd7/HXW8S8hcXBAwR4jscfJ9O6ZViCvylCkTKsh3QhEyKKyh9JoVcoKzVLzKuVpRYpgb/1XjFAKeVylR+WCCgihHiOCIfj4acoP/yBkZ/7x1o7h83+MXvvmT1rzX49H495zMx31qz1nvnuPZ+1vmvNrFQVkiSpP+436QCSJGlpLN6SJPWMxVuSpJ6xeEuS1DMWb0mSesbiLUlSz1i8pRFI8ook75x0jqVIcmqSV086x2xJnpvko5POIXWZxVtTJ8mJSc6e1bZlnrbDVzfd6khyQJLtq7i8SvLIUcyrqt5VVU9fYZ6/SPKb7e3rkvzAKLJJXWHx1jQ6H3hSkp0AkvwgsDOw76y2R7bTDi0N/2+673HApUl2B75dVbdPOpA0Sn4IaRpdTFOsH9vefyrwMeDqWW1frKodAEmemOTiJLe310+cmVmSjyf5sySfAL4BPDzJw5Kcl+TOJOcA6xcKlOTQJJcluSPJF5Mc3LbvneTMJLcl2Tqztdg+dq9h7dlb00muTfKSJJe3ud+X5IFJvg/4Z2DvJHe1l73nibY+yTnt6zgvyUPbeb8pyetnvYazkhw3x2ubWQH6bLus/9G2/2b7mm5rX+PeA8+pJL+b5JoktyZ53cxKUZJfS3LhwLQ/0ma8LclNSf5gkfc6wI8AVwA/CXxmoemlPrJ4a+pU1beBi2gKNO31BcCFs9rOB0iyG/Bh4I3Ag4E3AB9O8uCB2T4POBZ4EHAd8G7gUpqi/afAUfPlSfJ44O3AS4Fd22Vf2z78HmA7sDfwHODPkxy4hJf7y8DBwMOAHwN+raq+DjwT2FFVu7SXHfM8/7lt/vXAZcC72vbTgCMGCup64MA2771U1cx7+uPtst6X5GeA17T59qJ5z94766m/SFNc9wUOBX599ryTPAj4v8C/0LxHjwTOneuFJNmc5GvA7e3ruRn4IPCsJF9L8rx53gOpdyzemlbn8d1C/RSa4n3BrLbz2ts/B2ypqndU1T1V9R7g88CzBuZ3alVdWVX30BSjnwL+qKrurqrzgbMWyHIMcEpVnVNV36mqL1fV55NsAJ4MvKyqvlVVlwFvpVlRGNYbq2pHVd3WZnjsYk+Y5cNVdX5V3Q38IfCEJBuq6lM0RXBmReJw4ONVddOQ830uzWv+dDvvE9t5bxqY5rVVdVtVbQP+Cjhijvn8PHBjVb2+fY/urKqL5lpgVW2pql1pVsKOB9YBXwAeWVW7VtU7hswudZ7FW9PqfODJSdYBu1fVFuBfgSe2bT/Kd/d3702zZTjoOmCfgfvXD9zeG/hqu4U7OP18NgBfnKN9b+C2qrpzgeUu5saB298AdlnCc2HgdVXVXcBtbS5otr5/tb39q8BSit+93tN23l9h/vf0uoHlDprvvbuPJP/abnmfCLwKuAN4NHBlkv+zhOxS51m8Na3+DfgBmqHuTwBU1R3AjrZtR1V9qZ12B/DQWc/fCHx54P7g6fduANa1+5YHp5/P9cAj5mjfAezWDg3PtdyvA9878NgPLrCM2YY9XeCGmRtJdgF2a3MBvBM4NMmP0xTBf1rC8u/1nrbv1YO593u6YeD2xoHlDprvvbuPqnoi8CiaUZQfAF5Os3W/a1U9ZwnZpc6zeGsqVdU3gUuA36cZLp9xYds2eJT52cAPJfmVJPdvD7h6DPCheeZ9XTvvVyZ5QJInc+8h9tneBhyd5MAk90uyT5IfrqrraUYDXtMeaPZjNEPsM/udLwMOSbJbe3T8fQ4WW8BNwIOH+IrUIUmenOQBNPu+L2pzUVXbaQ7+ewfwgfY9XWh5Dx+4/+72NT82yfcAf97O+9qBaV6aZF27++DFwPvmmO+HgB9MclyS70nyoCT7LZBj8AC1fWn6SZo6Fm9Ns/OAPWgK9owL2rb/Kt5V9RWafavH0wzt/k/g56vq1gXm/SvAfjTDzH9Cc0DanNr9x0cDf0mzH/k8vrtVegSwiWar84PAn1TVOe1j7wA+S3Nw20eZu7jNt8zP0xxcdk17sNZ8R5u/u81/G83Xq5476/HTgP/O4kPmrwBOa5f1y1V1LvBHwAdoRioeQbPffNAZNAf9XUZzwODb5ngddwIH0awc3QhsAX56gRyPAz7d3t63nb80dVI17OiapLUmyVNphs83VdV3RjjfAjZX1dZRzVNaS9zyljSnJDvTDGe/dZSFW9LKWbwl3UeSRwNfo/la3F9NOI6kWRw2lySpZ9zyliSpZyzekiT1jMVbkqSesXhLktQzFm9JknrG4i1JUs9YvCVJ6hmL9woleWGSS5LcneTUSefRyrUnwHhbkuuS3JnkM0meOelcWrkk70xyQ5I7knwhyW9MOpNGJ8nmJN9K8s5JZxk3i/fK7QBeDZwy6SAamfvTnIryaTSnFf0j4P1JNk0wk0bjNTS/0/79wC8Ar07yuAln0ui8ieZMeFPP4r1CVXV6Vf0TzdmoNAWq6utV9YqquraqvlNVHwK+RHPGKvVYVV1ZVXfP3G0vQ50vXN2W5HCan/Q9d9JZVoPFW1pEkj2BHwKunHQWrVySNyf5BvB5mtOVnj3hSFqhJN8PvIrmtL5rgsVbWkB7Zq13Aae158hWz1XV84EHAU8BTgfuXvgZ6oE/Bd5WVddPOshqsXhL80hyP+AdwLeBF044jkaoqv6zqi4EHgL8zqTzaPmSPBb4WeAvJ51lNd1/0gGkLkoS4G3AnsAhVfUfE46k8bg/7vPuuwOATcC25t+WXYCdkjymqvadYK6xcst7hZLcP8kDgZ1o/mAemMSVov57C/Bo4FlV9c1Jh9HKJdkjyeFJdkmyU5JnAEcA/2/S2bQiJ9OsgD22vfwd8GHgGZMMNW4W75V7OfBN4ATgV9vbL59oIq1IkocCv0XzQXBjkrvay3MnHE0rUzRD5NuBrwL/Cziuqs6YaCqtSFV9o6punLkAdwHfqqpbJp1tnFJVk84gSZKWwC1vSZJ6xuItSVLPWLwlSeoZi7ckST3Tia80rV+/vjZt2jTpGAIuvfTSW6tq91HMy37tjlH1q33aLfbr9Bm2TztRvDdt2sQll1wy6RgCklw3qnnZr90xqn61T7vFfp0+w/apw+aSJPWMxVuSpJ6xeEuS1DMWb0mSesbiLUlSz1i8JUnqGYu3JEk9Y/GWJKlnLN6SJPWMxVuSpJ6xeEuS1DMWb0mSeqYTJyaZpBx/1n/drtc/a4JJJEkazqJb3kk2JPlYkquSXJnkxW37bknOSbKlvV7XtifJG5NsTXJ5kn3H/SIkaFbEZi6SNM2GGTa/Bzi+qh4N7A+8IMljgBOAc6tqM3Buex/gmcDm9nIs8JaRp5YkaQ1bdNi8qm4Abmhv35nkKmAf4FDggHay04CPAy9r299eVQV8MsmuSfZq5yOtioW2vt09IqnvlnTAWpJNwE8AFwF7zhTk9nqPdrJ9gOsHnra9bZMk9Zi7pLpj6APWkuwCfAA4rqruSDLvpHO01RzzOxZ4KbDr7rvvPmwMdZz9On2W2qczH/COcHTbUvrVot09Q215J9mZpnC/q6pOb5tvSrJX+/hewM1t+3Zgw8DTHwLsmD3Pqjq5qjZX1e4bN25cbv6x8cCn5el6v2rp7NPpZL/22zBHmwd4G3BVVb1h4KEzgaPa20cBZwy0H9kedb4/cLv7uyVJGp1htryfBDwP+Jkkl7WXQ4CTgIOSbAEOau8DnA1cA2wF/gF4/uhjS+oDR6+k8RjmaPMLmXs/NsCBc0xfwAtWmEsam9kFxX2z42Hhnk45/iz/ZzrAn0eVJKlnLN6SJPWMxVuStCTuEpm8NX9ikkFz/UF64hJJUte45S1Jmpdb2d1k8ZYkqWcs3pIk9YzFW5KknrF4S5LUMxZvSZJ6xuItSVLP+D1v9ZZfYZG0Vq3Z4u0Hv2b4Qzzj5YkspNFz2FyStGRuAE2WxVuSpJ6xeEtaNTn+LLfYpBFYU/u8/dCQJsP/vek0068e07D61lTxXikPbJIkdYHD5pJGxi1saXVYvCVJK+JK2+qzeEuS1DMWb0mSesbiLUlSzyxavJOckuTmJFcMtO2W5JwkW9rrdW17krwxydYklyfZd5zhh+V3SyVp6fzc7K5htrxPBQ6e1XYCcG5VbQbObe8DPBPY3F6OBd4ympjS6phZ0XOFb7x8b6WVWfR73lV1fpJNs5oPBQ5ob58GfBx4Wdv+9qoq4JNJdk2yV1XdMKrAWtv80Jek5e/z3nOmILfXe7Tt+wDXD0y3vW27jyTHtsPut2zbtm2ZMdQ19uv0sU+nk/3ab6M+YC1ztNVcE1bVyVW1uap237hx44hjaFLs1+ljn04n+7Xfllu8b0qyF0B7fXPbvh3YMDDdQ4Ady48naVp5XIG0fMst3mcCR7W3jwLOGGg/sj3qfH/gdvd3S5I0WosesJbkPTQHp61Psh34E+Ak4P1JjgG2AYe1k58NHAJsBb4BHD2GzJ0w1xaDJyuZPp6MZnhuRUurZ5ijzY+Y56ED55i2gBesNJQ0w4Ig9UOOP8sV3FXkL6xJktQzFm9Jknpm0WHzvnK4VZI0rdzyliSpZyzekiT1jMVbkqSesXhLkkbCY41Wz9QesCaN2uwPJr/TKmlSLN7qHNfe1xZ/3ENaOofNJUkj48r36piqLW//aCRJa8FUFe9J8yQWkqTVYPGWlsmVNUmT4j5vSZJ6xuItacU83kRaXRZvSRNn8ZeWxn3emrhp+OD2B1wkrSa3vMckx581FUVJkpbKz7/xs3hLktQzvR02d62u3+w/qdtG8T/qT9+Oj1ve0hjMDBu6kjI83ytpeL3d8u4Lf8hDHsw2PLfUppP9OnpjKd5JDgb+GtgJeGtVnTSS+fZ8zXwmv3/EktaKvn9ud9XIi3eSnYA3AQcB24GLk5xZVf8+6mVJfeRozMJcyZUWN44t78cDW6vqGoAk7wUOBZZUvKd5bW2+1+aHlaRp5dD5aI2jeO8DXD9wfzuw3xiWM3XmKur+sU+3hVZSZ/f9Wtti98N++jiqMjrjKN6Zo63uM1FyLPBSYFfgW0muHHh4I7BtDNnGYaxZ84aRzWrYnA9dyULs19FZqO8HHht7vy7Sp0vJsPRlj+7vHzrQp0tgvw6vL/060j5N1X3q6ookeQLwiqp6Rnv/RICqes0S5nFLVe0+0mBj0pesXcjZhQzD6kvWLuTsQoZh9CUndCNrFzIMY63mHMf3vC8GNid5WJIHAIcDZy5xHl8bfayx6UvWLuTsQoZh9SVrF3J2IcMw+pITupG1CxmGsSZzjnzYvKruSfJC4CM0XxU7papmD8cs5vZR5xqjvmTtQs4uZBhWX7J2IWcXMgyjLzmhG1m7kGEYazLnWL7nXVVnA2evYBYnjyrLKuhL1i7k7EKGYfUlaxdydiHDMPqSE7qRtQsZhrEmc458n7ckSRovf9tckqSesXhLktQzFm9JknrG4i1JUs9YvCVJ6hmLtyRJPWPxliSpZyzekiT1jMVbkqSesXhLktQzFm9JknrG4i1JUs9YvEcgyceTfCvJXe3l6kln0solOTzJVUm+nuSLSZ4y6UxavoH/z5nLfyb5m0nn0sol2ZTk7CRfTXJjkr9NMpazZnaFxXt0XlhVu7SXR006jFYmyUHAa4GjgQcBTwWumWgorcjA/+cuwJ7AN4H/PeFYGo03AzcDewGPBZ4GPH+iicZsqtdMpBV4JfCqqvpke//LkwyjkXsOzYf9BZMOopF4GPC3VfUt4MYk/wL8yIQzjZVb3qPzmiS3JvlEkgMmHUbLl2Qn4CeB3ZNsTbK9HYb7b5POppE5Cnh7VdWkg2gk/ho4PMn3JtkHeCbwLxPONFYW79F4GfBwYB/gZOCsJI+YbCStwJ7AzjRbZ0+hGYb7CeDlkwyl0UiykWZY9bRJZ9HInEezpX0HsB24BPiniSYaM4v3CFTVRVV1Z1XdXVWnAZ8ADpl0Li3bN9vrv6mqG6rqVuAN2KfT4kjgwqr60qSDaOWS3A/4CHA68H3AemAdzTErU8viPR4FZNIhtDxV9VWatXeHVKfTkbjVPU12AzbQ7PO+u6q+AvwjU76ybfFeoSS7JnlGkgcmuX+S59IcmfyRSWfTivwj8KIkeyRZBxwHfGjCmbRCSZ5Is3vLo8ynRDsy9iXgd9rP4F1pjmn47GSTjZfFe+V2Bl4N3ALcCrwIeHZV+V3vfvtT4GLgC8BVwGeAP5toIo3CUcDpVXXnpINopH4JOJjmc3grcA/wexNNNGbxYEtJkvrFLW9JknrG4i1JUs9YvCVJ6hmLtyRJPWPxliSpZzpxYpL169fXpk2bJh1DwKWXXnprVe0+innZr90xqn61T7vFfp0+w/ZpJ4r3pk2buOSSSyYdQ0CS60Y1L/u1O0bVr/Zpt9iv02fYPnXYXJKknrF4S5LUMxZvdV6OP4scf9akY0hSZ1i8JUnqGYu3JEk9Y/FWJzlULknzs3hLktQzFm9JknrG4q1Oc+hcku7L4i1JUs9YvCVJ6hmLtyRJPbNo8U6yIcnHklyV5MokL27bd0tyTpIt7fW6tj1J3phka5LLk+w77hchSdJaMsyW9z3A8VX1aGB/4AVJHgOcAJxbVZuBc9v7AM8ENreXY4G3jDy1JElr2KLFu6puqKpPt7fvBK4C9gEOBU5rJzsNeHZ7+1Dg7dX4JLBrkr1GnlySpDVqSfu8k2wCfgK4CNizqm6ApsADe7ST7QNcP/C07W3b2Ph1IknSWjJ08U6yC/AB4LiqumOhSedoqznmd2y7v/yWbdu2DRtDHWe/Th/7dDrZr/02VPFOsjNN4X5XVZ3eNt80MxzeXt/ctm8HNgw8/SHAjtnzrKqTq2pzVe2+cePG5eZXx4yzXx1hmQz/V6eT/dpvwxxtHuBtwFVV9YaBh84EjmpvHwWcMdB+ZHvU+f7A7TPD6+PgB7rUL550Rlq5+w8xzZOA5wGfS3JZ2/YHwEnA+5McA2wDDmsfOxs4BNgKfAM4eqSJJfXKTKGu1z9rwkmk6bFo8a6qC5l7PzbAgXNMX8ALVphrQa61S5LWMn9hTZKknrF4q3McWZGkhQ2zz7sz/FCXJKlnxVtSf7nyLY2Ow+aSJPWMxVuSpJ6xeEuS1DMWb0mSesbiLUlSz3i0uaSxWOzocn82VVo+i7ckSUMYXCGd9Epnb4bN/Y6oJEmN3hRvSZLUsHhLktQzFm9Jknpmaop3jj/L/eKSpDVhaoq3JElrhcVb0kQ5aiYtnd/zVmf4AS5Jw3HLW5KknulF8XaLTJKk7+pF8V4KC33/LGWfp/tHJWkKi7ckSdNu0eKd5JQkNye5YqBttyTnJNnSXq9r25PkjUm2Jrk8yb7jDC9J0lo0zJb3qcDBs9pOAM6tqs3Aue19gGcCm9vLscBbRhNTUl+4a0Mav0WLd1WdD9w2q/lQ4LT29mnAswfa316NTwK7JtlrVGElSdLy93nvWVU3ALTXe7Tt+wDXD0y3vW2TpAW5xS4Nb9QHrGWOtppzwuTYdp/5Ldu2bRtxDE2K/Tp97NPpZL/223KL900zw+Ht9c1t+3Zgw8B0DwF2zDWDqjq5qjZX1e4bN25cZgx1jf06fezT6WS/9ttyi/eZwFHt7aOAMwbaj2yPOt8fuH1meF2SJI3GMF8Vew/wb8CjkmxPcgxwEnBQki3AQe19gLOBa4CtwD8Azx9Laq157huVtJYtemKSqjpinocOnGPaAl6w0lCSJGl+/sKaJEk9M5XF26+cSJKm2VQWb0mSppnFWxPlCIkkLZ3FW5KknrF4S+oUj1mRFmfxliSpZ6a6eLv2LkmaRov+SMskWXyl/hjX/+vMfOv1zxrL/KU+muotb0mSppHFW5Kknun0sLm0EIdTp5u7zaT5WbwlSVNjcKVvmBX7pU7fFRZvTYRbVZLGra+FeRgWb0lS70xzYR7G1BfvHH/WmuxYSdLSzDUi2NX6MfXFW5I03YbZDTdtW+p+VUxSL/ib59J3rYktb79SJEn9Ms4t5aWsBHZ1hXFNFG91y6j/GTyuYW1ypXz6zPfZ0NUCOkmdLd52liRNp2nb/zwJnS3ekjQXV+yni/25PGuueDvUJkmrw8I8PmMp3kkOBv4a2Al4a1WdNI7lLJV/SNLo+X+lQf49rI6RF+8kOwFvAg4CtgMXJzmzqv591MtSv4zzn9oRlbVr9t+VfwNaC8bxPe/HA1ur6pqq+jbwXuDQMSxnRVw7nE4z3wW2f9cu+3/1+F5PzjiGzfcBrh+4vx3YbwzLWbHFjnh0a67f5urf2V8rs4+n13xb5LP73L+BuU3Dd6GnWapqtDNMDgOeUVW/0d5/HvD4qnrRrOmOBV4K7ArsAlw58PBGYNtIg41PX7IOm/OhVbX7chdiv666sffrIn26lAyT1pecYL8uxbTlHKpPx1G8nwC8oqqe0d4/EaCqXrOEedyykgKymvqStQs5u5BhWH3J2oWcXcgwjL7khG5k7UKGYazVnOPY530xsDnJw5I8ADgcOHOJ8/ja6GONTV+ydiFnFzIMqy9Zu5CzCxmG0Zec0I2sXcgwjDWZc+T7vKvqniQvBD5C81WxU6pq9nDMYm4fda4x6kvWLuTsQoZh9SVrF3J2IcMw+pITupG1CxmGsSZzjuV73lV1NnD2CmZx8qiyrIK+ZO1Czi5kGFZfsnYhZxcyDKMvOaEbWbuQYRhrMufI93lLkqTx8nzekiT1TOeKd5KDk1ydZGuSEzqQ59okn0tyWZJL2rbdkpyTZEt7va5tT5I3ttkvT7LvmLOdkuTmJFcMtC05W5Kj2um3JDlqTFnt1+Gz9aJfu9an0N1+7UuftsvoVL92tU/b5U2uX6uqMxeaA9y+CDwceADwWeAxE850LbB+VttfACe0t08AXtvePgT4ZyDA/sBFY872VGBf4IrlZgN2A65pr9e1t9fZr/Zr3/q0y/3ahz7tar92tU8n3a8T65B53ognAB8ZuH8icGIH/3CuBvZqb+8FXN3e/nvgiLmmG2O+TbP+cJaUDTgC+PuB9ntNZ7/ar33p0673a9f7tKv92uU+nWS/dm3YfK6fVt1nQllmFPDRJJem+UUigD2r6gaA9nqPtr0L+ZeabTUyd+F9mc1+XZkuvCdz6VO/dq1PF1r2JPWpT5eTbVmZu3Y+78zRNunD4Z9UVTuS7AGck+TzC0zbxfwz5su2Gpm7+L7Yr+NZ7qRNQ7/6v3pv09CnMOJ+7dqW93Zgw8D9hwA7JpQFgKra0V7fDHyQ5qxpNyXZC6C9vrmdvAv5l5ptNTJ34X25F/t1xbrwntxHz/q1a3260LInpmd9yjKyLStz14r3KH5adWSSfF+SB83cBp4OXNFmmjki8CjgjPb2mcCR7VGF+wO3zwyfrKKlZvsI8PQk69qjIp/eto2S/bpyXevXTvUp9LJfu9an0LF+7WGfzmQYf79O8kCEeXb+HwJ8geaIxz+ccJaH0xxt+VmaM+78Ydv+YOBcYEt7vVvbHuBNbfbPAT855nzvAW4A/oNm7e2Y5WQDfh3Y2l6Otl/t1771adf7tS992rV+7XKfTrpf/YU1SZJ6pmvD5pIkaREWb0mSesbiLUlSz1i8JUnqGYu3JEk9Y/GWJC0oSSV5/cD9lyR5xYjmfWqS54xiXoss57AkVyX52LiX1S7v15L87bjmb/GWJC3mbuCXkqyfdJBBSXZawuTHAM+vqp8eQ44kWdV6avGWJC3mHuBk4PdmPzB7yznJXe31AUnOS/L+JF9IclKS5yb5VJrzcz9iYDY/m+SCdrqfb5+/U5LXJbk4zfmvf2tgvh9L8m6aHzuZneeIdv5XJHlt2/bHwJOBv0vyulnTvznJL7S3P5jklPb2MUle3d7+/XZ+VyQ5rm3b1G7Jvxn4NLAhydHtazgPeNLAMg5rn/vZJOcv8b2fU9dOTCJJ6qY3AZcn+YslPOfHgUcDt9Gcp/qtVfX4JC8GXgQc1063CXga8AjgY0keCRxJ8xOiP5Xke4BPJPloO/3jgR+tqi8NLizJ3sBrgccBX6U5G9mzq+pVSX4GeElVXTIr4/nAU2h+vnQfmtN0QlPs35vkccDRwH40v5J2UVucvwo8iuYX0Z6f5nfMX9ku+3bgY8Bn2nn9MfCMqvpykl2X8P7Nyy1vSdKiquoO4O3A7y7haRdX1Q1VdTfNz4LOFN/P0RTsGe+vqu9U1RaaIv/DNL/xfWSSy4CLaH52dHM7/admF+7WTwEfr6pbquoe4F3AUxfJeAHwlCSPAf6d755Y5AnAv9IU8Q9W1der6i7gdJpiD3BdVX2yvb3fwLK/DbxvYBmfAE5N8pvAUob65+WWtyRpWH9FM0T8jwNt99BuCCYJ8ICBx+4euP2dgfvf4d71Z/bvdM+cKvNFVXWvk3QkOQD4+jz55jq95oLareF1wME0W+G7Ab8M3FVVd7avaT6zc8z5e+NV9dtJ9gN+DrgsyWOr6itLzTrILW9J0lCq6jbg/TQHf824lmaoGOBQYOdlzPqwJPdr94M/HLia5sxav5NkZ4AkP9SeWWwhFwFPS7K+PZjtCOC8IZb/bzRD+OfTbIm/pL2mbXt2ku9tl/+LA4/NXvYBSR7cZj5s5oEkj6iqi6rqj4FbufcpQJfFLW9J0lK8HnjhwP1/AM5I8imas2jNt1W8kKtpiuyewG9X1beSvJVmaP3T7dbvLcCzF5pJVd2Q5ESa/c0Bzq6qMxZ6TusC4OlVtTXJdTRb3xe08/x0klOBT7XTvrWqPpNk0xzLfgXNisANNCMUM0Pkr0uyuc10Ls1Z0lbEs4pJktQzDptLktQzFm9JknrG4i1JUs9YvCVJ6hmLtyRJPWPxliSpZyzekiT1jMVbkqSe+f95+IFiiJHo1AAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 504x288 with 8 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "training_set.hist(column='word_count', by='topic', bins=25, sharey=True, sharex=True, layout=(2, 4), figsize=(7,4), rot=0) \n",
    "plt.suptitle('Word count by topic #')\n",
    "plt.xlabel('Number of words')\n",
    "plt.tight_layout(rect=[0, 0.03, 1, 0.95])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>min</th>\n",
       "      <th>max</th>\n",
       "      <th>count</th>\n",
       "      <th>nunique</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>12</td>\n",
       "      <td>1783</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>1800</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>1726</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>1770</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>1805</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>1800</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>2</td>\n",
       "      <td>24</td>\n",
       "      <td>1569</td>\n",
       "      <td>23</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>10</td>\n",
       "      <td>60</td>\n",
       "      <td>723</td>\n",
       "      <td>34</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       min  max  count  nunique\n",
       "topic                          \n",
       "1        2   12   1783       11\n",
       "2        1    6   1800        6\n",
       "3        0    3   1726        4\n",
       "4        0    3   1770        4\n",
       "5        0    4   1805        5\n",
       "6        0    4   1800        5\n",
       "7        2   24   1569       23\n",
       "8       10   60    723       34"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "training_set.groupby(['topic'])['target_score'].agg(['min','max','count','nunique'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x720 with 8 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "topic_number = 0\n",
    "fig, ax = plt.subplots(4,2, figsize=(8,10))\n",
    "for i in range(4):\n",
    "    for j in range(2):\n",
    "        topic_number += 1\n",
    "        sns.violinplot(x='target_score', y='word_count', data=training_set[training_set['topic'] == topic_number], ax=ax[i,j])\n",
    "        ax[i,j].set_title('Topic %i' % topic_number)\n",
    "ax[3,0].locator_params(nbins=10)\n",
    "ax[3,1].locator_params(nbins=10)\n",
    "plt.suptitle('Word count by score')\n",
    "plt.tight_layout(rect=[0, 0.03, 1, 0.95])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see a reasonable correlation between word count and score for all but topic 8 where the word count apparently reaches a maximum at the upper third of the scores."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 648x648 with 8 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "topic_number = 0\n",
    "fig, ax = plt.subplots(4,2, figsize=(9,9), sharey=False)\n",
    "for i in range(4):\n",
    "    for j in range(2):\n",
    "        topic_number += 1\n",
    "        training_set[training_set['topic'] == topic_number]\\\n",
    "            .groupby('target_score')['essay_id']\\\n",
    "            .agg('count')\\\n",
    "            .plot.bar(ax=ax[i, j], rot=0)\n",
    "        ax[i,j].set_title('Topic %i' % topic_number)\n",
    "ax[3,0].locator_params(nbins=10)\n",
    "ax[3,1].locator_params(nbins=10)\n",
    "plt.suptitle('Histograms of essay scores')\n",
    "plt.tight_layout(rect=[0, 0.03, 1, 0.95])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Many scores are underrepresented. Classification could be difficult without rebalancing."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Essay processing:\n",
    "\n",
    "1. Language correction with languagetool (add number of corrections as feature)\n",
    "2. Sentence tokenization with Spacy\n",
    "3. Sentence count and length\n",
    "4. Word tokenize with Spacy\n",
    "5. Word token classification (punctuation, stop words and anonymized entities, pos, ent)\n",
    "6. Document similarity (based on arbitrary high scoring essay for each topic)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Grammar and spelling\n",
    "\n",
    "As to be expected with student essays, many essays exhibit grammar and spelling errors. \n",
    "\n",
    "Parts-of-speech (POS) and named-entity-recognition (NER) is hampered in part by the lack of consitent spelling and punctuation. Therefore, the essays will be corrected using languagetool and the nlp parsing will be performed with Spacy on the corrected essays."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "text = 'Some people are still using Myspoce instead of facdbook'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Textblob is an open source nlp package and something to keep an eye on. Unlike Spacy, textblob includes a statistics based spell checker. It only claims 70% accuracy, and in my testing it didn't perform well.\n",
    "\n",
    "https://textblob.readthedocs.io/en/dev/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Some people are still using Myspoce instead of facdbook\n"
     ]
    }
   ],
   "source": [
    "from textblob import TextBlob\n",
    "data = TextBlob(text)\n",
    "print (data.correct())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Languagetool.org has a python wrapper for spelling and grammatical errors at \n",
    "https://github.com/myint/language-check\n",
    "\n",
    "It appears to work quite well, although it requires intermediate storage of a list of errors ('matches')."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'Some people are still using MySpace instead of Facebook'"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tool = language_check.LanguageTool('en-US')\n",
    "\n",
    "matches = tool.check(text)\n",
    "language_check.correct(text, matches)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'I do think that there should be a censorship in not just in libaries, but everywhare.       Personlly, I think that the way that the libraries have the books are appropite and if the parents do not want thier children going any where  that is not privy to them keep a hand lenght away     As for the parents, the parents know the aera that intrest them ,therefor the parents should go there'"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "text = training_set.essay[1871]\n",
    "text"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'I do think that there should be a censorship in not just in libraries, but everywhere.  Personally, I think that the way that the libraries have the books are appropriate and if the parents do not want their children going anywhere that is not privy to them keep a hand length away As for the parents, the parents know the area that interest them,therefor the parents should go there'"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "matches = tool.check(text)\n",
    "language_check.correct(text, matches)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As seen in the last sentence, languagetool did not correct `\",therefor\"`. Nonetheless, it should be good enough to proceed."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Processing time: 0:14:07.330476\n"
     ]
    }
   ],
   "source": [
    "\"\"\"\n",
    "use language tool to correct for most spelling and grammatical errors. Also count the applied corrections. \n",
    "Using language_check python wrapper for languagetool:\n",
    "https://www.languagetool.org/dev\n",
    "\"\"\"\n",
    "tool = language_check.LanguageTool('en-US')\n",
    "\n",
    "t0 = datetime.now()\n",
    "\n",
    "training_set['matches'] = training_set['essay'].apply(lambda txt: tool.check(txt))\n",
    "training_set['corrections'] = training_set.apply(lambda l: len(l['matches']), axis=1)\n",
    "training_set['corrected'] = training_set.apply(lambda l: language_check.correct(l['essay'], l['matches']), axis=1)\n",
    "\n",
    "t1 = datetime.now()\n",
    "print('Processing time: {}'.format(t1 - t0))\n",
    "\n",
    "# save work\n",
    "training_set.to_pickle('training_corr.pkl')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here's a very special example of poor writing skills (or perhaps a digitization error?). None of the spell checkers I tried were able to make much sense out of this."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Original:\n",
      "I aegre waf the evansmant ov tnachnolage. The evansmant ov tnachnolige is being to halp fined a kohar froi alnsas. Tnanchnolage waf ont ot we wod not go to the moon. Tnachnologe evans as we maech at. The people are in tnacholege to the frchr fror the good ov live. Famas invanyor ues tnacholage leki lena orde dvanse and his fling mashine. Tnachologe is the grat\n",
      "Corrected with languagetool:\n",
      "I Segre weigh the evansmant of tnachnolage. The evansmant of tnachnolige is being to half fined a coheir from Kansas. Tnanchnolage weigh Ont OT we won not go to the moon. Tnachnologe Evans as we match at. The people are in tnacholege to the arch for the good of live. Lamas inventor us tnacholage Levi Lena order dance and his fling machine. Tnachologe is the great\n"
     ]
    }
   ],
   "source": [
    "print('Original:')\n",
    "print(training_set.essay[18])\n",
    "print('Corrected with languagetool:')\n",
    "print(training_set.corrected[18])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## NLP with SpaCy\n",
    "\n",
    "Although much of the analysis could be performed with other NLP packages, SpaCy was chosen due to its combination of speed and simplicity."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "training_set = pd.read_pickle('training_corr.pkl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Processing time: 0:10:03.031798\n"
     ]
    }
   ],
   "source": [
    "sents = []\n",
    "tokens = []\n",
    "lemma = []\n",
    "pos = []\n",
    "ner = []\n",
    "\n",
    "stop_words = set(STOP_WORDS)\n",
    "stop_words.update(punctuation) # remove it if you need punctuation \n",
    "\n",
    "nlp = spacy.load('en_core_web_sm')\n",
    "\n",
    "t0 = datetime.now()\n",
    "\n",
    "# suppress numpy warnings\n",
    "np.warnings.filterwarnings('ignore')\n",
    "\n",
    "for essay in nlp.pipe(training_set['corrected'], batch_size=100, n_threads=3):\n",
    "    if essay.is_parsed:\n",
    "        tokens.append([e.text for e in essay])\n",
    "        sents.append([sent.string.strip() for sent in essay.sents])\n",
    "        pos.append([e.pos_ for e in essay])\n",
    "        ner.append([e.text for e in essay.ents])\n",
    "        lemma.append([n.lemma_ for n in essay])\n",
    "    else:\n",
    "        # We want to make sure that the lists of parsed results have the\n",
    "        # same number of entries of the original Dataframe, so add some blanks in case the parse fails\n",
    "        tokens.append(None)\n",
    "        lemma.append(None)\n",
    "        pos.append(None)\n",
    "        sents.append(None)\n",
    "        ner.append(None)\n",
    "\n",
    "training_set['tokens'] = tokens\n",
    "training_set['lemma'] = lemma\n",
    "training_set['pos'] = pos\n",
    "training_set['sents'] = sents\n",
    "training_set['ner'] = ner\n",
    "\n",
    "t1 = datetime.now()\n",
    "print('Processing time: {}'.format(t1 - t0))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "training_set.to_pickle('training_spacy.pkl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "training_set = pd.read_pickle('training_spacy.pkl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>tokens</th>\n",
       "      <th>pos</th>\n",
       "      <th>sents</th>\n",
       "      <th>ner</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>[Dear, local, newspaper, ,, I, think, effects, computers, have, on, people, are, great, learning...</td>\n",
       "      <td>[ADJ, ADJ, NOUN, PUNCT, PRON, VERB, NOUN, NOUN, VERB, ADP, NOUN, VERB, ADJ, NOUN, NOUN, SYM, NOU...</td>\n",
       "      <td>[Dear local newspaper, I think effects computers have on people are great learning skills/affect...</td>\n",
       "      <td>[Facebook, MySpace, all day]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>[Dear, @CAPS1, @CAPS2, ,, I, believe, that, using, computers, will, benefit, us, in, many, ways,...</td>\n",
       "      <td>[ADJ, PROPN, PUNCT, PUNCT, PRON, VERB, ADP, VERB, NOUN, VERB, VERB, PRON, ADP, ADJ, NOUN, ADP, V...</td>\n",
       "      <td>[Dear @CAPS1 @CAPS2, I believe that using computers will benefit us in many ways like talking an...</td>\n",
       "      <td>[Facebook, MySpace, millions, one, MySpace, Facebook, one, millions, @MONTH1, @LOCATION2, A Mill...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>[Dear, ,, @CAPS1, @CAPS2, @CAPS3, More, and, more, people, use, computers, ,, but, not, everyone...</td>\n",
       "      <td>[ADJ, PUNCT, PROPN, PUNCT, PROPN, ADJ, CCONJ, ADJ, NOUN, VERB, NOUN, PUNCT, CCONJ, ADV, NOUN, VE...</td>\n",
       "      <td>[Dear, @CAPS1 @CAPS2 @CAPS3 More and more people use computers, but not everyone agrees that thi...</td>\n",
       "      <td>[today, one, @CAPS4, one, @MONTH1, thousand, @ORGANIZATION1]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>[Dear, Local, Newspaper, ,, @CAPS1, I, have, found, that, many, experts, say, that, computers, d...</td>\n",
       "      <td>[ADJ, PROPN, PROPN, PUNCT, PROPN, PRON, VERB, VERB, ADP, ADJ, NOUN, VERB, ADP, NOUN, VERB, ADV, ...</td>\n",
       "      <td>[Dear Local Newspaper, @CAPS1, I have found that many experts say that computers do not benefit ...</td>\n",
       "      <td>[Dear Local Newspaper, @PERSON1, @PERSON2, @CAPS4, @LOCATION2, @PERCENT3, @LOCATION1, A+, Newspa...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>[Dear, @LOCATION1, ,, I, know, having, computers, has, a, positive, effect, on, people, ., The, ...</td>\n",
       "      <td>[ADJ, ADP, PUNCT, PRON, VERB, VERB, NOUN, VERB, DET, ADJ, NOUN, ADP, NOUN, PUNCT, DET, NOUN, VER...</td>\n",
       "      <td>[Dear @LOCATION1, I know having computers has a positive effect on people., The computers connec...</td>\n",
       "      <td>[First, @NUM1 hours, one, Mae, Secondly, @LOCATION2, one, one]</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                                                tokens                                                                                                  pos                                                                                                sents                                                                                                  ner\n",
       "0  [Dear, local, newspaper, ,, I, think, effects, computers, have, on, people, are, great, learning...  [ADJ, ADJ, NOUN, PUNCT, PRON, VERB, NOUN, NOUN, VERB, ADP, NOUN, VERB, ADJ, NOUN, NOUN, SYM, NOU...  [Dear local newspaper, I think effects computers have on people are great learning skills/affect...                                                                         [Facebook, MySpace, all day]\n",
       "1  [Dear, @CAPS1, @CAPS2, ,, I, believe, that, using, computers, will, benefit, us, in, many, ways,...  [ADJ, PROPN, PUNCT, PUNCT, PRON, VERB, ADP, VERB, NOUN, VERB, VERB, PRON, ADP, ADJ, NOUN, ADP, V...  [Dear @CAPS1 @CAPS2, I believe that using computers will benefit us in many ways like talking an...  [Facebook, MySpace, millions, one, MySpace, Facebook, one, millions, @MONTH1, @LOCATION2, A Mill...\n",
       "2  [Dear, ,, @CAPS1, @CAPS2, @CAPS3, More, and, more, people, use, computers, ,, but, not, everyone...  [ADJ, PUNCT, PROPN, PUNCT, PROPN, ADJ, CCONJ, ADJ, NOUN, VERB, NOUN, PUNCT, CCONJ, ADV, NOUN, VE...  [Dear, @CAPS1 @CAPS2 @CAPS3 More and more people use computers, but not everyone agrees that thi...                                         [today, one, @CAPS4, one, @MONTH1, thousand, @ORGANIZATION1]\n",
       "3  [Dear, Local, Newspaper, ,, @CAPS1, I, have, found, that, many, experts, say, that, computers, d...  [ADJ, PROPN, PROPN, PUNCT, PROPN, PRON, VERB, VERB, ADP, ADJ, NOUN, VERB, ADP, NOUN, VERB, ADV, ...  [Dear Local Newspaper, @CAPS1, I have found that many experts say that computers do not benefit ...  [Dear Local Newspaper, @PERSON1, @PERSON2, @CAPS4, @LOCATION2, @PERCENT3, @LOCATION1, A+, Newspa...\n",
       "4  [Dear, @LOCATION1, ,, I, know, having, computers, has, a, positive, effect, on, people, ., The, ...  [ADJ, ADP, PUNCT, PRON, VERB, VERB, NOUN, VERB, DET, ADJ, NOUN, ADP, NOUN, PUNCT, DET, NOUN, VER...  [Dear @LOCATION1, I know having computers has a positive effect on people., The computers connec...                                       [First, @NUM1 hours, one, Mae, Secondly, @LOCATION2, one, one]"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "training_set[['tokens', 'pos', 'sents', 'ner']].head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Topic Modeling with Latent Dirichlet Allocation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Latent Dirichlet Allocation, or **LDA**, uses probabilities to allocate any number of documents to a pre-defined number of topics. A very good explanation is given here:\n",
    "\n",
    "https://tedunderwood.com/2012/04/07/topic-modeling-made-just-simple-enough/\n",
    "\n",
    "The *Hewlett ASAP* essays are already labeled as belonging to one of eight topics. A baseline excercise will determine how well essays are allocated to a topic using LDA.\n",
    "\n",
    "A second experiment will be performed using LDA to assign scores.\n",
    "\n",
    "Another important remark is that LDA is based on probability distributions. Probing these distributions introduces randomness so the results of running this notebook might not exactly match the comments or annotations."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To minimize confusion between the *LDA* derived topics and the *Hewlett ASAP* given topics, the given topic numbers will be replaced with a one-word summary.\n",
    "\n",
    "LDA uses the probability of finding certain words associated with documents. Stop words will not be very helpful, for example, the word \"the\" is going to have a high probability across all topics. In order to refine the word list, we'll also use the lemma generated by SpaCy instead of the regular essay. As a reminder, the lemma were generated on language corrected essays."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Replace topic numbers with meaningful one-word summary:\n",
    "topic_dict = {'topic':{1: 'computer', \n",
    "                       2: 'censorship', \n",
    "                       3: 'cyclist', \n",
    "                       4: 'hibiscus', \n",
    "                       5: 'mood', \n",
    "                       6: 'dirigibles', \n",
    "                       7: 'patience', \n",
    "                       8: 'laughter'}}\n",
    "\n",
    "training_set.replace(topic_dict, inplace=True)\n",
    "\n",
    "# Lemmatized essays re-joined (list to essay)\n",
    "training_set['l_essay'] = training_set['lemma'].apply(' '.join)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Convert essays to a matrix of token (lemma) counts:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "computer     20364\n",
       "book         14001\n",
       "building      7808\n",
       "library       7562\n",
       "dirigible     6275\n",
       "read          6130\n",
       "child         5762\n",
       "help          5203\n",
       "offensive     5115\n",
       "talk          4977\n",
       "dtype: int32"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Baseline: number of unique lemma\n",
    "vectorizer = CountVectorizer(max_df=.2, \n",
    "                             min_df=3, \n",
    "                             stop_words=STOP_WORDS, \n",
    "                             max_features=2000) # default: binary=False\n",
    "doc_term_matrix = vectorizer.fit_transform(training_set.l_essay) # using lemmatized essays\n",
    "\n",
    "# Most frequent tokens:\n",
    "words = vectorizer.get_feature_names()\n",
    "doc_term_matrix_df = pd.SparseDataFrame(doc_term_matrix, columns=words)\n",
    "word_freq = doc_term_matrix_df.sum(axis=0).astype(int)\n",
    "word_freq.sort_values(ascending=False).head(10) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Apply LDA on the word frequency matrix."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "iteration: 1 of max_iter: 40\n",
      "iteration: 2 of max_iter: 40\n",
      "iteration: 3 of max_iter: 40\n",
      "iteration: 4 of max_iter: 40\n",
      "iteration: 5 of max_iter: 40, perplexity: 396.4384\n",
      "iteration: 6 of max_iter: 40\n",
      "iteration: 7 of max_iter: 40\n",
      "iteration: 8 of max_iter: 40\n",
      "iteration: 9 of max_iter: 40\n",
      "iteration: 10 of max_iter: 40, perplexity: 387.6358\n",
      "iteration: 11 of max_iter: 40\n",
      "iteration: 12 of max_iter: 40\n",
      "iteration: 13 of max_iter: 40\n",
      "iteration: 14 of max_iter: 40\n",
      "iteration: 15 of max_iter: 40, perplexity: 387.3376\n",
      "iteration: 16 of max_iter: 40\n",
      "iteration: 17 of max_iter: 40\n",
      "iteration: 18 of max_iter: 40\n",
      "iteration: 19 of max_iter: 40\n",
      "iteration: 20 of max_iter: 40, perplexity: 387.2330\n",
      "iteration: 21 of max_iter: 40\n",
      "iteration: 22 of max_iter: 40\n",
      "iteration: 23 of max_iter: 40\n",
      "iteration: 24 of max_iter: 40\n",
      "iteration: 25 of max_iter: 40, perplexity: 387.1749\n",
      "iteration: 26 of max_iter: 40\n",
      "iteration: 27 of max_iter: 40\n",
      "iteration: 28 of max_iter: 40\n",
      "iteration: 29 of max_iter: 40\n",
      "iteration: 30 of max_iter: 40, perplexity: 387.1528\n",
      "iteration: 31 of max_iter: 40\n",
      "iteration: 32 of max_iter: 40\n",
      "iteration: 33 of max_iter: 40\n",
      "iteration: 34 of max_iter: 40\n",
      "iteration: 35 of max_iter: 40, perplexity: 387.1401\n",
      "iteration: 36 of max_iter: 40\n",
      "iteration: 37 of max_iter: 40\n",
      "iteration: 38 of max_iter: 40\n",
      "iteration: 39 of max_iter: 40\n",
      "iteration: 40 of max_iter: 40, perplexity: 387.1301\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "['lda_baseline.pkl']"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lda_base = LatentDirichletAllocation(n_components=8,\n",
    "                                     n_jobs=-1,\n",
    "                                     learning_method='batch',\n",
    "                                     max_iter=40,\n",
    "                                     perp_tol=0.01,\n",
    "                                     verbose=1,\n",
    "                                     evaluate_every=5)\n",
    "lda_base.fit(doc_term_matrix)\n",
    "\n",
    "# save base model\n",
    "joblib.dump(lda_base, 'lda_baseline.pkl')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Topic probabilities for all words. The numbered topics are generated from the latent Dirichlet allocation. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Topic 1</th>\n",
       "      <th>Topic 2</th>\n",
       "      <th>Topic 3</th>\n",
       "      <th>Topic 4</th>\n",
       "      <th>Topic 5</th>\n",
       "      <th>Topic 6</th>\n",
       "      <th>Topic 7</th>\n",
       "      <th>Topic 8</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>minor</th>\n",
       "      <td>1.507076e-04</td>\n",
       "      <td>0.000003</td>\n",
       "      <td>0.000028</td>\n",
       "      <td>5.680529e-05</td>\n",
       "      <td>0.000067</td>\n",
       "      <td>0.000002</td>\n",
       "      <td>0.000023</td>\n",
       "      <td>0.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>vow</th>\n",
       "      <td>1.685977e-03</td>\n",
       "      <td>0.000003</td>\n",
       "      <td>0.000001</td>\n",
       "      <td>7.512121e-07</td>\n",
       "      <td>0.000002</td>\n",
       "      <td>0.000002</td>\n",
       "      <td>0.000001</td>\n",
       "      <td>0.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>smart</th>\n",
       "      <td>1.792769e-04</td>\n",
       "      <td>0.000448</td>\n",
       "      <td>0.000065</td>\n",
       "      <td>1.503492e-04</td>\n",
       "      <td>0.000374</td>\n",
       "      <td>0.000002</td>\n",
       "      <td>0.000659</td>\n",
       "      <td>0.000093</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>bear</th>\n",
       "      <td>5.742768e-05</td>\n",
       "      <td>0.000842</td>\n",
       "      <td>0.000032</td>\n",
       "      <td>3.094863e-04</td>\n",
       "      <td>0.000002</td>\n",
       "      <td>0.001499</td>\n",
       "      <td>0.000265</td>\n",
       "      <td>0.000020</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>engine</th>\n",
       "      <td>5.606097e-07</td>\n",
       "      <td>0.000003</td>\n",
       "      <td>0.000001</td>\n",
       "      <td>5.825088e-05</td>\n",
       "      <td>0.000002</td>\n",
       "      <td>0.000002</td>\n",
       "      <td>0.000785</td>\n",
       "      <td>0.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>somebody</th>\n",
       "      <td>5.488834e-04</td>\n",
       "      <td>0.000003</td>\n",
       "      <td>0.000020</td>\n",
       "      <td>3.873019e-04</td>\n",
       "      <td>0.000207</td>\n",
       "      <td>0.000018</td>\n",
       "      <td>0.000317</td>\n",
       "      <td>0.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>tomorrow</th>\n",
       "      <td>5.609983e-07</td>\n",
       "      <td>0.000388</td>\n",
       "      <td>0.000001</td>\n",
       "      <td>3.139983e-04</td>\n",
       "      <td>0.000002</td>\n",
       "      <td>0.000002</td>\n",
       "      <td>0.000211</td>\n",
       "      <td>0.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>object</th>\n",
       "      <td>2.771226e-04</td>\n",
       "      <td>0.000317</td>\n",
       "      <td>0.000151</td>\n",
       "      <td>2.039656e-05</td>\n",
       "      <td>0.000040</td>\n",
       "      <td>0.000002</td>\n",
       "      <td>0.000121</td>\n",
       "      <td>0.000064</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>proper</th>\n",
       "      <td>1.093197e-04</td>\n",
       "      <td>0.000003</td>\n",
       "      <td>0.000090</td>\n",
       "      <td>1.203972e-05</td>\n",
       "      <td>0.000211</td>\n",
       "      <td>0.000018</td>\n",
       "      <td>0.000057</td>\n",
       "      <td>0.000087</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>apart</th>\n",
       "      <td>9.725760e-05</td>\n",
       "      <td>0.000160</td>\n",
       "      <td>0.000056</td>\n",
       "      <td>2.862382e-04</td>\n",
       "      <td>0.000358</td>\n",
       "      <td>0.000034</td>\n",
       "      <td>0.000086</td>\n",
       "      <td>0.000117</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               Topic 1   Topic 2   Topic 3       Topic 4   Topic 5   Topic 6   Topic 7   Topic 8\n",
       "minor     1.507076e-04  0.000003  0.000028  5.680529e-05  0.000067  0.000002  0.000023  0.000002\n",
       "vow       1.685977e-03  0.000003  0.000001  7.512121e-07  0.000002  0.000002  0.000001  0.000002\n",
       "smart     1.792769e-04  0.000448  0.000065  1.503492e-04  0.000374  0.000002  0.000659  0.000093\n",
       "bear      5.742768e-05  0.000842  0.000032  3.094863e-04  0.000002  0.001499  0.000265  0.000020\n",
       "engine    5.606097e-07  0.000003  0.000001  5.825088e-05  0.000002  0.000002  0.000785  0.000002\n",
       "somebody  5.488834e-04  0.000003  0.000020  3.873019e-04  0.000207  0.000018  0.000317  0.000002\n",
       "tomorrow  5.609983e-07  0.000388  0.000001  3.139983e-04  0.000002  0.000002  0.000211  0.000002\n",
       "object    2.771226e-04  0.000317  0.000151  2.039656e-05  0.000040  0.000002  0.000121  0.000064\n",
       "proper    1.093197e-04  0.000003  0.000090  1.203972e-05  0.000211  0.000018  0.000057  0.000087\n",
       "apart     9.725760e-05  0.000160  0.000056  2.862382e-04  0.000358  0.000034  0.000086  0.000117"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "topic_labels = ['Topic {}'.format(i) for i in range(1, 9)]\n",
    "topics_count = lda_base.components_\n",
    "topics_prob = topics_count / topics_count.sum(axis=1).reshape(-1, 1)\n",
    "topics = pd.DataFrame(topics_prob.T,\n",
    "                      index=words,\n",
    "                      columns=topic_labels)\n",
    "topics.sample(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "one_word = list(topic_dict['topic'].values())\n",
    "sns.heatmap(topics.reindex(one_word), cmap='Blues')\n",
    "plt.title('Topic probabilities for one-word-summary')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The heatmap suggests assignments for all but 1 topics. \n",
    "\n",
    "Below are the most probable words for each topic. We can already see our one-word summaries of the actual topic near the top of the list. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Topic 1</th>\n",
       "      <th>Topic 2</th>\n",
       "      <th>Topic 3</th>\n",
       "      <th>Topic 4</th>\n",
       "      <th>Topic 5</th>\n",
       "      <th>Topic 6</th>\n",
       "      <th>Topic 7</th>\n",
       "      <th>Topic 8</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>book</td>\n",
       "      <td>caps3</td>\n",
       "      <td>building</td>\n",
       "      <td>laugh</td>\n",
       "      <td>computer</td>\n",
       "      <td>mood</td>\n",
       "      <td>computer</td>\n",
       "      <td>cyclist</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>library</td>\n",
       "      <td>caps4</td>\n",
       "      <td>dirigible</td>\n",
       "      <td>patient</td>\n",
       "      <td>spend</td>\n",
       "      <td>memoir</td>\n",
       "      <td>help</td>\n",
       "      <td>water</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>read</td>\n",
       "      <td>caps5</td>\n",
       "      <td>obstacle</td>\n",
       "      <td>wait</td>\n",
       "      <td>exercise</td>\n",
       "      <td>love</td>\n",
       "      <td>learn</td>\n",
       "      <td>setting</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>offensive</td>\n",
       "      <td>caps6</td>\n",
       "      <td>empire</td>\n",
       "      <td>laughter</td>\n",
       "      <td>kid</td>\n",
       "      <td>narcs</td>\n",
       "      <td>talk</td>\n",
       "      <td>affect</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>child</td>\n",
       "      <td>person1</td>\n",
       "      <td>mast</td>\n",
       "      <td>tell</td>\n",
       "      <td>bad</td>\n",
       "      <td>create</td>\n",
       "      <td>information</td>\n",
       "      <td>feature</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>movie</td>\n",
       "      <td>person2</td>\n",
       "      <td>builder</td>\n",
       "      <td>mom</td>\n",
       "      <td>outside</td>\n",
       "      <td>grateful</td>\n",
       "      <td>online</td>\n",
       "      <td>road</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>shelf</td>\n",
       "      <td>caps7</td>\n",
       "      <td>face</td>\n",
       "      <td>start</td>\n",
       "      <td>nature</td>\n",
       "      <td>happy</td>\n",
       "      <td>hand</td>\n",
       "      <td>hill</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>music</td>\n",
       "      <td>organization1</td>\n",
       "      <td>dock</td>\n",
       "      <td>person1</td>\n",
       "      <td>play</td>\n",
       "      <td>rodriguez</td>\n",
       "      <td>place</td>\n",
       "      <td>desert</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>material</td>\n",
       "      <td>caps8</td>\n",
       "      <td>mooring</td>\n",
       "      <td>patience</td>\n",
       "      <td>reason</td>\n",
       "      <td>house</td>\n",
       "      <td>world</td>\n",
       "      <td>hot</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>magazine</td>\n",
       "      <td>caps9</td>\n",
       "      <td>wind</td>\n",
       "      <td>feel</td>\n",
       "      <td>child</td>\n",
       "      <td>paragraph</td>\n",
       "      <td>reason</td>\n",
       "      <td>town</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     Topic 1        Topic 2    Topic 3   Topic 4   Topic 5    Topic 6      Topic 7  Topic 8\n",
       "0       book          caps3   building     laugh  computer       mood     computer  cyclist\n",
       "1    library          caps4  dirigible   patient     spend     memoir         help    water\n",
       "2       read          caps5   obstacle      wait  exercise       love        learn  setting\n",
       "3  offensive          caps6     empire  laughter       kid      narcs         talk   affect\n",
       "4      child        person1       mast      tell       bad     create  information  feature\n",
       "5      movie        person2    builder       mom   outside   grateful       online     road\n",
       "6      shelf          caps7       face     start    nature      happy         hand     hill\n",
       "7      music  organization1       dock   person1      play  rodriguez        place   desert\n",
       "8   material          caps8    mooring  patience    reason      house        world      hot\n",
       "9   magazine          caps9       wind      feel     child  paragraph       reason     town"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "top_words = {}\n",
    "for topic, words_ in topics.items():\n",
    "    top_words[topic] = words_.nlargest(10).index.tolist()\n",
    "pd.DataFrame(top_words)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we can assign LDA topic probabilities to each essay and aggregate. It is now clear that, for example, LDA allocated topic 3 is aligned with the given topic \"computers\"."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Topic 1</th>\n",
       "      <th>Topic 2</th>\n",
       "      <th>Topic 3</th>\n",
       "      <th>Topic 4</th>\n",
       "      <th>Topic 5</th>\n",
       "      <th>Topic 6</th>\n",
       "      <th>Topic 7</th>\n",
       "      <th>Topic 8</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>computer</th>\n",
       "      <td>0.001149</td>\n",
       "      <td>0.001147</td>\n",
       "      <td>0.001148</td>\n",
       "      <td>0.001148</td>\n",
       "      <td>0.440254</td>\n",
       "      <td>0.001149</td>\n",
       "      <td>0.552857</td>\n",
       "      <td>0.001148</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dirigibles</th>\n",
       "      <td>0.002605</td>\n",
       "      <td>0.002606</td>\n",
       "      <td>0.981761</td>\n",
       "      <td>0.002605</td>\n",
       "      <td>0.002606</td>\n",
       "      <td>0.002605</td>\n",
       "      <td>0.002605</td>\n",
       "      <td>0.002607</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mood</th>\n",
       "      <td>0.002552</td>\n",
       "      <td>0.002552</td>\n",
       "      <td>0.002552</td>\n",
       "      <td>0.002553</td>\n",
       "      <td>0.002553</td>\n",
       "      <td>0.982134</td>\n",
       "      <td>0.002553</td>\n",
       "      <td>0.002551</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mood</th>\n",
       "      <td>0.002051</td>\n",
       "      <td>0.002050</td>\n",
       "      <td>0.002050</td>\n",
       "      <td>0.002052</td>\n",
       "      <td>0.002052</td>\n",
       "      <td>0.985644</td>\n",
       "      <td>0.002051</td>\n",
       "      <td>0.002051</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cyclist</th>\n",
       "      <td>0.005209</td>\n",
       "      <td>0.005210</td>\n",
       "      <td>0.005213</td>\n",
       "      <td>0.005226</td>\n",
       "      <td>0.005210</td>\n",
       "      <td>0.005209</td>\n",
       "      <td>0.005213</td>\n",
       "      <td>0.963509</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>censorship</th>\n",
       "      <td>0.773385</td>\n",
       "      <td>0.048939</td>\n",
       "      <td>0.000913</td>\n",
       "      <td>0.046385</td>\n",
       "      <td>0.033712</td>\n",
       "      <td>0.000915</td>\n",
       "      <td>0.094839</td>\n",
       "      <td>0.000913</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>computer</th>\n",
       "      <td>0.001171</td>\n",
       "      <td>0.117966</td>\n",
       "      <td>0.001170</td>\n",
       "      <td>0.001170</td>\n",
       "      <td>0.133341</td>\n",
       "      <td>0.001169</td>\n",
       "      <td>0.742844</td>\n",
       "      <td>0.001170</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>computer</th>\n",
       "      <td>0.096735</td>\n",
       "      <td>0.067008</td>\n",
       "      <td>0.000969</td>\n",
       "      <td>0.000971</td>\n",
       "      <td>0.000971</td>\n",
       "      <td>0.000971</td>\n",
       "      <td>0.831404</td>\n",
       "      <td>0.000970</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>patience</th>\n",
       "      <td>0.005685</td>\n",
       "      <td>0.005684</td>\n",
       "      <td>0.005683</td>\n",
       "      <td>0.960209</td>\n",
       "      <td>0.005685</td>\n",
       "      <td>0.005685</td>\n",
       "      <td>0.005684</td>\n",
       "      <td>0.005684</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hibiscus</th>\n",
       "      <td>0.719268</td>\n",
       "      <td>0.008933</td>\n",
       "      <td>0.008931</td>\n",
       "      <td>0.008937</td>\n",
       "      <td>0.008946</td>\n",
       "      <td>0.227115</td>\n",
       "      <td>0.008937</td>\n",
       "      <td>0.008934</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             Topic 1   Topic 2   Topic 3   Topic 4   Topic 5   Topic 6   Topic 7   Topic 8\n",
       "topic                                                                                     \n",
       "computer    0.001149  0.001147  0.001148  0.001148  0.440254  0.001149  0.552857  0.001148\n",
       "dirigibles  0.002605  0.002606  0.981761  0.002605  0.002606  0.002605  0.002605  0.002607\n",
       "mood        0.002552  0.002552  0.002552  0.002553  0.002553  0.982134  0.002553  0.002551\n",
       "mood        0.002051  0.002050  0.002050  0.002052  0.002052  0.985644  0.002051  0.002051\n",
       "cyclist     0.005209  0.005210  0.005213  0.005226  0.005210  0.005209  0.005213  0.963509\n",
       "censorship  0.773385  0.048939  0.000913  0.046385  0.033712  0.000915  0.094839  0.000913\n",
       "computer    0.001171  0.117966  0.001170  0.001170  0.133341  0.001169  0.742844  0.001170\n",
       "computer    0.096735  0.067008  0.000969  0.000971  0.000971  0.000971  0.831404  0.000970\n",
       "patience    0.005685  0.005684  0.005683  0.960209  0.005685  0.005685  0.005684  0.005684\n",
       "hibiscus    0.719268  0.008933  0.008931  0.008937  0.008946  0.227115  0.008937  0.008934"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_preds = lda_base.transform(doc_term_matrix)\n",
    "train_eval = pd.DataFrame(train_preds, columns=topic_labels, index=training_set.topic)\n",
    "train_eval.sample(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "train_eval.groupby(level='topic').mean().plot.bar(title='Avg. Topic Probabilities', rot=0, colormap='tab10', figsize=(10,5));"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The baseline model was successful in that each given topic is allocated with high probability to an LDA topic:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>topic</th>\n",
       "      <th>assignment</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>censorship</td>\n",
       "      <td>Topic 1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>computer</td>\n",
       "      <td>Topic 7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>cyclist</td>\n",
       "      <td>Topic 8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>dirigibles</td>\n",
       "      <td>Topic 3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>hibiscus</td>\n",
       "      <td>Topic 1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>laughter</td>\n",
       "      <td>Topic 4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>mood</td>\n",
       "      <td>Topic 6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>patience</td>\n",
       "      <td>Topic 4</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        topic assignment\n",
       "0  censorship    Topic 1\n",
       "1    computer    Topic 7\n",
       "2     cyclist    Topic 8\n",
       "3  dirigibles    Topic 3\n",
       "4    hibiscus    Topic 1\n",
       "5    laughter    Topic 4\n",
       "6        mood    Topic 6\n",
       "7    patience    Topic 4"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAgEAAAHICAYAAADX6TmAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAIABJREFUeJzs3XmczWX/x/HXZ2YMSpbsMZaoiLY7aaXSrZQtRUSLFrIkLZRSKtWvjbTYK5HQohQq3SXtiyVFcetG9jXJUjPM8vn9cY5pzAzmMMeZc+b9fDzOo3O+3+v7Pe9zdca5zue6zjnm7oiIiEjhExfpACIiIhIZGgSIiIgUUhoEiIiIFFIaBIiIiBRSGgSIiIgUUhoEiIiIFFIaBIiIiBRSGgSIiIgUUhoEiIiIFFIJkQ4gIiISTYqfdmtYvmo3ef5QC8d590eVABERkUJKgwAREZFCStMBIiIiobDYef+sQYCIiEgo7LBP3YdN7AxnREREJCSqBIiIiIQihqYDYueRiIiISEhUCRAREQlFDK0J0CBAREQkFJoOEBERkWinSoCIiEgoYmg6QJUAERGRQkqVABERkVBoTYCIiIhEO1UCREREQhFDawI0CBAREQmFpgNEREQk2qkSICIiEooYmg5QJUBERKSQUiVAREQkFDG0JkCDABERkVBoOkBERESinSoBIiIioYih6YDYeSQiUc7MxprZ3ZHOAWBmy8zs7EjnEJHwUiVAJARmtjPLzSOAXUB68PYt7j7hYM/t7p0PIRpm9h1QE6ji7mmHci53r3Uox0eCmT0BlHP3myOdRWJcDFUCNAgQCYG7l9hz3cxWADe7+yeRS5SZ5QTgDGA7cCkwLbKJRGJYnBYGikguzKy4mQ0zs/VmtsbMnjazIsF9zcxsqZk9bGZ/mNlyM2uX5djXzez+LLfbmdkCM9tuZv8zs4v2c9fXA58Bk4LXs2ZqbWb/NbMdZrbazG4Lbq9kZjPM7E8z22Jmn2Y5ZoOZnRe8XsLMJgbb/Wxm95rZ0mxt7wju22ZmE8wsMdtjvt/MfjeztWZ2WTDTsuD93pXlXPFm9kCwb34Pnqt0cF8dM0szsxuCfbvZzPoG910O3Alcb2Y7zWx2cHsXM1sRfOx79beIqBIgkt8eBk4GTgLigenA3cBjwf01gESgEtAYmGpmc939t6wnMbNGwGjgCuBzoCpQLLc7NLM44BrgIWAJMNPMyrj71mCTMcCl7j7bzMoC1YLb7wm2b0HgDcFZ+3hMjwLlgepAaeDDXNq0BS4CMoBvgY7A2OC+6kBq8DF3D+b5ONhPxwPfmtnr7r4W6AtcDJwH/AGMBIYANwTPFQ80AGoT6OOvzextd3/XzJ4hy3SAmZUBngZOd/dlZlYFKLmPxyiSdzE0HRA7j0SkYOgEPOjuv7v7RgIvoNdm2Z8GPOzuu4PTCJ8QeAHN7mZgpLvPcvcMd1/l7r/u4z6bABWAt4FvgPVAh2z3Wc/MjnL3Le4+P7g9FTgGqBbM88U+zn8V8Ki7b3P3lcDwXNoMcfeN7r4Z+AA4Ncu+v4Gng+sUXgcqAoPc/a9glmUEXtABbgH6ufs6d08hMKhqb7bXB7MfdPcUd58D/JfAYGJ/6ptZMXdf6+6LD9BWpFDRIEAknwRfqCoBK7NsXglUyXJ7c/DFLev+Y3I5XRKBF8e8uB54P/gi7eScErgcuBJYZWafmlmD4PbHgHXArGDJ/s59PKaKwOosm1dnbwdsyHL9b6BEltub3T0jeD05+N+NWfYnAyWC95UEfBCcevgTmE/g36mywbbp7v77fu4rU7AS0gm4DdhgZlPNrHZubUVCYhaeSwRoECCST4IvwBsIlL/3qAaszXK7nJkVy7Z/XS6nWw0ccIW+mZUgMGVwcXBufgOBkvuZwcWCuPu37t6CwIv5fwgMEggOGnq7e3UCg4T7zezcXB7TJgLTEXskHSjXwQje11qgibuXznIplu2Ff5+nyOWc77v7RQQGWquAEfmbWgoliwvPJQI0CBDJX5OAB82srJlVAPoDr2XZXwR4wMwSzawJ0JRAGT+7l4BbzKyxmcWZWZKZHZ9Lu3bATqAOgRL8qUBdYDZwnZkdaWYdzKwkgfL/DoIfaTSzVmZWM/gOfFtwe3ou9/Em0N/MSplZNQKDjHAZCTxhZknBjBXMrGUej90I7Hk8mFkVM2tuZns+yrmT3B+fSKGlQYBI/hoALAJ+AX4EvgaeyrJ/BYE5+g0EFsjd4O7Ls5/E3b8EuhGYf98GzGTvd+N7XA+8FJzv3rDnAgzjn7UINxKYdtgGXMc/UwV1CXyiYAfwBYF5+u9yuY/7ga3Bc3xIYFCw6wD9cLCeIrBO4lMz20FgjcO/8njs6wS+u+EPM/uGwCLCewn09RYCH6Hsle+JpfCJoekAC1TgRCTczKwZMNTdo3pe2szuAJq5+yWRziISCcWbPhmWF87kj+857CMBfURQRPYrWJqvQmCKoS7QG3gioqFEIimGPiKoQYCIHEhRAlMX1QlMC7xGYM2CiEQ5DQJEDhN3n0HgS26iirsvBU6MdA6RAiNC8/fhoEGAiIhIKDQdINkVbzQgqlZYbp01MNIRRETCplgCsfN2PYw0CBAREQlFDE0HxE5NQ0REREKiSoCIiEgotCZARESkkNJ0gIiIiEQ7VQJERERCEUPTAbHzSERERCQkqgSIiIiEIoYqARoEiIiIhEILA0VERCTaqRIgIiISihiaDoidRyIiIiIhUSVAREQkFFoTICIiItFOlQAREZFQxNCaAA0CREREQqHpABEREYl2GgQcJj3bnsXccT2Z9+qt3NrubAAG3NSE2WN78N2Y7kwbfB2Vyx6V67E7P3uI78Z057sx3Xnr8Y6Z21954Epmj+3Bw13/nbmt3/Xn0+K8Ovme/+svv6BV80to0awpL784Osf+V8e+QpuWl9G2TUu63Hg969atBWDFb8vp0O4K2rVpxU8/zgcgLS2Nrjd1Jjk5Od9zhpJ53tw5tG/bhn+dfCIffzQjc3tBzlzQ+jkW+7igZY6250ReMhe0Pg6VmYXlEgmFbhBgZg+ZWZ997PsmHPd5Ys0K3NDydBp1HU3DG4Zz6TnHU6vq0QyZ9DUNOw/nrBtH8OE3S7i38wW5Hp+8K5WzbhzBWTeOoN29EwGoX6siAA07D+fck6tT8siiVCpbggZ1qzL9q//ma/709HT+77GBDB/5ElOmvs+MD6azbOnSvdrUqVuXiW++zeQp02h68SUMGfw0AG+9+Qa977iLQc8+z7hXxgDw5huTaNGyNcWLF8/XnKFmrlS5Mo889jiXNm+x1/aCnLkg9XOs9nFByhxtz4m8Zi5IfVzYxewgwMziQz3G3c8JR5Y61csze9Eaknelkp6ewZc/rqB14xPZ8feuzDZHFE/E8TyfMzUtneJFi2BmJBaJJz3DeeCmJjzy8sx8z//zwgUkJVWnalISRRITaXZZcz6btff9NDzzrMw/0pNOOZVNGzYAkJCQwK6UFFKSU0goksD27dv5/LNZtGx9eb7nDDVzlSpVOf6EOsRlW+RTkDMXpH6O1T4uSJmj7TmR18wFqY8PRixVAsK6MNDMrgP6AA4sAO4ERgLVgk1ud/evzeyh4LZjg/991t2fN7MjgTeBqkA88Ii7v2FmFwGDgvnnAN3dfZeZrQDGABcDQ82sAtANSAMWuXuH4P2eaGafZb2vYN6d7l7CzC4ABgJbgBOAL4Ae7p5xMP3wy28beajrRRxdsjjJu9Jodtbx/LAkULJ7qMtFdLrkVLb9lUKz3q/kenyxxAS+evEW0tMzGDThS6Z9+V+WrPyd1Ru38e3L3Zj00U/UqnI0ZsZP/9twMBH3a9PGjVSqXCnzdoWKFVm4YME+2095ezLnNmoMQIerO3H/ffewe/duHnjwYUaNGEaXrt3C/oQPNXNW0ZI50v1cGPo4q2jo40g/Jw4mc1aRyhyyAhbnUIRtEGBm9YD+wLnu/ruZHQ0MBYa4+1dmVg34CKgbPKQOcCFwFLDEzEYAzYB17t48eM5SZlYMGAtc5O6/mtmrQHfg2eB5Utz9vGD7dUDN4AChdJZ4Oe7L3VOzPYSGwInASmAGcAUw+WD6YsnK3xk84SumD7mev/7ezYKlG0hLD4wnHnpxJg+9OJM+1zSi2xVn8uiYWTmOP77tM6zfsoMalcsw47nO/LxsI7+t20rfFz7MbDP5iU70enoqd1/bmJNrV2Lm3GW8Mm3ewcTNIbcKxb7+KKdPe49Fv/zMmHGvAVD5mGN4eex4AFatXMnmzZuoWfNY7uvXl9TUVHr26k2NGjXzJefBZs4uGjIXhH6O9T7OrqD3cUF4ToSaObtIZS7Mwjkd0ASY7O6/A7j7H8C/CbxD/xGYCpQ0sz2r4d53913B9puAisBC4N9m9qSZNXL3bQTemf/m7r8GjxsHNM5yv29kub4AmGBm1xCoBuyR231lN9vdl7t7OjAJOC97AzPramZzzWxu2oYf9tsZ497/gXNuGknTXmPYuiOZpau37LX/zY8XcPn5J+Z67PotOwBYsX4rX/y4glOPr7zX/hbn1eGH/67liOJFqHdsBa558E06XnwKxYsW2W+mvKpYsRIb1v9TYdi0cSMVKlTI0e67b7/hpdEjeW7oCBITE3Psf+H5IfTs1ZuJE8bTvHlLevTsxajhQ/Ml48FmPpCCmLmg9HMs9/GBFLQ+LijPiVAyH8jhzByqWJoOCOcgwCDHkDAOONvdTw1eqrj7juC+XVnapQMJwRf60wkMBh43swEcuBDzV5brzYFhwXPMM7M9lY8c95XLebJnzzG8dffR7t7A3RskVPrXfkOVL30kAEkVStG6cV3e/GQhtaoe/U/Q8+rw66rfcxxXukQxEosEljeULXUEZ9evxuIVmzP3J8TH0bPtWQyZ9DVHFC2CB1PGxVnmcYeqXv2TWLVqBWvWrCZ1925mfPA+51/YZK82ixcv4pGHB/Dc0BGULVs2xznmzplNhQoVqV69BinJyVhcHHHx8SSnpORLxoPJfCAFMXNB6udY7eOClDnanhN5zXwghztzYRbONQEzgSlmNsTdtwSnA/4D3Ao8DWBmp7r7j/s6gZkdA/zh7q+Z2U6gM/AUUMPMarv7UuBa4PNcjo0Dktx9lpl9BXQESoSQv6GZ1SQwHdAeyPk5lxBMerQDR5cqTmpaBrcPeZ8/d6Yw4p7WHFetHBnurNqwjdsGTQXgXyccw82Xn0GPJ9+jTo3yvNCnFRnuxJkxaMKX/DfLIKDbFWfy2owfSd6VysJlGzGDOWN78tF3v7JtZ/78wSQkJHBv/wF073ozGRnpXN7mSmrXPo5hLzxHvXr1uaDJRQwZ9BR///03fe/oDQRW/z4/bCQA7s6Lo0bw9ODAjM2V7dpz7z19SE9Po/8DD+VLxoPJ/PPCBdzR+9bMxUfDh73AlKnvF+jMBamfY7WPC1LmaHtO5DVzQerjg1Hg1igcAnPP+4r0kE9udj3Ql8C77fkEFgkOI7AOIAH4wt27BRcG7nT3QcHjfgZaECj9Pw1kAKkEFgDOPcDCwAbBNQhFgFlAKQLVg9fc/Yl93Ze7r8i2MHAAsBk4iTwsDCzeaED4OjIMts4aGOkIIiJhUywhfMv3SnZ4NSz/3m9//brDProI66cD3H0cgTn7rNrn0u6hbLfrB6+uILB4MHv7mcBpuWyvkeV6KrnM4+/nvnD3rJWCv909R1YREZFYod8OEBERCUEsTQdoEJALd/8M+CzCMURERMJKgwAREZFQxE4hIHa/NlhERET2T5UAERGREGhNgIiISCEVS4MATQeIiIgUUqoEiIiIhECVABEREYl6qgSIiIiEIJYqARoEiIiIhCJ2xgCaDhAREYkGZtbMzJaY2VIz65fL/mpmNsvM5pvZAjO77EDnVCVAREQkBJGYDjCzeAK/wtsUWAPMMbOp7r4oS7P7gTfdfYSZnQh8ANTY33lVCRARESn4GgJL3X25u+8GXgdaZ2vjQMng9VLAugOdVJUAERGREISrEmBmXYGuWTaNdvfRwetVgNVZ9q0Bzsx2ioeA/5hZL+BI4N8Huk8NAkREREIQrkFA8AV/9D5253annu321cBYdx9sZmcD482svrtn7Os+NQjIJ1tnDYx0BJFDtmjt9khHCNmJVUoeuFEBc/Zjn0Y6Qki+7d8k0hEk8M4/KcvtquQs998ENANw92/NrBhQDti0r5NqTYCIiEgoLEyX/ZsDHGdmNc0sEegATM3WZhVwEYCZ1QWKAZv3d1INAkRERAo4d08DbgU+AhYT+BTAL2Y20MxaBZvdBXQxs5+ASUBnd88+ZbAXTQeIiIiEIFLfGOjuHxD42F/WbQOyXF8EnBvKOVUJEBERKaRUCRAREQmBfjtARESkkIqlQYCmA0RERAopVQJERERCoEqAiIiIRD1VAkREREIRO4UADQJERERCoekAERERiXqqBIiIiIRAlQARERGJeqoEiIiIhCCWKgEaBIiIiIQidsYAmg4QEREprFQJEBERCUEsTQeoEhAhX3/5Ba2aX0KLZk15+cXROfbPmzuH9m3b8K+TT+Tjj2Zkbl/x23I6tLuCdm1a8dOP8wFIS0uj602dSU5OjljeV8e+QpuWl9G2TUu63Hg969atjWjevGQuaH0cDZm3bNrAI327cddN7ejT5So+nDIJgO+++IQ+Xa6i4yUNWfbron0eP3LwQG5pdzF9u7Tfa/vEl17g7luuZvhTD2Zu+/KTDzLPn58Keh8DlCiawNPt6vNOzzN5u8eZnFy1JMdXLMG4m07n9VvOYEKXBtQ75qh9Hn9kYjwf3XEu91x6PABF4o2hnU7hre4NadegSma7+1ucwAmVSuRrdoiOPpaAQjkIMLNTzeyySN1/eno6//fYQIaPfIkpU99nxgfTWbZ06V5tKlWuzCOPPc6lzVvstf2tN9+g9x13MejZ5xn3yhgA3nxjEi1atqZ48eIRy1unbl0mvvk2k6dMo+nFlzBk8NMRy5vXzAWpj6Mlc1x8Atd0vZ3BL7/FI8+9wn+mTmbNyuUk1ajFnQOeos5Jp+33+PObtqDf/z2/17a//9rJr4sW8NSoSWSkp7Pqt6Xs3pXC5/+ZRtOW7fItO0RHHwPc3ew4vlm6hSuGfU/7kbNZvvlvbm9am9Gf/0aHUXMYMes3bm9ae5/H92hyLPNWbs28fU6tsixev4OrRszmytOPAeD4iiWIM2PJhp35mj1a+vhQmFlYLpFQKAcBwKlASIMAM8u3qZOfFy4gKak6VZOSKJKYSLPLmvPZrJl7talSpSrHn1CHONv7f1FCQgK7UlJISU4hoUgC27dv5/PPZtGy9eX5Fe+g8jY886zMP9KTTjmVTRs2RCxvXjMXpD6Olsxlypaj5nF1ACh+xJFUqVaDP37fTJVqNTkmqcYBj6978r8ocVTJvbaZGWmpqbg7u3fvIj4+gWlvjafZ5R1ISMjfGcto6OMjE+P5V/XSTJm/HoC0DGfnrjTcnSOLBvqjRLEENu/YlevxdSsfRdkjE/l22R+Z29IyMiiWEEd83D8vND0urMmIWcvzNTtERx/LPyKyJsDMrgP6AA4sAO4HxgDlgc3ADe6+yszGAslAHaA6cANwPXA28L27dw6ebycwCrgQ2Ap0cPfNZvYZ0Mfd55pZOWAucDwwEChuZucBjwPTgReAkwj0yUPu/p6ZdQaaA8WAI4Em+fH4N23cSKXKlTJvV6hYkYULFuTp2A5Xd+L+++5h9+7dPPDgw4waMYwuXbuFdRQZat4pb0/m3EaNI5b3YDJnpcx5s3nDOlYsXULtOvUO6TzFjziSho2acG/3TtQ77QyOOLIEy5cs4spruuRT0n9EQx9XKVOcrX+n8nDruhxfsQSL1+/gqRm/Muij/zHsmlO5o2lt4szoPGZejmMNuPPi2tw/ZRENa5bJ3P7dsq00P7kS429uwNivV3H+8eVYtH4Hm3fuztfsEB19fKgKWp5DcdgHAWZWD+gPnOvuv5vZ0cA44FV3H2dmNwLPA3uGfmUIvPi2AqYB5wI3A3PM7FR3/5HAC/QP7n6XmQ0AHgRuze3+3X13sE0Dd781mOn/gE/d/UYzKw3MNrNPgoecDZzs7n9kP5eZdQW6AgwdPoqbunTNUx84nlu/5OnYysccw8tjxwOwauVKNm/eRM2ax3Jfv76kpqbSs1dvatSomadz5VUoeadPe49Fv/zMmHGvRSxvqJmzU+YDS0n+myED7+G67ndyxJGHPqfc6qrraHXVdQCMfuZR2l7fjU8/fJcF876nWs3aXNHppkO+D4iOPk6IM+pULsGTH/7Kz2u307fZcdx4XnVKFE1g8Ef/Y+bizTQ9sQIPtqpDt/E/7nXsVWdU4av/bWHj9r2rBOnu3PfOoszzD7vmVG6ftIC7Lq5NpVLFmP7TBj7/9fdDzg7R0ceHKpYGAZGYDmgCTHb33wGCL65nAxOD+8cD52VpP83dHVgIbHT3he6eAfwC1Ai2yQDeCF5/LdvxeXEx0M/MfgQ+I/DOv1pw38e5DQCC2Ue7ewN3b5DXAQBAxYqV2LB+Q+btTRs3UqFChRAjwwvPD6Fnr95MnDCe5s1b0qNnL0YNHxryeQ4kr3m/+/YbXho9kueGjiAxMTFieUPJfCDKnFNaWhpDBt7DuU2a0fC8fCmOZfpt6RIAKlepxpcff8Dt9z/OmhXLWL92Vb6cPxr6eOP2XWzavouf124H4JNFm6hT6ShanFKZmYs3A/Dxok3Uq1Iyx7EnVy1F+4ZVeb/32dxxcW1anFKJ2y6qtVebdmdUYfpP6zk5qSSp6c49k3/m5sY18iU7REcfyz8iMQgwyGWouLes+/cMaTOyXN9ze1+VjD3Hp/HPYyx2gExXuvupwUs1d18c3PfXAbKGrF79k1i1agVr1qwmdfduZnzwPudfGNo/pnPnzKZChYpUr16DlORkLC6OuPh4klNS8jtunvIuXryIRx4ewHNDR1C2bNmI5s1r5gNR5pzcndHPPMIx1WrQvG2nfDlnVm+NG0nb628hPT2NjIx0ACwujt35lD8a+njLX7vZsG0X1cseAUDDmkez/Pe/2LxjF6dXLx3cVoZVW/7OcWz/KYu47NlvaP7ctwz5z1Km/7SB52cuy9x/VLEEGh9Xjmk/baB4kXgy3HGHxIT8eymIhj4+ZBamSwREYk3ATGCKmQ1x9y3B6YBvgA4EqgCdgK9CPGcc0BZ4HeiY5fgVwOnA7OD+PXYAWT9f8xHQy8x6ubub2WnuPj/EDHmWkJDAvf0H0L3rzWRkpHN5myupXfs4hr3wHPXq1eeCJhfx88IF3NH71syFMcOHvcCUqe8DgX+IXxw1gqcHPwvAle3ac+89fUhPT6P/Aw9FJO+QQU/x999/0/eO3kBg9e/zw0ZGJG9eMxekPo6WzEt++YkvP/mApJq16detIwDtb+xJ2u7djB0+iO3btvLU/XdQo9bx3Pv4C/yxZTMvPvMo9zz2HADP/19/Fi+Yx45tf9KzY3PaXtuVCy9tDcCcrz/j2ONP5Oiy5QE47sSTuLtrB6rVrE31WsfnS/5o6GOAJz/8lf+74kQS4uNYuzWZB99bzGf//Z2+zY4jIc7YlZbBo9MDVZMTKx9F2wZVGDjtvwc8b9fza/DSlysA+GbpH1x1RlXe6t6QyfPW5Vv2aOljCbBApf0w36nZ9UBfIB2YDzxEYGFgOXIuDJzu7pPNrEbwev3gObLu2wkMIbDifxvQPrgwsA7wJrAT+BS4xt1rBAceHwFFCCwMnAo8C5xDYDy2wt1bBBcGZq4d2J+UtANWN0QKvEXBEnQ0OTGXsnhBd/Zjn0Y6Qki+7Z+/0z6HQ7GE8L23PvbOD8Ly7/3yZy477PWAiAwC8puZ7XT3/P/GixBoECCxQIOAw0ODgPAL5yCg1l0fhuXf+2WDLz3sg4DC+j0BIiIihV5M/HZApKsAIiJSeMTQJwRVCRARESmsYqISICIicrjE0pcFaRAgIiISghgaA2g6QEREpLBSJUBERCQEsTQdoEqAiIhIIaVKgIiISAhiqBCgSoCIiEhhpUqAiIhICOLiYqcUoEGAiIhICDQdICIiIlFPlQAREZEQ6COCIiIiEvVUCRAREQlBDBUCNAgQkX+cWKVkpCMUCt/2bxLpCCEpc06fSEcIWfLsQWE7t6YDREREJOqpEiAiIhICVQJEREQk6qkSICIiEoIYKgRoECAiIhIKTQeIiIhI1FMlQEREJAQxVAhQJUBERKSwUiVAREQkBFoTICIiIlFPlQAREZEQxFAhQIMAERGRUGg6QERERKKeKgEiIiIhiKFCgCoBIiIihZUqASIiIiGIpTUBGgSIiIiEIIbGAJoOiJSvv/yCVs0voUWzprz84uh9tvv4oxmcUu8Efvl5IQDzf5hH2zYt6XjVlaxauRKA7du3063LTbi78iqznhfKHJHMPdufx9xJfZj3eh9u7dAoc3v3q87lp7fuZt7rfXisV/Ncj2161gn89Nbd/Px2P/pcd2Hm9lcGdmT2hDt5uPulmdv63fhvWjSul6/ZC7NCNwgwsxpm9nPwegMze/4AbTvmd4b09HT+77GBDB/5ElOmvs+MD6azbOnSHO3++msnEyeM56STT8nc9uq4Vxj87Av0uv1O3nxjEgCjRw7n5q63hK1EFW15lVnPC2U+vJlPPLYSN1x+Fo06P0fDTs9w6Xl1qZVUjsan16JF43qc0XEwp3cYxLOvfZ7j2Lg449m729C690uc1v5p2l1yGnVqVqR+7coANOz0DOeeVpOSRxajUtmjaFCvGtO/+CXfsh8MMwvLJRIK3SAgK3ef6+637adJDSDfBwE/L1xAUlJ1qiYlUSQxkWaXNeezWTNztBv2/HN0vvFmihYtmrktISGBXSkppKQkk5CQwOpVq9i0aSMNzmiY3zGjNq8y63mhzIc3c52aFZj980qSd6WSnp7Blz8sp/UF9el65TkMGjeL3anpAGzeujPHsWfUq8ayNVtYse4PUtPSees/P9KicT1S09IpXrQIZkZiQgLpGRk8cMslPDJqRr5mL+yichBgZteZ2QIz+8nMppjZb2ZWJLivpJmtMLMiZlbbzD4JtvvBzGplO88FZjY9eP18M/sxeJlvZkcBTwCNgtvHwSmMAAAgAElEQVTuyK/8mzZupFLlSpm3K1SsyMaNG/dqs3jxIjZs2MD5F1y41/abbr6FgQ8N4LVXx3F1x2t44fkh9OzVO7+ixUReZdbzQpkPb+Zflm3gvNOO5ehSR1C8aBGanVuHqhVLU7taOc49tSZfjLmN/4zszul1k3Ice0z5UqzZ+Gfm7bWb/qRK+VIsWbGJ1Ru28u3423n7k5+oVbUcZsZPv67L9/yhMgvPJRKibmGgmdUD+gPnuvvvZnY0MBhoDrwLdADedvdUM5sAPOHuU8ysGIFBT4V9nLoP0NPdvzazEkAK0A/o4+4t8vMxODnn4rKWgjIyMhj05OMMfOzxHO3q1K3La5PeBGDe3DmUL18Bd6fvXbeTkJBAn779KFuuXH7Gjbq8yqznhTIf3sxLVmxi8KuzmP5CV/5K3s2C/60nLT2DhPh4ypQsTuMbn6fBiUm89vi11L38/7I9lpzn2/OY+w6Zmrlt8uAb6fX4ZO6+4SJOPu4YZn7/K6+89/0hZz8YsfTpgGisBDQBJrv77wDu/gfwEnBDcP8NwCvBd/JV3H1KsF2Ku/+9n/N+DTxjZrcBpd097UBBzKyrmc01s7n7W6yTXcWKldiwfkPm7U0bN1Khwj9jk7/++oul//uVmztfx6VNm7Dgpx/pfWv3zMU+wcfD6FEjuKVbD0YNH0qPnr1o0aIVEyeMz3OOWM2rzHpeKPPhzzxu6mzOue5Zmt4ynK3b/mbpqt9Zu+lP3p31MwBzF60mIyODcqWP3Ou4tZu2UbVi6czbVSqUZt3m7Xu1adG4Hj8sXs0RxROpd2wlrrlvPB0vO53iRYvkW/7CKuoqAYDB3kPj4Lv3GmZ2PhDv7j+bWclQTuruT5jZ+8BlwHdm9u88HDMaGA2QkpbLcH0f6tU/iVWrVrBmzWoqVqjIjA/e5/GnB2fuP+qoo/j8639GuDd1vpY7+9xNvfonZW6b+u4UGjc+n5KlSpGckoLFxWFxcaQkJ+c1Rp5FW15l1vNCmQ9/5vJlSrB5606SKpam9YUnccFNL5DhzgUNavPlD8uoXa0ciUUS+P3Pv/Y6bu6i1dROKkf1Y45m3aZttLv4VDo/MCFzf0J8HD07NOKKO16mdrVymVWCODMSi8STvCs13x5DXsVQISAqBwEzgSlmNsTdt5jZ0cFqwKvAJOARAHffbmZrzOxyd3/XzIoC8fs6qZnVcveFwEIzOxuoA6wGjsrvB5CQkMC9/QfQvevNZGSkc3mbK6ld+ziGvfAc9erV54ImF+33+OTkZKa+N4WRL44B4Lrrb+Cu22+jSJEiPJHlH4fCmleZ9bxQ5sOfedKT13F0ySNJTU/n9qff4c8dyYybOptRD1zF3El92J2axs0Pvw5A5XIlGd6/HW3ueJn09AzueHoK057vQnycMW7aHBYv/2fNQ7d25/La+3NJ3pXKwv+txzDmTLyLj775L9t2puRb/sLKwv351nAws+uBvkA6MN/dO5tZJeA3oLK7/xlsdxwwCigHpALtgAxgurvXN7MLCM75m9kLwIXBcy4COgfbzggeP9bdh+wrUyiVABGRaFLmnD6RjhCy5NmDwvZ+vdHgr8Ly7/2Xd5132GsM0VgJwN3HAeOybT6PwFqBP7O0+x+BNQTZ1Q/u/wz4LHi91z7ubv/DbBERkSgVlYOA7ILv4i8lMJ8vIiISNrH06YCYGATs5128iIhIvoqhMUBUfkRQRERE8kFMVAJEREQOl1iaDlAlQEREpJBSJUBERCQEMVQI0CBAREQkFJoOEBERkainSoCIiEgIYqgQoEqAiIhIYaVBgIiISAjizMJyORAza2ZmS8xsqZn120ebq8xskZn9YmYTD3ROTQeIiIiEIBLTAWYWDwwDmgJrgDlmNtXdF2VpcxxwL3Cuu281swoHOq8qASIiIgVfQ2Cpuy93993A60DrbG26AMPcfSuAu2860ElVCRAREQlBhD4iWAVYneX2GuDMbG2OBzCzr4F44CF3n7G/k2oQICIiUgCYWVega5ZNo9199J7duRzi2W4nAMcBFwBVgS/NrL67/7mv+9QgQEREJARxYSoEBF/wR+9j9xogKcvtqsC6XNp85+6pwG9mtoTAoGDOvu5TawJERERCYGZhuRzAHOA4M6tpZolAB2BqtjbvAhcGM5YjMD2wfH8nVSVAokaZM26NdISQbZ0zNNIRRA7Z1m8GRTpCoefuaWZ2K/ARgfn+Me7+i5kNBOa6+9TgvovNbBGQDvR19y37O68GASIiIiGI1DcGuvsHwAfZtg3Ict2BO4OXPNF0gIiISCGlSoCIiEgILNeF+tFJlQAREZFCSpUAERGREITrI4KRoEGAiIhICCL0jYFhoekAERGRQkqVABERkRDEUCFAlQAREZHCSpUAERGREMTFUClAgwAREZEQxNAYQNMBIiIihZUqASIiIiHQRwRFREQk6qkSICIiEoIYKgRoECAiIhKKWPp0gKYDRERECikNAiLk6y+/oFXzS2jRrCkvvzg6x/5Xx75Cm5aX0bZNS7rceD3r1q0FYMVvy+nQ7gratWnFTz/OByAtLY2uN3UmOTm50OftefUFzH3rPuZN7s+tHS8A4OTjq/D5uLv47vV+fDXhbhrUq57juGqVy/D1hLv57vV+zJvcn5vbngdAYpEE3hvag7lv3UfXdo0y2w+9/2pOOaFqvuePln6O1rzRmDna8kZr5lBYmC6RUGAGAWb2kJn1MbOBZvbvfbTZ574sbVqZWb/g9bFm1jaXNheY2fT8SR669PR0/u+xgQwf+RJTpr7PjA+ms2zp0r3a1Klbl4lvvs3kKdNoevElDBn8NABvvfkGve+4i0HPPs+4V8YA8OYbk2jRsjXFixcv1HlPrFWZG644h0bXPk3D9o9zaeP61KpWnsduv5zHRn/IWR2e4JER03ns9stzHLt+83Yu7PwMZ3V4gsbXPk2fG5pSuXwpmp5Tl/mLV3PGVY9z45XnAnDS8VWIizN+WrImX/NHSz9Ha95ozBxteaM1c2FWYAYBe7j7AHf/JPt2M4vf175sx0919yfCl/DQ/bxwAUlJ1amalESRxESaXdacz2bN3KtNwzPPynzSn3TKqWzasAGAhIQEdqWkkJKcQkKRBLZv387nn82iZeucL2yFLW+dmpWYvXAFySmppKdn8OW8pbS+8BTcoeSRxQAoVaI46zdvy3Fsalo6u1PTACiaWCRzzi81LZ3ixYqQEP/Pn8qAHi14ZMT7+Z4/Wvo5WvNGY+ZoyxutmUNlZmG5REJEFwaaWX/gOmA1sBmYZ2ZjgenuPtnMVgBjgIuBoWbWLMu+y4BngN+BH4Bj3b2FmXUGGrj7rcG7+beZ9QYqAne6+14VADM7EngBOIlAfzzk7u+ZWT3gFSCRwGDpSnf/X3487k0bN1KpcqXM2xUqVmThggX7bD/l7cmc26gxAB2u7sT9993D7t27eeDBhxk1YhhdunYL6xMoWvL+smwdD93akqNLHUnyrt00O68ePyxaRd9Bk5k2rCeP39GGuDjjws6Dcz2+asXSvPN8d2ollee+Z99l/eZtbPpjBx2bN+SL8X0YMvYTmp9/EvMXrcp1IHGooqWfozVvNGaOtrzRmrkwi9ggwMxOBzoApwVz/ADMy6VpirufFzymWfC/xYBRQGN3/83MJu3nrmoA5wO1gFlmVjvb/v7Ap+5+o5mVBmab2SdAN+A5d59gZolA/EE+1Bwcz7FtX0/y6dPeY9EvPzNm3GsAVD7mGF4eOx6AVStXsnnzJmrWPJb7+vUlNTWVnr16U6NGzfyKGlV5l/y2kcFjP2b6iFv5K3kXC35dS1paOl3bNeLuwe/w7swfubLpaYx4sBPNuw3NcfyajX/SsP3jVC5fijef6cKUT+az6Y8ddL5vLAAJCXFMG9aTtreP4sm7riCpUhkmTJ/N+58vzJf80dLP0Zo3GjNHW95ozRyquBgak0RyOqARMMXd/3b37cDUfbR7I5dtdYDl7v5b8Pb+BgFvuntG8F388uCxWV0M9DOzH4HPgGJANeBb4D4zuweo7u45VqWYWVczm2tmc3Nb/LIvFStWYsP6DZm3N23cSIUKFXK0++7bb3hp9EieGzqCxMTEHPtfeH4IPXv1ZuKE8TRv3pIePXsxanjOF7dDFU15x737Led0fJKmNz3L1m1/sXTVZjq1OJN3Z/4IwNsfz891YWBW6zdvY9GyDZz7r1p7bb+lXWNemz6bM08+lt2paVxzzxj63XxJvmWPpn6OxrzRmDna8kZr5lDF0nRApNcE5Bwy5vRXLttC6a3s95H9thEo9Z8avFRz98XuPhFoBSQDH5lZkxwndh/t7g3cvcFNXbrmOVC9+iexatUK1qxZTeru3cz44H3Ov3Dv0y9evIhHHh7Ac0NHULZs2RznmDtnNhUqVKR69RqkJCdjcXHExceTnJKS5xyxmLd8mRIAJFUqQ+smp/DmjLms37yNRqcfB8AFDY9n6arNOY6rUqE0xYoWAaD0UcU5+9Rj+XXFpsz9pY8qzqWN6zNh2vccUbwIGRmOe2D9QH6Jpn6OxrzRmDna8kZr5sIskmsCvgDGmtkTwRwtCZT48+K/wLFmVsPdVwDt99O2nZmNA2oCxwJLgLOy7P8I6GVmvdzdzew0d59vZscSqDY8H7x+MvBpKA9wXxISEri3/wC6d72ZjIx0Lm9zJbVrH8ewF56jXr36XNDkIoYMeoq///6bvnf0BqBS5co8P2wkAO7Oi6NG8PTgZwG4sl177r2nD+npafR/4KH8iBi1eScNupmjSx9Jalo6tz/xJn/uSKbnIxN5um9bEhLi2LUrjVsfDRSO/nViNW5uex49Bk7khJqVeOLONjiOYTz76kx+Wbou87z3db2UJ16aAcDH3yzmlqsaM/et+3hp8lf5lj2a+jka80Zj5mjLG62ZQxVLSxTMPS9vxsN05/8sDFwJrAEWAfXZe2FgA3f/Pdh+bJZ9LYGnCSwMnA1UdPdOWRcGBttvBRqQZWGgmV0A9AkuJCwOPAucQ6AqsCK4/V7gGiAV2AB0dPc/9vVYUtLyVNWQQ1DmjFsP3KiA2TqnYJQvRQqbYgnh++j9tRN+Csu/9+M7nXLYhxcRHQQcCjMr4e47LTCRMgz4n7sPiVQeDQLCT4MAEcmrcA4Crpu4ICz/3r/a8eTDPgiI5t8O6GJm1xP4CN988j6VICIictBi6dMBUTsICL7rj9g7fxERkWgXtYMAERGRSIilLy+K9EcERUREJEJUCRAREQlB7NQBNAgQEREJSZymA0RERCTaHXAQYGbjgj+ss+d2GTMbE95YIiIiBZNZeC6RkJdKwMnu/ueeG+6+lcAv/4mIiEgUy8uagDgzKxN88cfMjs7jcSIiIjEnlj4imJcX88HAN2Y2OXi7HfBY+CKJiIjI4XDAQYC7v2pmc4EmBD4ZcYW7Lwp7MhERkQIohgoB+x4EmFlJd98eLP9vACZm2Xf0/n5RT0REJFbF0kcE91cJmAi0AOYBzt7fj+DAsWHMJSIiImG2z0GAu7cI/rfm4YsjIiJSsMVQISBvq/zN7ArgPAIVgC/d/d2wphIREZGwO+AgwMyGA7WBScFN3cysqbv3DGuyKFOm1XORjhCSP97rHekIIftj9tBIRwiZe6QTiEh+K2wfETwfqO8e+OfMzMYBC8OaSkREpICKpe/bz8tjWQJUy3I7CVgQnjgiIiJyuOSlElAWWGxms4O3zwC+NbOpAO7eKlzhRERECprCNh0wIOwpRERE5LDLyzcGfm5mFQlUAABmu/um8MYSEREpmOJipxCQp58SvgqYTeA3A64CvjeztuEOJiIiUhDFWXgukZCX6YD+wBl73v2bWXngE2Dyfo8SERGRAi1PPyWcrfy/hdj6hISIiEieFbaFgR+a2Uf882VB7YEPwhdJREREDoe8DAIcGEXga4MNGA2cFc5QIiIiBVUsLQzMyyCgqbvfA7yzZ4OZPQzcE7ZUIiIiEnb7HASYWXegB3CsmWX9hsCjgK/DHUxERKQgiqElAfutBEwEPgQeB/pl2b7D3f8IayoREZECKi6GRgH7HAS4+zZgG3D14YsjIiIih0te1gSIiIhIUCx9Rj6WHouIiIiEQJUAERGREMTQkgBVAg6nXpefxrwR1zB3eCfG3d2MokXiueCUJL55/mq+e6EjM59ux7GVS+U4rkhCHKPuaMqc4Z34fmhHGp1UBYDEhHjeG9iaucM70bX5yZnth/ZqwinHls/X7F9/9QWtW1xCy0ubMual0Tn2v/fuO1zY6CyuurI1V13ZmncmvwXAit+Wc/VVV3DVFa346cf5AKSlpXHLzZ1JTk7O14yhZgb4aMYHXNHqMq5o3Zx+d99V4DOvX7+Om2+4lvZtL6ddm5Z8+cXnAMz/YR7t2rSkY/srWbVqJQDbt2+ne9ebcPeI5Y3G54X6OPyZoeD97YUiziwsl0goUJUAM6sBTHf3+tm2DwS+cPdPzGwF0MDdf8/Wphvwt7u/epjihuSYskfSo9UpnNZtPCm703nt3ktpd/7x3N3+DNoNnMaS1Vvp2vxk+nVoSNchH+917I3NAt1xRo8JlC9VnHcHtua821+n6enVmL90E5c/+B7fPt+R0e8v4KSa5YiLM35avjnfsqenp/P4owMZ+eIrVKxUkU7t23L+hU2oVav2Xu0ubnYZ9/bf+5enJ7/1BrfdfhfHVKnC80MGM/jZF3jrjUk0b9ma4sWL51vGg8m8cuUKxrw0mrHjJ1GyVCn+2LKlwGd+cdQILr7kUq7q0JFly5Zya/eufPifTxk/7hUGPfsC69au5a03JnFX3368OGo4N3W5JWxfcRqrzwv1cfgzF7S/vcIsKioB7j7A3T85QJuRBXUAsEdCfBzFExOIjzOKFy3C+i1/4Q4lj0gEoOSRiaz/468cx9WpdjSzflwFwOZtyWz7azenH1eR1PQMihdNICH+n/+NA649m0fGf5evuX9euICkatWpmpREkSKJXHJpcz77dGaejk1ISGDXrhRSklNISEhg+/btfP7ZLFq2ujxfMx5M5ncmv0n7Dp0oWSpQfTm6bNkCn9nM+OuvnQDs3LGD8uUr/JM5JYWUlGQSEhJYvWoVmzZupMEZDSOad1/Ux/mXd18Kch8XtL+9UJmF5xIJBaoSEBRvZi8C5wBrgdbACAIVgj2/XNjXzC4MXu/o7kvN7CFgp7sPMrPbgG5AGrDI3TuYWQngBaABga9Cftjd3zazne5eAiD4E8kt3L2zmbUDHgTSgW3u3vhQHtS6LX/x7Ds/8Ou4G0nencbMH1Yxc/4qejz3CVMebk3K7jS2/72b8+94M8exC5f/TsuzavHW579StfxRnFa7AlXLH8W0b5fRsUldvhjSniFvz6P5mTWZv3RTrgOJQ7Fp00YqVaqUebtixYosXLggR7uZH/+HH+bOoXqNmvS5+14qVa5M+6s7cf+995Caupv7BzzM6JHDuLlrt7D/AEdeMq9cuQKA66/pQEZGBt163Mq55zUu0Jm79biV7l1vYtLE10hOTmbUi68AcGOXW3jk4QEULVqUxx5/mmcGPUmPXr0jnhei73mhPg5/5oL2t1eYFcRBwHHA1e7exczeBK7Mpc12d29oZtcBzwItsu3vB9R0911mVjq47QECL+YnAZhZmQPkGABc4u5rs5zjoJUuUZQWZx1L3RvG8udfu5h432V0uPAELj+nNm0efI85SzZyx5X/4smujejx3N6j5nH/+YU6SUfz9XNXs2rTdr5bvJ609AzSM5zOT80AAlWGaY9eTtuHp/Fkl0YklT+KCTMX8/73vx1q9FznO7P/UZ5/wYVcelkLEhMTeeuNSTzQ/x5eHPMqlSsfw8tjxwOwatVKNm/aRM2ax9K/X19SU1Pp2as31WvUPOSMB5M5PS2dVStX8tIr49m0cQM3XN+JyVOmF+jMMz54n1at23Bd5xv56cf53H/v3Ux+dzp16tRl/MTAAHLe3DmUr1AB3Ln7rttJSEjgrr79KFuu3GHPG43PC/Vx+DMXtL+9UMXSbwcUxOmA39z9x+D1eUCNXNpMyvLfs3PZvwCYYGbXEKgGAPwbGLangbtvPUCOr4GxZtYFiM+tgZl1NbO5ZjY3bdU3+z1Zk1OTWLFhO79vTyYtPYN3v17K2Scew0nHlmPOko0ATP7if5xVt3KOY9MznLtf/IKzek3kqkemU/rIRJau/XOvNre0OJnXPlnMmXUrszs1g2ue+JB+HfKnTFmxYiU2bNiQeXvjxo2ZJdI9SpcuQ2JiYFrjirZXsXjRLznOM/S5IfTo1ZuJE8ZzWYuWdO/Zi5EjhuZLxoPJXLFiRS5ochFFihShStUkatSoyargO5SCmnnKO5O5+JJLATjl1NPYtXsXf27956ns7rw4agRdb+nByBFD6d6zF81btmLihPERyRuNzwv1cfgzF7S/vVDF0sLAgjgI2JXlejq5Vyt8H9f3aE7gBf90YJ6ZJRD4BcTc2mbdVixzo3s34H4gCfjRzMrmONB9tLs3cPcGCdXO2cfDCVi9eQcN61SieNHAw7nw1CT+u+oPSh5RlNpVAoWGJqdVY8nqnGOT4kUTOCJ4XJPTqpGW4fx39T/f3Fy6RFEubViTCTMXc0TRBDLccXeKJuZPoade/ZNYtWoFa9esJjV1Nx99+D7nX9hkrzabN2/KvP75rE+peWytvfbPnTObChUrUr16DVJSkjGLIy4+npTklHzJeDCZL7zo38yZ/T0AW7f+wcoVK6ialFSgM1euXJnvv/8WgOXLlrF71y7KHH105v6p702hUePzKVmqFCnJKcTFxWEWR0pK/q+sjtXnhfo4/JkL2t9eYVYQpwPyoj3wRPC/32bdYWZxQJK7zzKzr4COQAngP8CtwO3BdmWC1YCNZlYXWAK0AXYE99dy9++B782sJYHBwJaDDTxnyUamfLWUb5+/mrT0DH5avpmXP/yZtb/vZFL/5mRkOH/u3MUtzwY+GdD8zJr867iKPPLad5QvVZxpj7YhI8NZt2UnNw36aK9z33f1mTwxaTYAH89byS0tTmbu8Gt46YOFBxt3LwkJCfS7bwDdb7mZjPR0Wre5ktq1j2P40Oc4sV59LrjwIia9Np7PPvuUhPh4SpYqxcBHH888fs87p6cHPwvAlW3bc1+/PqSnpXHfAw/lS8aDyXzOuY349puvuaLVZcTFx3PHXXdTunSZAp35zr79GPjg/Ux4dSyY8fCjT2SWWpOTk5n23hRGjB4DwLXX38Bdd9xGkSJFeOKpwRHJG43PC/Vx+DMXtL+9UMXSEgUL5+dbQ5X9I4Jm1ofAC/ie7ZODHxF8BbiMQCXj6qwLA4HngFlAKQLv/l9z9yeCCwP3VAfSCSwMfCe4GPBJYDXwM1AiuDDwHQLrEwyYCdzu++ms4pc9V3A6Mg/+eC+8C5pERCKpeBHC9lL9yCdLw/Lv/QP/rn3YhxcFahAQzTQIEBEpOMI5CHhsZngGAf0vOvyDgIK4JkBEREQOg2hdEyAiIhIRFr4iw2GnQYCIiEgI9D0BIiIiEvVUCRAREQmBKgEiIiIS9VQJEBERCUEs/aCRBgEiIiIh0HSAiIiIRD1VAkREREIQQ7MBqgSIiIhEAzNrZmZLzGypmfXbT7u2ZuZm1uBA51QlQEREJARxESgFmFk8gR/BawqsAeaY2VR3X5St3VHAbcD3eTmvKgEiIiIhiLPwXA6gIbDU3Ze7+27gdaB1Lu0eAZ4CUvL0WEJ43CIiIhIZVQj85P0ea4LbMpnZaUCSu0/P60k1HSAiIhKCcM0GmFlXoGuWTaPdffSe3bkckvmTxmYWBwwBOodynxoEiIiIFADBF/zR+9i9BkjKcrsqsC7L7aOA+sBnwS8zqgRMNbNW7j53X/epQYCIiEgI4iLzU8JzgOPMrCawFugAdNyz0923AeX23Dazz4A++xsAgAYB+Wbr1N6RjhCS5N3pkY4QsuKJ8ZGOELKyHV6JdISQbHn9hkhHEJFcuHuamd0KfATEA2Pc/RczGwjMdfepB3NeDQJERERCEKkvC3L3D4APsm0bsI+2F+TlnBoEiIiIhEC/HSAiIiJRT5UAERGREETiGwPDRZUAERGRQkqVABERkRDEUCFAgwAREZFQaDpAREREop4qASIiIiGIoUKAKgEiIiKFlSoBIiIiIYild88aBIiIiITAYmg+IJYGNCIiIhICVQJERERCEDt1AFUCRERECi1VAkREREKgLwsSERGRqKdBQIR8/eUXtGp+CS2aNeXlF0fn2D9v7hzat23Dv04+kY8/mpG5fcVvy+nQ7gratWnFTz/OByAtLY2uN3UmOTk5rJnT09O5rsMV3HVb9xz73nnrdTq1a8217dvQ9YZr+G3ZUgB++vEHOl11OTd0uorVq1YCsGPHdnr36IK7hzVvNPTxrS1O5P/bu/M4m8v+j+OvzzgziOx7hNKKUtGmiOqXIsuNUrlb7oRI3GmTFuW+S7dKJQlps5Q20b6hRclSlnTfdatshUnpFs0YM/P5/THHZJnJHM6ZOd9z3s8e59H5nu/2Ppcz51znc13fmQWjOrPggc48Nag1pVNLMa7/aSwf041PR3bk05EdOaZBlQL3vaR1I5aM7sqS0V25pHUjANJCKbwy9GwWPNCZq845Mn/b0X1O5diGBR9nfwShjYOeOWh5g5o5EhajW0mIm06AmW2JwTEvN7NHCll3S7TPV1Q5OTnc/c+7ePSxx5k+83XeeuM1vl2xYpdtatWuzfB/3sO57Tvs8vgLz09j4N8Hc9+DD/P0k08A8Py0Z+lwfifKli0b09zTpk6iQcNDC1x3zrkdmPLCDCZNm07Py/7GQw/8C4Cpk57inpEP0nfAIF5+4TkAnhg/lsv+1juml9kEoY1rVzmAq889mtNvepUW171CSorRvWVDAIZOWsApN8zklBtmsnTlL3vsW7l8GkMuaMYZQ16l9c2vMuSCZlQql3mfixcAACAASURBVMZZzQ7ii+9+5sTBr/C3sw4HoGn9yqSYseT7PY+zP4LQxkHPHLS8Qc0cKbPY3EpC3HQCSkDEnQAzKxWNE3+5bCn16tWnbr16pKal0e689syZ/f4u2xx0UF0OP+JIUmzXf6JQKMS2zEwyMzIJpYbYvHkzH8yZzfmdOkcjWqHSN6znk48/oGOXrgWuL1e+fP79zJ167KFQiG3btrEtI4NQKJW1a1bzU3o6xzdvEdO8QWnjUKkUyqaVolSKcUDpEOs2/V6k/c469iBmLfmRTVuy+HVrFrOW/MjZzQ4iOyeXsmmlCJX64znd1uN4hk/7POrZg9LGQc4ctLxBzZzM4m5ioJmVB2YAlYFU4FZ3n2FmDYDX3L1JeLvrgfLuPszMWgATga3Ax8C5O7YD6pjZW8ChwHR3v9HMRgBlzWwxsNzdLzGznsC1QBrwGdDP3XPCFYoHgHOAweHj75f0DRuoVbtW/nKNmjVZtnRpkfbtcdEl3HrLTWRlZXHbHXcybuwYrurdN+a/vGLUyBFcM/B6tv6+tdBtXpw2lWcnP8327dt5ZFxeL/6yv13FiH/cQenSZRj2jxE8/MBIevcbENOsEIw2XvfL7zw080v+M/YCMrJymLX0B95f8iMXnHYId1x0AkO6N2POsnXcNnkhWdm5u+xbp2o51v78x7/FD79spU7Vcrz86UouanUoH9zTgVEzvuS85vX44rufWb8p+qXUILTx7oKWOWh5IZiZIxVvefZH3HUCgEygi7tvNrNqwDwzm7mXfZ4Eerv7J+EP+J01A44DtgFfm9lod7/ZzK5x92YAZnYUcCHQ0t23m9mjwCXAM0A54Et3vz1aT9DZcyy8qC+q2nXqMPGpSQCsXrWKn35Kp2HDQ7jl5hvYvn07/QcMpEGDhtGKCsDHH86hcpUqHHl0YxYtnF/odt0uvJhuF17M22++xlOPj+P24fdw+BFHMfGZvGGALxYtpFr1GgAMvek6QqEQ1153I1WrVotqXghGG1cql0aHFgfTuP8L/Lo1i8mD29Dj9EO4Y8oi1v+aQVoohUf6tuS6zk0Z8eKSXZ9LAcdzd3JynSse+hCAUClj5q3n0P3e9xhxWQvqVivP1A9W8MbCNfudHYLRxrsLWuag5YVgZk5m8TgcYMDdZrYUeA84CKhZ6MZmlYAD3f2T8ENTd9vkfXf/n7tnAl8B9Qs4zJnACcCCcHXgTOCQ8Loc4KVCzt3bzBaa2cKCJr8UpmbNWqxftz5/OX3DBmrUqFHk/XcY/fAo+g8YyNQpk2jf/nz69R/AuEcLnAKxX5Yu/pyPPphN5/PO4rabB7NwwWfcMfTGQrc/+5zz+GDOruU/d+fJxx/jb7378vi4MVzV9xranXc+zz87Oep5IRht3OaYOqxM/42Nm7eRnePM/GwVJx1Rg/W/5n1rz8rOZdLs/9L8sOp77PvDz1upW7Vc/vJBVcqx7pddhxJ6n3MUU+as4KTDa5CVnculo+ZwU9djo5IdgtHGuwta5qDlhWBmjlRKjG4lIR47AZcA1YETwt/UNwBlgGx2zVsm/P+9dTG37XQ/h4KrHwY87e7Nwrcj3H1YeF2mu+cUdGB3H+/uzd29+ZVX9d5LjD80btKU1atXsnbtGrZnZfHWG6/Tuk3bIu8PsHDBfGrUqEn9+g3IzMjAUlJIKVWKjMzMiI5TFP2uvY5X357NK2+8x/AR99O8xUnc+c9/7bLN6lUr8+/P/egD6tXbta/1+quv0PL01lSoUJFtmZmkpKSQkpLCthjkhWC08ZqNW2hxeHXKpuVNNTmjaR2+/uF/1Kr0xwSo81sczFerN+2x73tLfuDMY+tQqVwalcqlceaxdXhvyQ/56yuVS+PcE+oy5YMVlC0dItcdd6dMWlSmtQDBaOOgZw5a3qBmjpSZxeRWEuJxOKAikB4uy7fhj2/uG4AaZlYV2AJ0AN5y901m9puZnezu84AeRTzPdjNLdfftwPvADDMb5e7pZlaFvOrCqug+tTyhUIghQ2/n6t69yM3NoXOXrjRqdBhjRj9E48ZNOKPtmXy5bCl/H3hN/sSYR8eMZvrM14G8b9UTxo1l5P0PAtC1+4UMuel6cnKyGXrbsFhELtD4R0dz5NGNaXVGW16cNpUFn31KKBTiwAoVuX343fnbZWZk8MarM3j40QkAXNTzMoZcP5BQairD77kvJtmC0MYL/7uRVz5dydyRHcnJcZZ8/zNPvPs1rwz9P6pVKIMZLF35C9eOzytyHXdoVXqdfST9H5vLpi1Z3PvSEj4ccT4AI15czKYtWfnHHtK9Gfe+lDeE8N7iH+jT7kjmP9CZie98HZXsEIw2DnrmoOUNauZkZrG+VruozGyLu5cPzwN4lbxJgYuBluRN9FtpZteSN3nve+AHYGV4YuBJwATyJgbOAVq5e0szuxxo7u7XhM/xGnCfu88xs3uBjsDn4YmBFwJDyKs2bAf6u/u8Hbn2lj8zu4CBsDiWkVVgcSOulY3it9jiUrXHkyUdISI/P3dFSUcQiYoyodhdev/C4h9j8n7fvVmdYi8HxE0nYH+YWXl33xK+fzNQ290HFmcGdQJiT52A2FMnQBKFOgFFE4/DAfuivZkNIe/5rAIuL9k4IiKSqHSJYJxx92nAtJLOISIiiS8eZ9Tvq0R6LiIiIhKBhKgEiIiIFJdEGg5QJUBERCRJqRIgIiISgcSpA6gSICIikrRUCRAREYlAAk0JUCdAREQkEikJNCCg4QAREZEkpUqAiIhIBBJpOECVABERkSSlSoCIiEgELIHmBKgTICIiEgENB4iIiEjgqRIgIiISAV0iKCIiIoGnSkCSKptWqqQjJIWfn7uipCMkvMotbyjpCBHbNHdkSUeQ/ZBIcwLUCRAREYlAInUCNBwgIiKSpFQJEBERiUAi/Z4AVQJERESSlCoBIiIiEUhJnEKAOgEiIiKR0HCAiIiIBJ4qASIiIhHQJYIiIiISeKoEiIiIREBzAkRERCTwVAkQERGJgC4RFBERSVIaDhAREZHAUyVAREQkArpEUERERAJPlQAREZEIJFAhQJWAkjL3ow/p2P4cOrQ7m4kTxhe63btvv8WxjY9g+ZfLAPji80V063I+F1/QldWrVgGwefNm+l51Je6uvMqs10UB+l94GgunDmbRs4O5psdpAAztdTbfvnor8yb9nXmT/s45px5Z4L4DepzOomcHs3DqYJ4efjGl0/K+Oz1550XMn3wdd17dLn/bm/92Fh1aNY56/qC0c1DzRirFLCa3EnkuJXLWgDGzlWZWLVrHy8nJ4e5/3sWjjz3O9Jmv89Ybr/HtihV7bLd16xamTplE02OOzX/smaef5P4HRzNg0HU8P+1ZAMY/9ii9evfBYvQiClpeZdbrYmdHH1KTKzqdxOlXPMyJPUdxbsujObRe3o/z6Oc+4uS/juLkv47i7U/+s8e+dapXoN+Fp9Hy8odofvH9lEpJofvZzWjSqDYAJ/Z8gJbNGlKhXBlqVT2Q5kfX47UPl0c1f1DaOah5k506ASXgy2VLqVevPnXr1SM1LY1257Vnzuz399huzMMPcfnfelG6dOn8x0KhENsyM8nMzCAUCrFm9WrS0zfQvMWJyqvMel0U4MgGNZn/5Soytm0nJyeXj774jk6tmxR5/1CpFMqWTqVUqRTKlkll3cbNbM/OoWzpVMyMtFCInNxcbut9DsPHvx31/EFp56Dm3RcWo1tJCHwnwMwamNl/zOxxM/vSzKaY2VlmNtfM/mtmJ5pZFTN7xcyWmtk8MzsmvG9hj1c1s3fM7AszG0eU/33SN2ygVu1a+cs1atZkw4YNu2zz739/xfr162l9RptdHr+yVx/uGnY7k595mosu7snoh0fRf8DAaMYLfF5l1utiZ8u/W89pxx1ClQoHULZ0Ku1OPZK6NSsC0LfbqcyffB2P3dqdSgeW3WPfH3/azINTPuCbGUP5/vXb2Lwlk/c/+4avV6azZsOvfPrMIF56fwmH1q2GGSz55seo5w9KOwc1b7JLlImBjYDuQG9gAXAxcBrQEbgFWAN84e6dzawt8AzQDLizkMfvAD5297vMrH34uHsws9471j3y6DiuvKrAzfbg7Dm2tXOpKzc3l/vuvYe7/nnPHtsdedRRTH72eQAWLVxA9eo1cHduGDyIUCjE9TfcTNVqURu5CGReZdbrYmdfr0zn/mdm89roq9iakcXS//5Idk4uE17+lHueeA93uKPPOYwY2IG+/3hhl30rHViWDq0ac1SXe/j1twym3vNXerQ7nufe+pwbRs3M3+7F+65gwIiXuPHythxzWB3en/8NT86YH5X8QWnnoObdJwk0MhH4SkDY9+6+zN1zgeXA+543i2QZ0IC8DsEkAHefBVQ1s4p/8ngrYHL48deBTQWd1N3Hu3tzd29e1A4AQM2atVi/bn3+cvqGDdSoUSN/eevWraz47zf0uvxSzj27LUuXLGbgNVfnT54Jn5vx48bSp28/xj36CP36D6BDh45MnTKpyDkSNa8y63Wxu6dfXcCplz3E2X3Hsmnz76xYs5H0X7aQm+u4O0/M+IzmRx+8x35tWxzGyh9/YeOvW8nOyeWV2V9yctP6u2zToVVjPv/3Wg4om0bjQ2vRc+hkLj73BMqWTo1K9iC1cxDz7guL0X8lIVE6Adt2up+703IuedWOglrX/+Txnf8fdY2bNGX16pWsXbuG7VlZvPXG67Ru0zZ//YEHHsgHcz/jzXdn8ea7szjm2GY89MhYGjdpmr/NzFem06pVaypUrEhGZiaWkoKlpJCZkZH0eZVZr4vdVa9cDoB6NSvR6YymPP/OYmpVPTB/fafWTfjqu/V77LdmwyZObHJw/gd6mxaN+Hplev76UKkU+l94GqMmz+GAMqnsmMCekmKkpZaKSvYgtXMQ8ya7RBkO2JsPgUuA4WZ2BrDR3Teb2d4e/4eZnQtUjmaYUCjEkKG3c3XvXuTm5tC5S1caNTqMMaMfonHjJpzR9sw/3T8jI4OZM6bz2IQnALj0sisYPOhaUlNTGTHy/mhGDWReZdbrYnfPjriUKhXLsT07h0Ejp/PrbxlMHNaDYw6rgzusWvcLA0a8BEDtahV4dGg3uvz9CRYsX8P0Wcv49JlBZOfksuSbH5j4yrz84/btfiqTX19IxrbtLPvvOsxgwZTrePuT//C/LZlRyR6kdg5i3n2RSBcqWDxde7kvzKwB8Jq7NwkvPxVefnHHOvLK+08CDYHfgd7uvtTMqhTyeFXgWaAa8AHwF+AEd99YWI7M7NhVDkSkcJVb3lDSESK2ae7Iko6Q8MqEYldfn//d/2Lyfn/iIRWLvXsR+E5AvFAnQKRkqBMgBYllJ2BBjDoBLUqgE5AocwJEREQkQskyJ0BERCQ6EmhOgDoBIiIiESipy/liQcMBIiIiSUqVABERkQgk0iWCqgSIiIgkKVUCREREIpBAhQB1AkRERCKSQL0ADQeIiIgEgJm1M7OvzWyFmd1cwPrrzOwrM1tqZu+bWf2CjrMzdQJEREQiUBJ/RdDMSgFjgHOBo4GLzOzo3Tb7Amju7scALwL/2ttzUSdAREQk/p0IrHD379w9C3gO6LTzBu4+291/Dy/OA+ru7aCaEyAiIhKBErpE8CBgzU7La4GT/mT7K4E393ZQdQJEREQiEKs+gJn1Bnrv9NB4dx//J6ct8A8ZmVlPoDnQem/nVCdAREQkDoQ/8McXsnotUG+n5brAj7tvZGZnAUOB1u6+bW/n1JwAERGRSFiMbn9uAXCYmTU0szSgBzBzl1hmxwHjgI7unl6Up6JOgIiISJxz92zgGuBt4N/A8+6+3MzuMrOO4c1GAuWBF8xssZnNLORw+cy9wCEFiVBmdsFjM5LcKrcdVtIRIrJp1rCSjiASFWVCsfuVPkvXbInJ+/0x9coX+5RDVQJERESSlCYGioiIRCCR/oqgOgEiIiIRSKA+gIYDREREkpUqASIiIpFIoFKAKgEiIiJJSpUAERGRCOztL/4FiToBIiIiEUikqwM0HCAiIpKkVAkQERGJQAIVAlQJEBERSVaqBIiIiEQigUoB6gSIiIhEIJGuDtBwgIiISJJSJUBERCQCukRQREREAk+VABERkQgkUCFAlYCSMvejD+nY/hw6tDubiRPG77H++WnP0rXz+Vzwl05c1vMivl2xAoAvPl9Ety7nc/EFXVm9ahUAmzdvpu9VV+LuJZZ3xvSXOeO0k7ngL5244C+dePnFFwBY+f139Oj+F7p36ciSxV8AkJ2dTe8rLycjIyNmeYuSOR7auH+3k1j4VD8WPd2Pa7qfDMDtV7Zh/pNXM29iX169/6/UrnrgHvsdXLMicyf0Zt7Evix6uh+9OjYHIC21FDNG9mThU/3o3blF/vaPXH8+xx5WK6rZIRhtrMzx934Rb3mTmQWxYc3scuAdd/8xvPw48IC7f1VSmTKzKXJD5uTk0LH9OYyb8CQ1a9bk4gu7MWLkAxzaqFH+Nlu2bKF8+fIAzJn1PtOem8rY8RP5+8BrGHTd9fz4ww/M/fgjrr/xZu771wjOaNOW5i1OjP4TK2LeGdNfZvnyL7nl1tt32Xfkvfdw2umtqHPQQTz0wP088NBopk6ZRPly5enYuUtM8hY1c3G0ceW2wwpdd3TDGjxzRzdO7zOBrOwcZo7sybUPvEb6L1v57fdtAPTrehJHNqjOtfe/tsu+qaFSmEHW9hzKlU1j0VP9aNNvIscfWYcWRx3EnRNn8+njfTj5ysdoemhNru56Ev3+NXOveTfNKjzv7uKljSOhzPHxflEcecuEYveF/ZsNv8fkg/PwmgcUe5EhqJWAy4E6OxbcvVdJdgAi9eWypdSrV5+69eqRmpZGu/PaM2f2+7tss+MHBCAjIwMLz0QJhUJsy8wkMzODUCjEmtWrSU/fENM3oaLkLUx+3oxMQqkhNm/ezAdzZnN+p84xy1vUzCXdxkfWr8b8r9aSsW07OTm5fLR4JZ1OPyq/AwBwQJnUAr8Bbc/OIWt7DgClU0uRkmL5j5ctnUqo1B8/2rf3asvwibOjmh2C0cbKHJ/vF/GUd19YjP4rCXExJ8DMGgBvAZ8BxwHfAJcC1wPnA2WBT4A+QFegOTDFzDKAU4A3gevdfaGZ/R9wJ1Aa+Ba4wt23mNlK4Onw8VKB7u7+HzMrD4wOH9OBO939pcKOE43nm75hA7Vq/1GarVGzJsuWLt1ju+emTmHSM0+yfft2JjzxNABX9urDXcNup3Tp0tw9YiT333cv/QcMjEas/c77/rvv8PmiBdSv35AbbhpCrdq16XHRJdx6y01kZWVx2x13Mm7sGK7q3Tf/h76kM5dkGy//Pp1hV51JlQplydiWTbuTD+Pzr38EYFivtlzS7lj+t2Ub7QY+VeD+dWtU4OV7L+HQg6pwy9h3WPfzb6T/upWL/+9YPnysF6OenUv7lkfwxdc/su7n36KePwhtrMzx+34RL3mTXTxVAo4Axrv7McBmoB/wiLu3cPcm5HUEOrj7i8BC4BJ3b+bu+QPLZlYNuBU4y92PD2933U7n2Bh+fCx5HQyA24D/uXvT8LlnFeE4+8ULGDko6EOxx8WX8Ppb7zHo79cz4bGxABx51FFMfvZ5Jj41ibVr11C9eg3cnRsGD2LITdfz88aN0YoZUd7Wbdrw5ruzeHH6q5x0yincestNANSuU4eJT01i0tRplClTlp9+Sqdhw0O45eYbuGHwIFau/D7qeYuaGUq2jb9etZH7p37Maw9cysz7erL02w1k5+QCMOzxWRzWbRTPvbuUvn8p+FvQ2vTNnHjFWJpc9DA92zWjRuVy5OTkcvnwlzil1zhemr2ca7qfzEPTPuXe/ucw9a4LaN/yiKhkh2C0sTLH5/tFPOXdF2axuZWEeOoErHH3ueH7k4HTgDZm9pmZLQPaAo33coyTgaOBuWa2GLgMqL/T+pfD/18ENAjfPwsYs2MDd99UhOMAYGa9zWyhmS0saPJLYWrWrMX6devzl9M3bKBGjRqFbt/uvPbMnvXeLo+5O+PHjaVP336Me/QR+vUfQIcOHZk6ZVKRc0Qzb6VKlUlLSwOga7cL+PdXy/c4zuiHR9F/wECmTplE+/bn06//AMY9+kjU8xY1885Kqo2ffv0LTu01jrMHPMmmzRmsWPvLLuuff28ZnVsf/afHWPfzb3z1fTotj9n1JdqnSwsmv7WYkxrXJSs7h57DXuDmS1tFLXtQ2liZ4+/9Ip7yJrt46gTs3n104FGgm7s3BSYAZfZyDAPeDVcImrn70e5+5U7rdwy25vDHUIgVcO69HScvoPt4d2/u7s2vvKr33p9hWOMmTVm9eiVr165he1YWb73xOq3btN1lm1WrVubf//CDORxcf9c3+JmvTKdVq9ZUqFiRjMxMLCUFS0khMwYz7ouS96ef0vPvz5k9i4aHHLrL+oUL5lOjRk3q129AZkYGlpJCSqlSZGRmRj1vUTPHQxtXr1QOgHo1KtKp1VE8/94yDq1bJX99+5ZH8M3qPb/9HFS9AmXS8l7ClcqX4ZSmB/PNmj+2q1S+DOeecjhT3lrCAWVSyc113KF0WvRGAIPSxsocf+8X8ZR3X1iMbiUhLuYEhB1sZqe4+6fARcDHwKnAxvC4fTfgxfC2vwF7XjcF84AxZtbI3VeY2QFAXXf/5k/O+w5wDTAIwMwq7+NxiiwUCjFk6O1c3bsXubk5dO7SlUaNDmPM6Ido3LgJZ7Q9k+emTmbep5+SGgpxYIUKDL/73vz9MzIymDljOo9NeAKASy+7gsGDriU1NZURI++PRsSI806dPIk5s2cRKlWKChUrMvyf9+Tv7+5MGDeWkfc/CEDX7hcy5KbrycnJZuhtw6Ket6iZ46GNnx1+AVUqHsD27BwGjXqdX7dkMvamjhxWrxq57qxe/2v+lQHHH1GHXp2a0+9fMzmifjVG9D8Hd8fMePC5T1j+3R8dsVsub82IZz4E4N3539Kny4ksfOpqHp+xMGrZg9LGyhx/7xfxlHefJNAvCoiLSwTDEwPfAD4k74P/v8BfgVuAHsBKYA2wyt2HmVlX4G6goImBbYF7yZvQB3Cru88MTwxs7u4bzaw5cJ+7nxHuYIwBTiCvQnCnu79c2HEKew6RXCIoyePPLhGMR5FcIigSz2J5ieC3P2XE5P3+0Opli717EU+dgNfCEwADSZ0AKYg6ASIlI5adgO9+yozJ+/0h1cvo9wSIiIhI8YiLOQHuvhIIbBVARESSRyL9FcG46ASIiIgERQL1ATQcICIikqxUCRAREYlEApUCVAkQERFJUqoEiIiIRKCk/uJfLKgSICIikqRUCRAREYmALhEUERFJUgnUB9BwgIiISLJSJUBERCQCiTQcoEqAiIhIklIlQEREJCKJUwpQJ0BERCQCGg4QERGRwFMlQEREJAIJVAjA3L2kM8hemFlvdx9f0jmKKmh5QZmLQ9DyQvAyBy0vBDPzj79mxeSDs06ltGLvX2g4IBh6l3SACAUtLyhzcQhaXghe5qDlhQBmNovNrSRoOEBERCQC+gNCIiIiEniqBARDoMbLCF5eUObiELS8ELzMQcsLQcycOIUATQwUERGJxPrN22PywVmrQmqxdy9UCRAREYlAAhUCNCdAREQkWakTECNmVtXMFodv683sh52W0yI81pNmdkQRt61hZnPMbKuZPRiQzO3M7HMzW2Zmi8zsjABkPsXMloTPs8TMOsZz3p32aRh+bQyKYJ+SauNGZpax07nGxHvm8PbNzGyemS0Pv6ZT4zWvmV2603kWm5mbWZMi7ltSmdPMbHK4bf9tZjdGcq5oSKRLBDUnoBiY2TBgi7vfVwznKg8cCxwHNHL3Ir/Z73acYRRf5uOBde6+zsyOBV5z93r7cJxhFF/mA4Asd882szrAF0Btd8+N4BjDKKa8O51zOpADfOzuEXUSw/sPo/jauBHwors328/jDKP4MqcCnwMXu/syM6sG/BLvr4vweY8Dprn74fuw7zCKr40vBf7P3XuaWTngP8Ap7r421ufe4affsmPywVn9wJB+WVAyMLMbzezL8G1A+LFG4W8Ok8I93OfNrGx43cdm1ix8v334W/MSM3tn92O7+xZ3nwtkBijz5+6+Lry4DChflG9PJZz5d3fPDi+W3XHKeM0b3qYbeW+Y/9mfnMWZORZinPlcYJG7LwNw942RdABKIO/OLgKe3Z+sxZTZgXJmVoq8n71M4Ldo5E5G6gQUMzM7EbgEOBE4BehnZseEVx8NjHH3puS9sPvstm8tYCzQxd2PBXokYOYLgM/cfXu8ZzazU81sObAEuMrdc+I1r5kdCAwG/rGvGYs7c1gjM/vC8oa4Tg1A5sPzNrV3wh9kg+M8745tjbyfvf3uBBRD5ueAbGAdsAoY4e7/29/cEbEY3UqAOgHF73TgpfA3yd+AV4DTwuu+d/d54fuTd3p8h1OA2e6+CsDdfymOwBRTZjNrSt6H1NVByOzun7h7Y+AkYKhFOA5azHmHAyPdfet+ZCzuzGuBg939OOBGYJrlDXfFc+YQ0JK8b9WnAxeaWes4zrvDqcAmd49GlSjWmU8hrwNxEHAIcJOZ1Y9C7qSkSwSL35/193YfZ9p92Qp4rDjEPLOZHQy8DPR09+8ji1fwIf9kXVTb2d2Xm1kWed9yFhd1vwLOWegp9rJclLwnAp3N7AGgEpBrZtvcfWxkMfc4b2H2O7O7ZxIe1nL3+Wa2CmjEvrfxjvMWesq9LBelndcCc9z9ZwAzexM4HvggkpC7nbMw0Xwd9yBKQwHEPvMlwBvhauEGM5sHnEBeVaBY6BJB2R8fAl3MrGz4W00n4KPwuoZm1iJ8/yLg4932nQu03dHrNbMqxRGYGGc2s8rA68D1O31LiPfMDcNjkphZQ/I+nPbnTSimed39VHdv4O4NgEeAu/azAxDzzGZWfac2bkTet7797SDG+ufvTeC48PFDQCvgqzjOS7iNJ2eguQAAAfNJREFUu5FXZo+GWGdeDbQNry9PXiXu6yhlL5JEujpAnYBi5u7zyetxLwDmAWN3TCIClgNXmdlSoBy7/TpNd99AXql8hpktAaYUdA4zWwv8C7jSzNZahJeRlUDmgUBD4E774xKjqnGeuTWw1MwWAy8Cfdx9UxznjbpiyNyGP9p4GnnzLvZr7DfWmcMVgIeBReRVLOa5+9vxmjesDfCtu6/e15zFnPlhoIrlzceZDzzm7sujkT0Z6RLBOGFRuhyqOClz7AUtLyhzcQhaXghm5sL8sjUnJh+cVcqV0iWCIiIiUjxUCRAREYnApt9jUwmofIAqASIiIlJM1AkQERFJUvo9ASIiIhEoqcv5YkGVABERkSSlSoCIiEgELIF+Z6AqASIiIklKlQAREZEIJNKcAHUCREREIpBAfQANB4iIiCQrVQJEREQikUClAFUCREREkpQqASIiIhFIpEsE1QkQERGJQCJdHaDhABERkSSlSoCIiEgEEqgQoEqAiIhIslIlQEREJBIJVApQJ0BERCQCiXR1gIYDREREkpQqASIiIhFIpEsEzd1LOoOIiIiUAA0HiIiIJCl1AkRERJKUOgEiIiJJSp0AERGRJKVOgIiISJJSJ0BERCRJqRMgIiKSpNQJEBERSVLqBIiIiCQpdQJERESS1P8DFH6cRoHtN+MAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 576x576 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df = train_eval.groupby(level='topic').agg('median')\n",
    "fig, ax = plt.subplots(figsize=(8,8))\n",
    "g = sns.heatmap(df, annot=True, fmt='.1%', annot_kws={\"size\": 10}, cmap='Blues', square=True)\n",
    "loc, labels = plt.yticks()\n",
    "g.set_yticklabels(labels, rotation=0)\n",
    "g.set_title('Topic Assignments');\n",
    "\n",
    "df = train_eval\\\n",
    "            .idxmax(axis=1)\\\n",
    "            .reset_index()\\\n",
    "            .groupby('topic', as_index=False)\\\n",
    "            .agg(lambda x:x.value_counts().index[0])\\\n",
    "            .rename(columns={0:'assignment'})\n",
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualization with PyLDAVis\n",
    "\n",
    "#### Lambda\n",
    "\n",
    "- **$\\lambda$ = 0**: how probable is a word to appear in a topic - words are ranked on lift P(word | topic) / P(word)\n",
    "- **$\\lambda$ = 1**: how exclusive is a word to a topic -  words are purely ranked on P(word | topic)\n",
    "\n",
    "The ranking formula is $\\lambda * P(\\text{word} \\vert \\text{topic}) + (1 - \\lambda) * \\text{lift}$\n",
    "\n",
    "User studies suggest $\\lambda = 0.6$ works for most people."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "<link rel=\"stylesheet\" type=\"text/css\" href=\"https://cdn.rawgit.com/bmabey/pyLDAvis/files/ldavis.v1.0.0.css\">\n",
       "\n",
       "\n",
       "<div id=\"ldavis_el1146022842250645362787206531\"></div>\n",
       "<script type=\"text/javascript\">\n",
       "\n",
       "var ldavis_el1146022842250645362787206531_data = {\"mdsDat\": {\"x\": [0.07175606223644311, 0.08068894289840188, 0.10703733789857596, -0.37578474058767836, 0.07484893711937005, 0.05677954053827845, -0.10377635975169079, 0.08845027964829978], \"y\": [-0.12307604802937175, 0.07375227982007691, -0.12683295546513473, -0.015804887909771866, -0.0987446493906184, -0.08860034481376315, 0.055200835792635024, 0.32410576999594803], \"topics\": [1, 2, 3, 4, 5, 6, 7, 8], \"cluster\": [1, 1, 1, 1, 1, 1, 1, 1], \"Freq\": [25.791481326715477, 19.30042039480357, 14.366069036606946, 12.629812628324954, 8.2376094344009, 7.5193590961976575, 7.491409275776115, 4.6638388071743915]}, \"tinfo\": {\"Category\": [\"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Default\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic1\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic2\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic3\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic4\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic5\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic6\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic7\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\", \"Topic8\"], \"Freq\": [20459.0, 13892.0, 7777.0, 6250.0, 4649.0, 4307.0, 3450.0, 3904.0, 7502.0, 3989.0, 4648.0, 2464.0, 3443.0, 2889.0, 5073.0, 2581.0, 3709.0, 2978.0, 3614.0, 1905.0, 2172.0, 3266.0, 2571.0, 2427.0, 1952.0, 2375.0, 2388.0, 2376.0, 2322.0, 3293.0, 5073.101666794062, 3292.8483529286136, 2254.450078115149, 2140.394741859809, 1980.7175793252154, 1836.9087982753401, 1422.3428789264315, 1148.6102812720121, 820.3296857492816, 684.4548649462706, 373.03476960419596, 332.3713396499431, 228.23433193524622, 223.27540857897299, 2898.6448544686814, 202.44793084556485, 197.48898255057216, 498.1645607204922, 2759.77518008278, 133.02304347144738, 128.06409497739864, 124.09673404080841, 257.9612585913149, 1045.5356863355082, 112.19537852766892, 601.7796370778298, 97.31867297133465, 93.35161256364646, 89.38431388381045, 192.42211085300144, 7330.093978655566, 13358.762261243339, 777.7433621723358, 1434.582093501369, 449.18136749497404, 399.195335211002, 4145.944547867875, 828.4239899762076, 1063.2965438523468, 2497.132355070101, 5262.276458484105, 1799.727499513262, 803.3854382400057, 739.0597994425234, 2904.47403741603, 4368.284964568868, 772.0655300728381, 1924.9118630154735, 2086.0958668549038, 1809.2340860194145, 1592.4542314765588, 1821.5975146296346, 2000.7467349377446, 1381.8150881376732, 1092.0282108313236, 1068.4945543785345, 1510.109833203271, 1448.1572120482097, 1317.156353119678, 1260.5429919872195, 1290.936669171758, 1269.4879399386602, 4647.271707938316, 3265.3647992305046, 2460.4958574531065, 1489.479852558859, 596.0651266986436, 241.88276921458117, 154.3319665483857, 150.35304993662928, 149.35790907407016, 133.4396729674675, 128.46540594757306, 126.4754125452783, 81.70535373738974, 81.70532168965053, 79.71524576392022, 78.72062191052136, 78.72034582813788, 76.73066433463278, 75.7360628704754, 73.74626075327335, 63.797077916318045, 60.81247175168179, 58.822744190441206, 113.53931773019255, 114.4520329812711, 50.8639225036437, 128.12326254818484, 49.86886882472185, 49.86870949472781, 48.873970756443875, 2916.0009274106756, 265.79269909727657, 98.67319954520325, 96.60822342442009, 2067.0888829865007, 1182.2511167454536, 913.4453397381118, 845.6003794993501, 691.5412570645694, 225.66694957881882, 816.149009356815, 275.93137876122324, 135.64887611133565, 409.426220825949, 181.41811330877778, 145.91224628570956, 414.49434621544907, 420.30895106620534, 651.2886015769362, 800.5062700851389, 346.72549390755637, 892.6353170452072, 469.5130742171171, 1227.3434249099573, 1276.2832507301325, 731.588513130496, 2036.02556178762, 571.1681087350381, 1002.1293685206545, 2080.0084706163207, 942.4515814888674, 1624.5915665538232, 906.471095929262, 1073.4655162862432, 1103.9188513485428, 950.6999509559131, 680.0255075373972, 730.9594336995324, 878.0743923916061, 1317.6450186572997, 914.218532227504, 1294.0121162634105, 1107.8942205816954, 1150.4786111427152, 890.2251795965267, 937.1267324538204, 898.9987414789437, 893.987331401672, 975.7606970937507, 266.57171390160295, 140.36007736011194, 81.26088438560913, 50.208951170780615, 49.20738907392445, 47.20400032278923, 138.09248945300192, 54.88751214228894, 187.69382527415553, 662.0948438088103, 48.909521449164814, 253.43351751540135, 92.37264467528033, 140.71516360743283, 53.66161024243834, 45.152832417063365, 429.7563720264615, 190.4218147731416, 52.35839320185377, 48.12183114033862, 788.467125995345, 416.5120441198349, 96.71717613335774, 229.6410205232765, 190.94004194943255, 55.874963053883974, 335.525698479379, 481.88015434907135, 362.1185124362962, 884.5021977773919, 462.87423681244496, 1531.8906634458615, 1411.8537673061803, 850.3450605528401, 676.4021136730943, 1206.3847649664845, 280.34564476927096, 12266.06600334665, 913.7512429949627, 596.9740389291155, 420.3734872757406, 3173.3158104890063, 1338.902922727126, 2440.498532493095, 1120.0512919500918, 381.726025998885, 877.4297981388589, 513.4114147220221, 1131.3794832367882, 790.850918769141, 667.381071855458, 2070.8042770095876, 1002.4832473383851, 764.2391033096322, 1076.6496410643858, 659.5898904373474, 1313.09712207395, 1239.9553895335578, 1207.479494708858, 720.3020861525973, 1238.6943680965112, 1057.193171958776, 915.772793847189, 985.8393039198423, 1038.9489128316527, 891.4280774602762, 840.6065540477289, 6249.961724796936, 3442.2659204062206, 2977.1385845121963, 2570.7746684336003, 2387.51245777508, 2321.7771772187784, 1368.6147509519112, 868.627772134419, 846.715947769345, 789.9445624818662, 765.044743321116, 651.5019505925807, 561.8628527649053, 558.8748029048916, 423.42012986514413, 417.4443055976519, 412.46435527331374, 898.2822523303126, 362.66463977506567, 322.82525325269506, 320.83324126870235, 314.85715795695023, 312.86534411835424, 296.9292082968804, 257.08991414004413, 245.1380345629191, 231.19417142135862, 209.28239584888018, 198.32630164998324, 192.3504300289836, 3922.229953075689, 1989.974905694369, 347.53585110354254, 362.6185224096233, 7466.580636450003, 611.3509487810494, 1174.6934804724183, 1335.4456031432835, 716.6068897106268, 617.1913045681586, 833.5814134316888, 753.2292230018073, 1561.57305338355, 2412.5120832155435, 799.3631174268828, 854.4261398421853, 698.5211133443285, 925.5814112358422, 1457.137284473018, 1331.4613220824574, 851.289250924839, 746.4475223099356, 1645.0293498857604, 288.7407813436347, 218.10102792008678, 184.79926653733546, 94.98545941698515, 82.87555454041134, 73.79332236024256, 49.57397317469463, 145.41778577053938, 52.35213052902089, 90.41832823138272, 383.33596320400454, 193.20307927843515, 76.56125117912508, 55.02911081047401, 59.740619714654436, 93.44486509598033, 136.3965417580455, 150.97622055589693, 2054.745941163401, 45.360304490235876, 93.34466713097312, 336.80936958423626, 50.535499795280415, 59.61821305457535, 72.45029152828123, 67.46876761213498, 37.74974275936271, 37.33413621293409, 401.4516466373638, 244.76872424930247, 220.4157451172945, 396.1495201196217, 945.9880928141953, 8192.979481272554, 855.7919064704615, 227.054567911945, 620.0353242872726, 537.6104049229572, 495.1613122143333, 275.64544893861955, 809.623354930588, 1109.2227198905089, 286.12148389606267, 949.8833774344795, 677.4233599772822, 652.8110889712062, 518.9640295039585, 784.731794896608, 498.73270182004023, 330.986329921881, 416.0968386527397, 682.5278104894737, 526.6287735649239, 553.8482169614194, 447.8319560455208, 382.22608213642053, 316.81957092230533, 362.0117789336048, 338.13306495150806, 362.4983763319009, 362.9979436951356, 353.36180625315245, 342.1559133001972, 4306.581416085522, 2580.620392742128, 1951.6430821168597, 1640.6321014941589, 1171.6317160309113, 738.4024708084457, 634.0696760015951, 354.8554742091913, 345.9126027727912, 340.9440628335647, 262.4463607169936, 232.6370931292875, 192.89131991515802, 184.94207146857303, 173.0184395916659, 173.01841515958668, 153.14554071021584, 140.2281537431641, 132.278993303057, 131.28520954771636, 128.30409622351652, 123.33616715892549, 120.3549561279039, 118.36784469098652, 117.37365500427013, 112.40591153733716, 109.42478564919696, 108.43146776532795, 106.44421721052767, 105.45056502315039, 378.3539928195161, 1944.393691214454, 425.14570603468627, 2539.8338268811594, 525.2940025675864, 328.5089150132579, 1511.4027543661778, 469.09865263900554, 535.601143095772, 296.03346437948113, 1145.553615821127, 299.09611254685854, 499.2798177053468, 1063.8074923881775, 447.45332789416193, 429.22941937495625, 567.3406940988737, 378.82602065132926, 599.5159543076807, 684.8151317609376, 426.95079448670043, 529.9758382100531, 469.09134556528346, 431.39206848242543, 438.7221818039654, 417.9045945282471, 3903.593944058445, 2171.8148251611346, 1416.4008161548186, 909.8058492105233, 600.2751124348678, 442.02658862102385, 398.2343803115743, 396.24394412966956, 358.4232414180661, 309.65500981438163, 265.86289845638464, 259.8912445667572, 241.97628884540637, 208.13655899405597, 172.3070237738076, 167.3306098933198, 153.39678127652232, 138.46766449594668, 131.50054557972493, 129.51012875787595, 126.52406799523722, 112.59026454488145, 112.59013142522274, 457.731017537257, 93.68027946826153, 82.73191190850847, 81.73679894358574, 79.74591860541861, 70.788835930596, 66.80779235910019, 340.9223800278022, 919.9166509578733, 221.3911864037126, 626.5646800096303, 2693.858145866417, 1187.5478967469871, 2088.2221807002725, 256.5002490498573, 225.3955987868245, 887.6809191699297, 194.34288753588675, 569.5707692071942, 766.924836978328, 367.75057877958403, 447.64142798993, 263.1646451271351, 397.66068392872035, 546.2658462170649, 299.1816306912245, 321.89793559596893, 444.81879807788545, 617.2667292774015, 467.4458400956687, 370.996053640803, 310.1035885869321, 321.54939225881026, 354.9348818647189, 400.675173327872, 386.5005559320401, 380.1903433450125, 2464.07059176197, 1904.223528163824, 1339.2495925337182, 954.7392054071638, 704.551103852907, 606.1164898228498, 587.6597492008653, 551.7723024480739, 339.52258295201847, 338.4972241759402, 294.4067216966395, 290.3051391107572, 286.2038452231713, 179.56627945297944, 124.19678800467918, 123.17141808546144, 108.81636753762002, 94.46132432914459, 90.35973674369728, 80.1062656055492, 67.80193796088408, 50.37079264089971, 3283.128967310615, 1364.8648996433133, 3705.1800649257802, 35.24724441335786, 31.588466947516302, 454.04046948328784, 129.84054552590277, 27.797401752187533, 1037.2185864310748, 1373.2865648699226, 71.87121138008212, 327.8420056340192, 387.2798201767673, 167.36996147117495, 212.28303999959135, 250.757531493991, 178.87260755042112, 188.8086377903327, 214.09915073554043, 154.0703797140462, 145.75546284545396, 150.40043234864902], \"Term\": [\"computer\", \"book\", \"building\", \"dirigible\", \"caps3\", \"mood\", \"caps4\", \"cyclist\", \"library\", \"obstacle\", \"laugh\", \"caps5\", \"empire\", \"water\", \"offensive\", \"memoir\", \"love\", \"mast\", \"person1\", \"caps6\", \"setting\", \"patient\", \"builder\", \"affect\", \"narcs\", \"spend\", \"dock\", \"create\", \"mooring\", \"shelf\", \"offensive\", \"shelf\", \"sang\", \"remove\", \"hibiscus\", \"censorship\", \"censor\", \"offend\", \"geese\", \"bud\", \"vow\", \"violence\", \"katherine\", \"paterson\", \"material\", \"librarian\", \"vietnam\", \"melt\", \"magazine\", \"retake\", \"sangs\", \"mining\", \"literature\", \"spring\", \"genre\", \"freedom\", \"profanity\", \"slavery\", \"uncensored\", \"ho\", \"library\", \"book\", \"flower\", \"conclude\", \"appropriate\", \"medium\", \"movie\", \"section\", \"plant\", \"test\", \"read\", \"certain\", \"winter\", \"inappropriate\", \"music\", \"child\", \"public\", \"month1\", \"right\", \"believe\", \"different\", \"person\", \"kid\", \"watch\", \"listen\", \"age\", \"paragraph\", \"bad\", \"world\", \"try\", \"reason\", \"feel\", \"laugh\", \"patient\", \"laughter\", \"patience\", \"joke\", \"bus\", \"boyfriend\", \"giggle\", \"ice\", \"asleep\", \"patiently\", \"stomach\", \"funniest\", \"pant\", \"deer\", \"shower\", \"cream\", \"coaster\", \"roller\", \"laughing\", \"tolerant\", \"wet\", \"puppy\", \"impatient\", \"shot\", \"dentist\", \"clock\", \"van\", \"skate\", \"pee\", \"wait\", \"ibm\", \"hug\", \"hilarious\", \"mom\", \"dad\", \"brother\", \"funny\", \"smile\", \"seat\", \"sister\", \"wake\", \"truck\", \"cry\", \"dance\", \"red\", \"moment\", \"time1\", \"relationship\", \"minute\", \"team\", \"line\", \"night\", \"sit\", \"hour\", \"car\", \"start\", \"didn\", \"walk\", \"tell\", \"date1\", \"person1\", \"finally\", \"long\", \"little\", \"ask\", \"girl\", \"ride\", \"fun\", \"feel\", \"year\", \"look\", \"try\", \"talk\", \"hard\", \"school\", \"play\", \"lot\", \"coordination\", \"faraway\", \"google\", \"keyboard\", \"textbook\", \"earthquake\", \"skype\", \"typing\", \"dictionary\", \"cam\", \"communicate\", \"reconnect\", \"program\", \"update\", \"assignment\", \"efficient\", \"convenient\", \"communication\", \"tool\", \"application\", \"easier\", \"research\", \"email\", \"engine\", \"improve\", \"instant\", \"handy\", \"mail\", \"helpful\", \"web\", \"benefit\", \"touch\", \"information\", \"online\", \"website\", \"chat\", \"internet\", \"useful\", \"computer\", \"positive\", \"project\", \"search\", \"help\", \"hand\", \"learn\", \"society\", \"business\", \"easy\", \"paper\", \"eye\", \"student\", \"technology\", \"talk\", \"type\", \"teach\", \"effect\", \"job\", \"place\", \"world\", \"school\", \"dear\", \"reason\", \"great\", \"game\", \"work\", \"look\", \"allow\", \"lot\", \"dirigible\", \"empire\", \"mast\", \"builder\", \"dock\", \"mooring\", \"hydrogen\", \"flammable\", \"york\", \"airship\", \"urban\", \"blimp\", \"helium\", \"architect\", \"populated\", \"tether\", \"moor\", \"frame\", \"pedestrian\", \"hindenburg\", \"smith\", \"modify\", \"densely\", \"steel\", \"modification\", \"marcia\", \"docking\", \"framework\", \"german\", \"transmit\", \"obstacle\", \"wind\", \"landing\", \"swivel\", \"building\", \"shift\", \"attempt\", \"law\", \"highly\", \"ship\", \"safety\", \"current\", \"area\", \"face\", \"stress\", \"low\", \"violent\", \"fly\", \"allow\", \"problem\", \"air\", \"nature\", \"exercise\", \"addict\", \"obesity\", \"obese\", \"addiction\", \"virus\", \"addicted\", \"predator\", \"unhealthy\", \"cuber\", \"overweight\", \"health\", \"lazy\", \"excessive\", \"outdoors\", \"bullying\", \"bully\", \"fat\", \"active\", \"spend\", \"distraction\", \"interaction\", \"healthy\", \"spending\", \"outdoor\", \"usage\", \"distract\", \"depression\", \"suicide\", \"interact\", \"screen\", \"activity\", \"negative\", \"outside\", \"computer\", \"nature\", \"waste\", \"enjoy\", \"dear\", \"instead\", \"social\", \"play\", \"kid\", \"percent1\", \"bad\", \"effect\", \"game\", \"hour\", \"reason\", \"sit\", \"technology\", \"stay\", \"child\", \"lot\", \"talk\", \"problem\", \"online\", \"newspaper\", \"fun\", \"internet\", \"away\", \"person\", \"world\", \"month1\", \"mood\", \"memoir\", \"narcs\", \"grateful\", \"rodriguez\", \"cuban\", \"sacrifice\", \"cuba\", \"immigrant\", \"cooking\", \"loving\", \"gratitude\", \"gratefulness\", \"passionate\", \"blueprints\", \"celebration\", \"selflessly\", \"culinary\", \"solidarity\", \"congregation\", \"warmth\", \"backdrop\", \"instill\", \"admiration\", \"perseverance\", \"courageous\", \"caring\", \"eternally\", \"generosity\", \"encompass\", \"thankful\", \"create\", \"neighborhood\", \"love\", \"happiness\", \"childhood\", \"happy\", \"warm\", \"describe\", \"blood\", \"house\", \"kitchen\", \"culture\", \"paragraph\", \"simple\", \"forget\", \"live\", \"care\", \"great\", \"talk\", \"grow\", \"feel\", \"tell\", \"example\", \"music\", \"child\", \"cyclist\", \"setting\", \"feature\", \"desert\", \"heat\", \"terrain\", \"california\", \"june\", \"snake\", \"dehydrate\", \"kurmaskie\", \"yosemite\", \"rolling\", \"factory\", \"dehydration\", \"dehydrated\", \"joe\", \"heatstroke\", \"pedal\", \"thirst\", \"battery\", \"acid\", \"cycle\", \"exceed\", \"tumbleweed\", \"ridiculously\", \"rider\", \"stroke\", \"biker\", \"bumpy\", \"journey\", \"hill\", \"thirsty\", \"rough\", \"water\", \"road\", \"affect\", \"pump\", \"abandon\", \"hot\", \"juice\", \"speed\", \"town\", \"ahead\", \"post\", \"dry\", \"bike\", \"limit\", \"sun\", \"direction\", \"man\", \"old\", \"ride\", \"rid\", \"drink\", \"travel\", \"set\", \"example\", \"effect\", \"feel\", \"caps5\", \"caps6\", \"caps7\", \"caps8\", \"caps9\", \"caps11\", \"person3\", \"caps10\", \"caps12\", \"caps13\", \"caps14\", \"person4\", \"caps15\", \"caps16\", \"caps18\", \"caps17\", \"caps21\", \"caps19\", \"person5\", \"caps20\", \"caps22\", \"caps24\", \"caps4\", \"person2\", \"caps3\", \"glance\", \"hello\", \"organization2\", \"reply\", \"said\", \"organization1\", \"person1\", \"organization3\", \"ask\", \"look\", \"location2\", \"date1\", \"location1\", \"man\", \"walk\", \"love\", \"begin\", \"run\", \"face\"], \"Total\": [20459.0, 13892.0, 7777.0, 6250.0, 4649.0, 4307.0, 3450.0, 3904.0, 7502.0, 3989.0, 4648.0, 2464.0, 3443.0, 2889.0, 5073.0, 2581.0, 3709.0, 2978.0, 3614.0, 1905.0, 2172.0, 3266.0, 2571.0, 2427.0, 1952.0, 2375.0, 2388.0, 2376.0, 2322.0, 3293.0, 5073.978725993901, 3293.725538811122, 2255.327156399382, 2141.2719406025963, 1981.594632212334, 1837.7858803407235, 1423.2199563179424, 1149.4874334073447, 821.2067591182401, 685.3322871920221, 373.91194691787774, 333.24878776655976, 229.11140999530477, 224.15248745821899, 2910.944949974289, 203.32501285985015, 198.36608884504238, 500.8660680123521, 2775.2033342051163, 133.90009566893102, 128.9411731718527, 124.97404092779713, 259.8745551536936, 1053.3058830863104, 113.07262403249102, 607.0138622289514, 98.1958564263114, 94.22871788771802, 90.26158191545102, 194.40282747546001, 7502.570507409226, 13892.23695716039, 795.5911867201871, 1491.1094460467057, 460.3497781447353, 408.7020906804571, 4475.9679254498915, 869.9162402454757, 1129.1685304322857, 2761.968429907069, 6089.034914588303, 1992.0120979265437, 857.1854480213874, 795.3720708590129, 3584.901166428925, 5731.350127661245, 871.5342586813721, 2905.482741744743, 3285.3424383666998, 3101.000349285713, 2638.3031359225133, 3448.5109708034083, 4054.6430447059497, 2215.78364456789, 1493.1170098274035, 1448.305937700689, 3100.6173364298097, 3786.1867388594383, 3341.4036936545363, 3338.5001854545308, 4305.799202738168, 3856.6710465938854, 4648.148688812487, 3266.2415634526787, 2461.3726929089767, 1490.3565993148843, 596.9422055014235, 242.76011164278972, 155.20948397735188, 151.2299019766754, 150.23500953333746, 134.31671415748292, 129.34224404410077, 127.35245905873451, 82.58224832792972, 82.58224785018547, 80.59246116566224, 79.5975687365083, 79.59757225800003, 77.60777823481764, 76.61288487981399, 74.62309725978427, 64.67415907967687, 61.68948455898733, 59.699697478630526, 115.4148326542966, 116.41598940314495, 51.740542126154146, 130.35628777858543, 50.745651128568596, 50.74565239068495, 49.75075980851345, 2982.2764613601203, 271.61391687839927, 100.51912131697047, 98.50093489465002, 2174.8072444886593, 1240.1490321826093, 957.5067555514395, 895.4971049394, 732.6826112205274, 233.9635370348072, 883.5026859924665, 298.04244304586143, 141.39526365955825, 460.9327172156984, 193.05504338277638, 153.3117895096495, 471.60231554270985, 486.1110376578947, 791.9870968070965, 1022.1616310057346, 403.415479959698, 1180.7138418468578, 588.9128742932428, 1866.3319018062416, 1970.9478696865674, 1034.6301901808768, 3685.3525935566145, 775.9586667316913, 1583.0918077216431, 3977.532308114809, 1494.9513972763366, 3614.935075584916, 1612.3417964738182, 2093.532557348062, 2229.2458757624004, 1807.944100964003, 1090.0343061883796, 1240.660025582234, 1743.850628673285, 3856.6710465938854, 1957.5291030302274, 4309.1138235342505, 3338.5001854545308, 4974.16536004499, 2061.555659006628, 3250.7380264252447, 2519.9086299514597, 3766.3487487952502, 976.6365930432929, 267.4476621620101, 141.2360733209967, 82.1369968915374, 51.084938214774155, 50.08325687085437, 48.07989899181, 143.27125658558478, 57.12648080051445, 195.37709695083754, 691.2569560988017, 51.07613735774226, 266.4507066134153, 97.16687947498407, 149.43055490154714, 57.075234520252955, 48.06602624579671, 467.0748534489015, 207.3105672609345, 57.154490548273266, 53.06174808929041, 871.8937028819693, 461.0950367209748, 107.11404620080529, 257.38297570100843, 214.49166506636934, 63.08263158480297, 384.74494336539146, 557.6843818462672, 419.11632985330573, 1056.8128702200115, 551.326346533645, 1944.902759624704, 1794.8296788511634, 1061.3300155682543, 834.610536961713, 1562.9393320502186, 330.52830481095384, 20459.795200398665, 1222.8604643808847, 766.1836878230796, 530.166296884258, 5199.401085294033, 1990.8337920148565, 4102.704781060302, 1732.1611701952793, 480.8856255567184, 1330.2284303679698, 700.3385282417499, 1900.9958402386626, 1239.9266690765198, 999.1173400979749, 4974.16536004499, 1818.1383449637474, 1312.5825303595689, 2263.2714783365136, 1070.5658393280326, 3370.704269228662, 3341.4036936545363, 3250.7380264252447, 1271.967949342838, 4305.799202738168, 3172.851337310015, 2314.76869061123, 3112.7966206138267, 4309.1138235342505, 3129.5303711408837, 3766.3487487952502, 6250.83820202156, 3443.1424007948, 2978.0150722977087, 2571.6511535420173, 2388.3889947048892, 2322.6536543030643, 1369.4912272754746, 869.5042488455525, 847.5924689858278, 790.8210390937926, 765.9212892910095, 652.3784295461738, 562.7393298127068, 559.7513603106115, 424.29672046207196, 418.3207802196681, 413.3408302435654, 900.4159452845419, 363.54133198851804, 323.7017305583563, 321.7097507689828, 315.73381181928437, 313.741830541211, 297.80599286066087, 257.96639074214556, 246.01451077327815, 232.07065087178168, 210.1588708595797, 199.20298231680422, 193.2270411625753, 3989.89294750287, 2016.1938993565434, 349.6324180254131, 365.53107510584954, 7777.468239178764, 620.7508854303086, 1232.0192686791208, 1497.7480262639062, 764.8785966590664, 652.4888044015033, 907.1263680172485, 819.1169458494967, 1988.3497878688866, 3492.7021037515733, 961.5816820392641, 1048.0396834630728, 816.2214137341691, 1182.222195523539, 3129.5303711408837, 2804.4262889213687, 1384.472250938791, 1677.2295013931591, 1645.9045077313233, 289.61594682986396, 218.9759189138996, 185.67419096106894, 95.8604379865028, 83.75071744753694, 74.66842996169208, 50.4489930308857, 151.31159716312442, 54.476256622565835, 94.90846774511121, 402.7142127456313, 202.97026517872268, 80.67811953690929, 58.50277992387664, 63.51707123791839, 99.84335718435409, 146.30670370351666, 162.3853984661983, 2375.5476693458486, 53.35977567674401, 109.85246367011906, 398.8551982598279, 61.41511690530743, 72.48170417676866, 103.55814990084093, 96.44841060753618, 54.23387144809367, 54.216141746519476, 594.9600980471673, 365.7471834046142, 334.62889256749816, 626.4951856883715, 1693.032309534904, 20459.795200398665, 1677.2295013931591, 370.32653224672856, 1463.2656220013564, 1271.967949342838, 1211.9819502294279, 543.9456250370436, 2519.9086299514597, 4054.6430447059497, 597.0527589907489, 3786.1867388594383, 2263.2714783365136, 2314.76869061123, 1970.9478696865674, 4305.799202738168, 1866.3319018062416, 999.1173400979749, 1719.4252372565186, 5731.350127661245, 3766.3487487952502, 4974.16536004499, 2804.4262889213687, 1794.8296788511634, 926.0870720127391, 1743.850628673285, 1562.9393320502186, 2875.9116214937976, 3448.5109708034083, 3341.4036936545363, 2905.482741744743, 4307.458632013637, 2581.4971721220686, 1952.519853977148, 1641.5089534997212, 1172.508488071662, 739.2792454159177, 634.946515897159, 355.73225535094804, 346.78945073015024, 341.8212294552371, 263.3232678886011, 233.51391385429858, 193.76811171562167, 185.81895174306996, 173.89521067104712, 173.89521072097912, 154.0223095939061, 141.10492395639793, 133.1557635398445, 132.16212084242747, 129.18118652076498, 124.21295829136425, 121.23202373173761, 119.24473420673203, 118.25108948999838, 113.28286339748136, 110.30193086777636, 109.30828232899532, 107.32099212310538, 106.32734709506774, 387.5393308359039, 2376.8255609700154, 485.3762634121261, 3709.2391009821213, 649.0117047117176, 388.7030651436482, 2543.5710711872985, 633.7884112291616, 758.3669497233822, 372.21072137458003, 2421.1196112204943, 397.07118644232963, 811.9743204027144, 3100.6173364298097, 808.3194522479494, 1024.4122326445843, 2126.1258249930593, 909.8261567744377, 3172.851337310015, 4974.16536004499, 1337.8804869285648, 3856.6710465938854, 3977.532308114809, 2746.2129990419267, 3584.901166428925, 5731.350127661245, 3904.470512069116, 2172.6915734245626, 1417.2776584440833, 910.682554348804, 601.1519514845193, 442.9031836763744, 399.11107359161525, 397.1205219969864, 359.30006679856024, 310.53157506644305, 266.7394639578465, 260.76781245617354, 242.85285793895142, 209.0135036725734, 173.18359022404894, 168.20721478787837, 154.27336042442425, 139.34423163861237, 132.37730791741083, 130.38675513305208, 127.40093068814997, 113.46707888641565, 113.46707908196899, 461.8580815589956, 94.55684525763503, 83.60882318663927, 82.61354520618347, 80.6229996716654, 71.66551405720357, 67.6844141996709, 347.5247496697807, 949.4872862416153, 225.93177892998276, 649.9118234329867, 2889.2151439707964, 1315.4741162166774, 2427.9888104949437, 267.73078133975434, 236.95555682424, 1047.6884079328568, 204.0337444246534, 685.0703784263059, 1049.5342588101278, 438.22746880221564, 563.3852607707654, 293.6005597394819, 564.9413592973957, 943.3156941596399, 415.5101010659825, 486.7930865559047, 1038.3742467194122, 2320.0737805939866, 1240.660025582234, 702.3294845264072, 458.78612402252173, 570.572016145695, 1080.8396994650461, 2746.2129990419267, 2263.2714783365136, 3856.6710465938854, 2464.94385177644, 1905.0966560120437, 1340.1226528876489, 955.6122154144463, 705.4240904295457, 606.9894195798837, 588.5329124539927, 552.64527646232, 340.39551512354456, 339.37015391288116, 295.27962445898265, 291.1781762973052, 287.0767354053105, 180.43917359833372, 125.0696707983593, 124.04430934594022, 109.68925313874402, 95.33419706886646, 91.23274905576433, 80.97914061968486, 68.6748068666663, 51.24366668627923, 3450.8542237430706, 1507.0038281871, 4649.370655765774, 53.76083077254738, 53.73129010942155, 784.8913889766428, 242.74600959167958, 52.56248432420436, 2571.3680953007374, 3614.935075584916, 153.8385867462375, 1807.944100964003, 4309.1138235342505, 782.7579843483426, 1494.9513972763366, 2498.985608538947, 1038.3742467194122, 1583.0918077216431, 3709.2391009821213, 1026.9311098335083, 1452.0108929703892, 3492.7021037515733], \"loglift\": [30.0, 29.0, 28.0, 27.0, 26.0, 25.0, 24.0, 23.0, 22.0, 21.0, 20.0, 19.0, 18.0, 17.0, 16.0, 15.0, 14.0, 13.0, 12.0, 11.0, 10.0, 9.0, 8.0, 7.0, 6.0, 5.0, 4.0, 3.0, 2.0, 1.0, 1.355, 1.3549, 1.3547, 1.3547, 1.3547, 1.3546, 1.3545, 1.3544, 1.3541, 1.3538, 1.3528, 1.3525, 1.3513, 1.3512, 1.3509, 1.3508, 1.3507, 1.3497, 1.3496, 1.3486, 1.3483, 1.3481, 1.3477, 1.3477, 1.3473, 1.3465, 1.3462, 1.3458, 1.3454, 1.3449, 1.3319, 1.316, 1.3324, 1.3165, 1.3306, 1.3316, 1.2785, 1.3063, 1.295, 1.2543, 1.2092, 1.2536, 1.2903, 1.2817, 1.1446, 1.0835, 1.2339, 0.9434, 0.9009, 0.8163, 0.8503, 0.7169, 0.6488, 0.8829, 1.0423, 1.051, 0.6357, 0.3941, 0.4242, 0.3811, 0.1505, 0.2439, 1.6449, 1.6448, 1.6447, 1.6445, 1.6436, 1.6414, 1.6394, 1.6392, 1.6392, 1.6385, 1.6382, 1.6381, 1.6344, 1.6344, 1.6341, 1.634, 1.634, 1.6337, 1.6335, 1.6332, 1.6314, 1.6307, 1.6302, 1.6287, 1.628, 1.628, 1.6278, 1.6276, 1.6276, 1.6273, 1.6226, 1.6234, 1.6265, 1.6256, 1.5942, 1.5972, 1.5979, 1.5877, 1.5873, 1.6089, 1.5657, 1.568, 1.6036, 1.5265, 1.5829, 1.5956, 1.516, 1.4996, 1.4495, 1.4006, 1.4936, 1.3653, 1.4185, 1.2259, 1.2105, 1.2985, 1.0517, 1.3386, 1.1878, 0.9968, 1.1837, 0.8452, 1.0692, 0.9771, 0.9422, 1.0023, 1.1732, 1.116, 0.9589, 0.5711, 0.8837, 0.4421, 0.542, 0.181, 0.8053, 0.4012, 0.6143, 0.2069, 1.9394, 1.937, 1.9341, 1.9296, 1.923, 1.9227, 1.9219, 1.9035, 1.9003, 1.9002, 1.8972, 1.897, 1.8902, 1.8897, 1.8802, 1.8786, 1.8778, 1.857, 1.8553, 1.8527, 1.8426, 1.8397, 1.8386, 1.8382, 1.8263, 1.824, 1.819, 1.8034, 1.7942, 1.7941, 1.7623, 1.7654, 1.7016, 1.7003, 1.7187, 1.7301, 1.6814, 1.7756, 1.4287, 1.6489, 1.6908, 1.7083, 1.4465, 1.5436, 1.4209, 1.5043, 1.7094, 1.5242, 1.6298, 1.4214, 1.4906, 1.5368, 1.064, 1.345, 1.3994, 1.1973, 1.456, 0.9976, 0.949, 0.95, 1.3717, 0.6944, 0.8413, 1.013, 0.7905, 0.5178, 0.6845, 0.4406, 2.069, 2.0689, 2.0688, 2.0688, 2.0687, 2.0687, 2.0685, 2.0681, 2.0681, 2.068, 2.068, 2.0678, 2.0676, 2.0675, 2.067, 2.067, 2.067, 2.0667, 2.0667, 2.0664, 2.0664, 2.0663, 2.0663, 2.0662, 2.0657, 2.0655, 2.0653, 2.0649, 2.0647, 2.0646, 2.052, 2.056, 2.0631, 2.0611, 2.0283, 2.0539, 2.0215, 1.9544, 2.0039, 2.0135, 1.9846, 1.9853, 1.8275, 1.6991, 1.8843, 1.8649, 1.9134, 1.8244, 1.3047, 1.3242, 1.5828, 1.2595, 2.4959, 2.4934, 2.4925, 2.4917, 2.4873, 2.486, 2.4847, 2.479, 2.4567, 2.4567, 2.448, 2.4471, 2.4471, 2.4441, 2.4352, 2.4352, 2.4302, 2.4263, 2.4236, 2.3514, 2.334, 2.3336, 2.3274, 2.3015, 2.3011, 2.1392, 2.1391, 2.1341, 2.1234, 2.1031, 2.0948, 2.079, 2.0381, 1.9144, 1.5813, 1.8236, 2.0073, 1.6378, 1.6353, 1.6013, 1.8167, 1.3611, 1.2003, 1.7609, 1.1137, 1.2902, 1.2307, 1.162, 0.7941, 1.1768, 1.3917, 1.0776, 0.3686, 0.5291, 0.3013, 0.6619, 0.9498, 1.4238, 0.9243, 0.9656, 0.4254, 0.2452, 0.2498, 0.3574, 2.5875, 2.5873, 2.5872, 2.5872, 2.5869, 2.5865, 2.5863, 2.5852, 2.5852, 2.5851, 2.5844, 2.5839, 2.5832, 2.583, 2.5826, 2.5826, 2.582, 2.5815, 2.5811, 2.581, 2.5809, 2.5806, 2.5804, 2.5803, 2.5802, 2.5799, 2.5797, 2.5796, 2.5795, 2.5794, 2.5637, 2.3869, 2.4552, 2.209, 2.3762, 2.4194, 2.0672, 2.2868, 2.2399, 2.3587, 1.8393, 2.3043, 2.1014, 1.5179, 1.9963, 1.7178, 1.2666, 1.7115, 0.9214, 0.6048, 1.4455, 0.603, 0.4501, 0.7367, 0.4871, -0.0308, 2.5912, 2.591, 2.5908, 2.5905, 2.59, 2.5894, 2.5892, 2.5892, 2.589, 2.5886, 2.5881, 2.588, 2.5878, 2.5872, 2.5863, 2.5862, 2.5857, 2.5851, 2.5848, 2.5847, 2.5845, 2.5837, 2.5837, 2.5824, 2.5821, 2.5809, 2.5807, 2.5805, 2.5791, 2.5784, 2.5722, 2.5598, 2.5711, 2.5548, 2.5214, 2.4891, 2.4407, 2.5486, 2.5414, 2.4257, 2.5428, 2.4068, 2.2777, 2.4161, 2.3614, 2.482, 2.2403, 2.0451, 2.263, 2.1778, 1.7437, 1.2674, 1.6153, 1.9532, 2.1997, 2.0179, 1.4779, 0.6666, 0.824, 0.2745, 3.065, 3.0649, 3.0647, 3.0644, 3.0641, 3.0639, 3.0638, 3.0638, 3.0628, 3.0628, 3.0624, 3.0623, 3.0623, 3.0605, 3.0583, 3.0583, 3.0573, 3.0561, 3.0557, 3.0545, 3.0525, 3.0482, 3.0155, 2.9663, 2.8383, 2.6432, 2.5341, 2.518, 2.4396, 2.4283, 2.1574, 2.0975, 2.3043, 1.3579, 0.656, 1.5227, 1.1134, 0.7662, 1.3066, 0.9389, 0.2132, 1.1684, 0.7666, -0.0798], \"logprob\": [30.0, 29.0, 28.0, 27.0, 26.0, 25.0, 24.0, 23.0, 22.0, 21.0, 20.0, 19.0, 18.0, 17.0, 16.0, 15.0, 14.0, 13.0, 12.0, 11.0, 10.0, 9.0, 8.0, 7.0, 6.0, 5.0, 4.0, 3.0, 2.0, 1.0, -3.7754, -4.2076, -4.5864, -4.6383, -4.7159, -4.7912, -5.047, -5.2608, -5.5974, -5.7785, -6.3854, -6.5008, -6.8767, -6.8987, -4.3351, -6.9966, -7.0214, -6.0962, -4.3842, -7.4166, -7.4546, -7.486, -6.7543, -5.3548, -7.5868, -5.9072, -7.7291, -7.7707, -7.8141, -7.0474, -3.4073, -2.8072, -5.6507, -5.0385, -6.1997, -6.3176, -3.9772, -5.5876, -5.338, -4.4842, -3.7388, -4.8117, -5.6182, -5.7017, -4.3331, -3.925, -5.658, -4.7444, -4.664, -4.8064, -4.9341, -4.7996, -4.7058, -5.0759, -5.3113, -5.3331, -4.9871, -5.029, -5.1239, -5.1678, -5.144, -5.1607, -3.5731, -3.926, -4.209, -4.711, -5.6268, -6.5287, -6.9781, -7.0042, -7.0108, -7.1235, -7.1615, -7.1771, -7.614, -7.614, -7.6387, -7.6513, -7.6513, -7.6769, -7.6899, -7.7165, -7.8615, -7.9094, -7.9426, -7.285, -7.277, -8.088, -7.1642, -8.1078, -8.1078, -8.1279, -4.0392, -6.4344, -7.4254, -7.4465, -4.3833, -4.942, -5.1999, -5.2771, -5.4782, -6.5981, -5.3126, -6.397, -7.1071, -6.0024, -6.8164, -7.0342, -5.9901, -5.9762, -5.5382, -5.3319, -6.1686, -5.223, -5.8655, -4.9046, -4.8655, -5.4219, -4.3984, -5.6695, -5.1073, -4.377, -5.1687, -4.6242, -5.2076, -5.0385, -5.0105, -5.16, -5.495, -5.4228, -5.2394, -4.8336, -5.1991, -4.8517, -5.0069, -4.9692, -5.2257, -5.1743, -5.2159, -5.2215, -4.8387, -6.1363, -6.7777, -7.3242, -7.8057, -7.8259, -7.8674, -6.794, -7.7166, -6.4871, -5.2265, -7.8319, -6.1868, -7.1961, -6.7752, -7.7392, -7.9119, -5.6587, -6.4727, -7.7638, -7.8482, -5.0518, -5.69, -7.1501, -6.2854, -6.4699, -7.6988, -5.9062, -5.5442, -5.8299, -4.9369, -5.5845, -4.3876, -4.4692, -4.9763, -5.2051, -4.6265, -6.0859, -2.3073, -4.9043, -5.33, -5.6808, -3.6594, -4.5223, -3.9219, -4.7008, -5.7772, -4.9449, -5.4808, -4.6907, -5.0488, -5.2185, -4.0862, -4.8117, -5.083, -4.7403, -5.2303, -4.5418, -4.5991, -4.6256, -5.1422, -4.6001, -4.7585, -4.9021, -4.8284, -4.7759, -4.9291, -4.9878, -2.8528, -3.4492, -3.5944, -3.7411, -3.8151, -3.843, -4.3715, -4.8262, -4.8517, -4.9211, -4.9532, -5.1138, -5.2618, -5.2672, -5.5447, -5.5589, -5.5709, -4.7926, -5.6996, -5.816, -5.8222, -5.841, -5.8473, -5.8996, -6.0437, -6.0913, -6.1498, -6.2494, -6.3032, -6.3338, -3.3187, -3.9972, -5.7422, -5.6997, -2.6749, -5.1774, -4.5243, -4.3961, -5.0186, -5.1679, -4.8674, -4.9687, -4.2396, -3.8047, -4.9093, -4.8427, -5.0441, -4.7627, -4.3089, -4.3991, -4.8463, -4.9778, -3.7602, -5.5002, -5.7808, -5.9465, -6.612, -6.7484, -6.8645, -7.2623, -6.1861, -7.2078, -6.6613, -5.2168, -5.902, -6.8277, -7.1579, -7.0757, -6.6284, -6.2502, -6.1486, -3.5378, -7.3511, -6.6294, -5.3462, -7.2431, -7.0778, -6.8828, -6.9541, -7.5348, -7.5458, -5.1707, -5.6654, -5.7702, -5.184, -4.3135, -2.1547, -4.4137, -5.7406, -4.736, -4.8786, -4.9609, -5.5466, -4.4692, -4.1543, -5.5093, -4.3094, -4.6475, -4.6845, -4.9139, -4.5004, -4.9537, -5.3637, -5.1348, -4.6399, -4.8993, -4.8489, -5.0613, -5.2197, -5.4074, -5.2741, -5.3423, -5.2727, -5.2714, -5.2983, -5.3305, -2.7066, -3.2187, -3.4981, -3.6717, -4.0084, -4.47, -4.6224, -5.2028, -5.2283, -5.2428, -5.5045, -5.625, -5.8124, -5.8545, -5.9211, -5.9211, -6.0431, -6.1312, -6.1896, -6.1971, -6.2201, -6.2596, -6.2841, -6.3007, -6.3092, -6.3524, -6.3793, -6.3884, -6.4069, -6.4163, -5.1387, -3.5018, -5.0221, -3.2347, -4.8106, -5.28, -3.7537, -4.9237, -4.7911, -5.384, -4.0309, -5.3738, -4.8614, -4.1049, -4.9709, -5.0125, -4.7336, -5.1374, -4.6784, -4.5454, -5.0178, -4.8017, -4.9237, -5.0075, -4.9907, -5.0393, -2.8011, -3.3875, -3.8149, -4.2576, -4.6734, -4.9794, -5.0838, -5.0888, -5.1891, -5.3353, -5.4878, -5.5105, -5.582, -5.7326, -5.9215, -5.9508, -6.0378, -6.1402, -6.1918, -6.207, -6.2304, -6.347, -6.347, -4.9445, -6.5309, -6.6552, -6.6673, -6.6919, -6.8111, -6.869, -5.2391, -4.2465, -5.6709, -4.6305, -3.1721, -3.9911, -3.4267, -5.5237, -5.6529, -4.2822, -5.8012, -4.7259, -4.4284, -5.1634, -4.9668, -5.498, -5.0852, -4.7677, -5.3697, -5.2966, -4.9731, -4.6455, -4.9235, -5.1546, -5.3339, -5.2976, -5.1989, -5.0776, -5.1137, -5.1301, -2.7873, -3.045, -3.397, -3.7354, -4.0393, -4.1898, -4.2207, -4.2837, -4.7693, -4.7724, -4.9119, -4.9259, -4.9402, -5.4063, -5.775, -5.7833, -5.9072, -6.0487, -6.0931, -6.2135, -6.3803, -6.6775, -2.5003, -3.3781, -2.3794, -7.0345, -7.1441, -4.4787, -5.7306, -7.2719, -3.6526, -3.3719, -6.322, -4.8043, -4.6377, -5.4767, -5.239, -5.0724, -5.4102, -5.3561, -5.2304, -5.5595, -5.6149, -5.5836]}, \"token.table\": {\"Topic\": [4, 7, 8, 7, 3, 5, 1, 2, 3, 5, 7, 5, 5, 5, 6, 1, 3, 5, 7, 1, 2, 3, 4, 5, 6, 8, 1, 2, 3, 7, 8, 2, 3, 4, 5, 6, 7, 8, 4, 1, 2, 3, 4, 5, 6, 7, 3, 6, 8, 1, 3, 5, 4, 1, 2, 3, 4, 7, 8, 1, 2, 3, 5, 7, 8, 2, 3, 8, 1, 2, 4, 8, 1, 2, 3, 4, 5, 6, 7, 8, 6, 1, 2, 3, 4, 5, 6, 7, 8, 7, 1, 2, 3, 4, 5, 6, 7, 8, 1, 2, 3, 5, 6, 7, 8, 2, 3, 5, 6, 2, 5, 7, 7, 4, 1, 2, 5, 6, 8, 6, 1, 3, 2, 2, 3, 5, 1, 4, 4, 7, 1, 5, 8, 1, 5, 7, 2, 1, 2, 3, 7, 7, 3, 5, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 1, 2, 3, 5, 8, 3, 5, 8, 8, 8, 8, 8, 8, 2, 3, 7, 8, 1, 2, 3, 5, 6, 7, 6, 6, 1, 1, 1, 2, 3, 4, 5, 6, 3, 5, 8, 1, 2, 3, 5, 6, 8, 1, 2, 3, 5, 6, 2, 5, 2, 2, 3, 5, 3, 5, 8, 3, 5, 1, 3, 4, 5, 6, 6, 2, 3, 6, 3, 6, 2, 1, 3, 4, 5, 6, 8, 1, 2, 8, 6, 6, 3, 5, 6, 1, 3, 6, 1, 3, 4, 5, 8, 7, 7, 2, 3, 5, 2, 3, 6, 1, 2, 3, 4, 5, 7, 8, 1, 3, 5, 2, 7, 7, 7, 4, 2, 1, 2, 5, 6, 1, 4, 6, 7, 7, 3, 8, 1, 2, 7, 1, 2, 3, 4, 5, 6, 7, 1, 2, 3, 4, 7, 8, 4, 1, 2, 5, 1, 5, 7, 4, 4, 1, 2, 5, 7, 8, 2, 7, 3, 2, 3, 7, 1, 2, 3, 4, 5, 6, 7, 1, 3, 4, 5, 7, 8, 1, 3, 4, 3, 5, 4, 6, 2, 3, 1, 2, 3, 5, 6, 7, 6, 1, 2, 3, 4, 5, 6, 7, 5, 7, 5, 7, 5, 1, 2, 3, 5, 7, 8, 1, 2, 3, 4, 5, 6, 7, 8, 7, 3, 2, 5, 8, 7, 1, 2, 3, 5, 6, 7, 8, 1, 2, 3, 4, 5, 6, 7, 8, 4, 1, 2, 5, 1, 2, 3, 4, 5, 8, 1, 2, 3, 4, 5, 6, 7, 8, 4, 8, 4, 1, 2, 5, 1, 2, 3, 5, 6, 8, 2, 1, 2, 3, 8, 1, 2, 3, 5, 8, 1, 6, 1, 4, 2, 1, 2, 5, 8, 1, 2, 8, 3, 6, 6, 6, 1, 2, 3, 4, 5, 6, 7, 8, 1, 2, 3, 5, 6, 7, 8, 1, 2, 3, 4, 6, 7, 8, 2, 3, 5, 2, 6, 1, 2, 3, 5, 6, 7, 1, 2, 3, 4, 5, 6, 7, 5, 7, 8, 2, 3, 5, 7, 7, 4, 1, 2, 3, 8, 1, 2, 3, 4, 5, 6, 7, 8, 1, 3, 5, 6, 7, 1, 1, 3, 4, 5, 2, 7, 2, 7, 4, 1, 7, 2, 3, 7, 8, 2, 3, 4, 5, 7, 1, 2, 3, 5, 6, 8, 2, 8, 4, 2, 7, 2, 6, 2, 4, 1, 2, 3, 4, 5, 1, 3, 5, 1, 3, 5, 6, 7, 2, 3, 5, 1, 2, 3, 4, 5, 6, 7, 6, 3, 5, 2, 3, 5, 1, 3, 5, 1, 2, 3, 4, 5, 6, 8, 7, 2, 7, 8, 2, 7, 7, 1, 3, 1, 2, 3, 5, 6, 8, 2, 5, 6, 8, 7, 4, 8, 2, 2, 2, 1, 3, 4, 5, 8, 5, 8, 1, 2, 3, 5, 6, 7, 1, 1, 3, 1, 5, 7, 1, 2, 3, 4, 5, 6, 7, 8, 1, 2, 3, 5, 6, 7, 1, 5, 1, 2, 3, 4, 5, 6, 7, 8, 1, 2, 3, 4, 5, 6, 7, 8, 1, 2, 3, 4, 5, 6, 7, 8, 1, 2, 3, 4, 5, 6, 8, 1, 2, 3, 4, 5, 6, 7, 1, 2, 3, 4, 5, 6, 7, 8, 1, 2, 3, 4, 5, 6, 7, 1, 2, 3, 5, 6, 8, 6, 1, 2, 3, 4, 5, 7, 1, 3, 5, 2, 3, 5, 8, 1, 2, 3, 4, 5, 6, 7, 8, 4, 4, 1, 4, 1, 3, 4, 5, 1, 2, 6, 1, 2, 3, 5, 7, 8, 4, 4, 2, 3, 5, 1, 2, 3, 4, 5, 6, 1, 2, 3, 5, 6, 7, 6, 4, 4, 1, 2, 3, 5, 8, 1, 2, 3, 5, 6, 6, 1, 4, 5, 7, 1, 2, 3, 5, 6, 7, 5, 6, 7, 8, 1, 3, 4, 5, 2, 3, 5, 7, 8, 5, 5, 4, 7, 1, 1, 1, 2, 3, 5, 6, 7, 8, 3, 5, 1, 2, 3, 4, 5, 6, 8, 2, 3, 4, 5, 8, 3, 4, 8, 2, 5, 7, 3, 5, 1, 2, 3, 4, 5, 7, 8, 5, 8, 2, 1, 2, 3, 5, 8, 1, 3, 4, 6, 7, 6, 1, 2, 2, 2, 7, 4, 2, 2, 3, 5, 6, 1, 2, 3, 5, 6, 7, 1, 2, 3, 4, 5, 6, 7, 8, 2, 6, 8, 8, 8, 8, 1, 2, 3, 4, 5, 6, 7, 8, 1, 2, 3, 5, 7, 1, 2, 3, 4, 5, 6, 7, 8, 4, 1, 2, 3, 5, 6, 7, 3, 5, 7, 5, 1, 2, 3, 4, 5, 6, 7, 8, 1, 2, 3, 5, 1, 3, 4, 1, 2, 3, 4, 5, 2, 7, 2, 1, 2, 3, 5, 6, 7, 8, 1, 2, 3, 4, 5, 6, 7, 2, 3, 2, 5, 2, 3, 5, 6, 1, 2, 3, 8, 1, 3, 5, 1, 1, 2, 5, 7, 2, 5, 7, 8, 7, 7, 1, 2, 3, 4, 5, 6, 7, 8, 2, 7, 8, 6, 2, 7, 2, 7, 1, 2, 3, 4, 5, 7, 8, 6, 1, 3, 4, 5, 8, 2, 6, 7, 8, 1, 1, 1, 2, 3, 5, 8, 3, 5, 8, 1, 2, 3, 5, 7, 8, 1, 2, 8, 1, 6, 7, 6, 1, 2, 3, 4, 5, 6, 7, 8, 7, 1, 4, 8, 2, 3, 4, 2, 3, 2, 1, 2, 3, 5, 6, 8, 2, 3, 5, 8, 1, 2, 3, 5, 8, 2, 3, 1, 2, 6, 7, 8, 4, 7, 3, 5, 1, 3, 5, 6, 2, 3, 4, 7, 2, 3, 4, 5, 2, 3, 5, 6, 1, 2, 1, 2, 3, 4, 5, 6, 7, 8, 1, 2, 3, 4, 5, 6, 7, 8, 4, 2, 1, 2, 3, 4, 5, 6, 7, 7, 1, 2, 3, 5, 8, 1, 3, 5, 2, 5, 7, 8, 2, 4, 1, 2, 3, 4, 5, 6, 7, 8, 1, 2, 3, 6, 8, 2, 3, 4, 5, 3, 5, 1, 2, 3, 4, 5, 6, 7, 8, 7, 1, 2, 3, 4, 5, 7, 4, 3, 2, 6, 7, 2, 7, 2, 3, 8, 2, 2, 3, 5, 7, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 7, 8, 4, 3, 4, 5, 7, 2, 7, 8, 1, 2, 3, 4, 5, 6, 7, 8, 7, 1, 2, 3, 5, 6, 7, 8, 3, 5, 1, 2, 3, 5, 1, 3, 7, 8, 4, 1, 3, 4, 5, 1, 3, 4, 5, 7, 2, 1, 1, 1, 4, 5, 1, 2, 3, 5, 2, 3, 7, 8, 1, 2, 4, 5, 8, 2, 5, 6, 7, 8, 6, 1, 2, 3, 5, 8, 1, 2, 3, 5, 8, 2, 7, 3, 5, 3, 5, 2, 2, 4, 8, 1, 2, 5, 6, 1, 2, 3, 4, 5, 6, 7, 8, 1, 2, 3, 4, 5, 7, 8, 1, 2, 3, 5, 6, 7, 8, 4, 7], \"Freq\": [0.03376160537114536, 0.9495451510634632, 0.00844040134278634, 0.9958835735351642, 0.06774008072092597, 0.92988656262362, 0.10459347885789669, 0.07769801286586611, 0.1404541001806041, 0.6574447242496363, 0.017930310661353716, 0.9978732288860261, 0.9910480244189543, 0.9910240553394514, 0.989561516363699, 0.04077448774560824, 0.025535537780077885, 0.07331170523957845, 0.8599710142710101, 0.7374132579305326, 0.03866586371171332, 0.06904618519948807, 0.013118775187902734, 0.07595080371943688, 0.04418955852767237, 0.021404317411841303, 0.03651071906499145, 0.07073951818842093, 0.03194687918186752, 0.8397465384948033, 0.022819199415619655, 0.10184386137345108, 0.0021668906675202356, 0.6146746526865735, 0.11556750226774591, 0.1394032996104685, 0.012279047115948002, 0.014445937783468239, 0.998961789010149, 0.22655156394648793, 0.006071198469652004, 0.2847072545505229, 0.4655650615938405, 0.003834441138727581, 0.007668882277455162, 0.00543212494653074, 0.9098147757275566, 0.017496437994760704, 0.05248931398428212, 0.9753453163581912, 0.006516783850945599, 0.015205828985539729, 0.9986576891743603, 0.12774412306611163, 0.010058592367410364, 0.022128903208302804, 0.7855760638947494, 0.04979003221868131, 0.005029296183705182, 0.12555697926665027, 0.5260118382492706, 0.08462651025461451, 0.042036697904252955, 0.040377354565927184, 0.18142153832361804, 0.9901969448423291, 0.9435821214268953, 0.05353657426535576, 0.017045187955960002, 0.02597361974241524, 0.9537188499168097, 0.003246702467801905, 0.41864986079600797, 0.13734775333423851, 0.24131478687078867, 0.010779190768003528, 0.12587313090378313, 0.009040611611873927, 0.015299496573940492, 0.04172589974711043, 0.9902348490201881, 0.3824428375754651, 0.17643081180967587, 0.04965418056919022, 0.06682184938300599, 0.2509120826634612, 0.004754123671518212, 0.06867067525526308, 0.0005282359635020236, 0.9968530003196652, 0.13924984707414692, 0.2979751972355871, 0.019475503087293275, 0.04187233163768054, 0.0535576334900565, 0.022396828550387264, 0.2736308183764705, 0.1499613737721582, 0.5833601406774064, 0.09126087330663682, 0.177039644683193, 0.10544984300802206, 0.01805868871085393, 0.021283454552077846, 0.0029022892571015243, 0.012301137094681301, 0.837423563753304, 0.14761364513617561, 0.0028387239449264544, 0.2106413312489592, 0.08496457058781547, 0.7044978977906367, 0.9907136079890203, 0.9994199232699384, 0.06716625439394816, 0.06447960421819023, 0.040299752636368896, 0.7952484520243462, 0.03223980210909511, 0.9948520107736574, 0.9616161919203698, 0.038366751275810844, 0.9922074093260411, 0.9535180767201924, 0.028198234470367137, 0.017754443925786716, 0.998055998211494, 0.9997467955398537, 0.9600810662761965, 0.039858729147665854, 0.040062755428127955, 0.9314590637039749, 0.010015688857031989, 0.04723139687538146, 0.9446279375076292, 0.9898881565606543, 0.996868877520092, 0.1663597241181507, 0.029112951720676374, 0.7943676826641696, 0.010397482757384419, 0.9972161293806843, 0.962241751638403, 0.035828150326961815, 0.9988323858181678, 0.9983699558048829, 0.99883807187236, 0.9959626564178272, 0.9956663977024246, 0.9962493115166915, 0.9975660850713529, 0.9915811587694217, 0.9914474005445825, 0.9860050526475544, 0.9879087304188204, 0.9937163111333049, 0.9901738803870762, 0.9757303337816725, 0.002150828733690829, 0.05678187856943789, 0.08452756923404958, 0.05957795592323597, 0.7968820458324521, 0.03390464865047023, 0.014778949411743434, 0.9513586454657588, 0.9996170899488197, 0.9994243567597566, 0.999162276016132, 0.9993593474376207, 0.9993988149323233, 0.7074991692171962, 0.0918202473710842, 0.12758181739982227, 0.07248966897717174, 0.2220204359876185, 0.2275159913338467, 0.01428844390019327, 0.11760488440928307, 0.4165630952440961, 0.0021982221384912724, 0.9881966629456647, 0.9948520104879972, 0.9991428195532764, 0.999572376548797, 0.9036089699824583, 0.013052129566413287, 0.047690473415740856, 0.011044109633118936, 0.00953809468314817, 0.015060149499707638, 0.809958621491752, 0.18451720075403819, 0.0035944909237799645, 0.7621240899101066, 0.0071536372908228875, 0.032104128329546615, 0.11916912852761054, 0.0729322045747309, 0.006630200415884627, 0.07203442038629766, 0.041162525935027235, 0.007717973612817607, 0.030871894451270428, 0.8464044395389976, 0.9819242491579105, 0.007671283196546176, 0.9921685912334883, 0.011573120428543733, 0.9576757154619939, 0.0303794411249273, 0.9206233151385926, 0.07493445588337382, 0.0042819689076213615, 0.599517242467852, 0.4004438910434625, 0.9623706722565097, 0.0194486059201664, 0.004694491084178096, 0.004694491084178096, 0.009388982168356192, 0.9912068538623633, 0.04160943094759985, 0.9362121963209966, 0.9975974884399489, 0.9993481781782216, 0.9886755740541258, 0.9924925818583623, 0.05890209290027328, 0.041652194265193246, 0.05721917596026547, 0.016408440165076126, 0.8178976328437947, 0.007993855465037088, 0.08027201936000879, 0.8873312410336107, 0.03037319651459792, 0.9979415548072085, 0.9982696045860209, 0.0183566210675674, 0.9545442955135047, 0.9921694868937433, 0.16749298171467808, 0.21675562339546575, 0.6145514549678261, 0.024416537957541864, 0.040287287629944074, 0.9192826541014512, 0.0036624806936312797, 0.010987442080893839, 0.9958835718188218, 0.9998794940139356, 0.9531112546366548, 0.03306054267352187, 0.012901675189667072, 0.9375564441542485, 0.03625908900044055, 0.02071947942882317, 0.03344590338595314, 0.6301208197913571, 0.08160800426172565, 0.026087804641043447, 0.0735809874490969, 0.012709443286662193, 0.14181063035644131, 0.010220383309749627, 0.5660519986938255, 0.42296663235733073, 0.9926486775922576, 0.9982881770836691, 0.9928230498947339, 0.9931656906839862, 0.9976355383025229, 0.9856873914396074, 0.14750929237380123, 0.11063196928035092, 0.7006691387755558, 0.03687732309345031, 0.07516150330758878, 0.0843918633629067, 0.7067818556643436, 0.1318622865045417, 0.9992505024439695, 0.9627759180905942, 0.017505016692556257, 0.05283786593001204, 0.7358639377082165, 0.21135146372004815, 0.6034181509788256, 0.09210465495468254, 0.19823347547859657, 0.011749976558004768, 0.010991913554262525, 0.0545805362694415, 0.02880639414220524, 0.01848834802415875, 0.041085217831463884, 0.1807749584584411, 0.057519304964049445, 0.6614720070865686, 0.03903095693989069, 0.9998659056602538, 0.06220942327826374, 0.22810121868696703, 0.6946718932739451, 0.0937035423516439, 0.8433318811647951, 0.03748141694065756, 0.9998371309255939, 0.9953865304907805, 0.050132292141579535, 0.2615597850865019, 0.008718659502883397, 0.6756961114734633, 0.002179664875720849, 0.10217964170987835, 0.8957748589899336, 0.9783708780431817, 0.018845967877221755, 0.9046064581066443, 0.05653790363166527, 0.09321707247355925, 0.11877659234534162, 0.6592852625750923, 0.04059453156106612, 0.021049016364997248, 0.008269256429106062, 0.05863654558820662, 0.04241647584873876, 0.475859838428038, 0.01060411896218469, 0.29912452239162646, 0.17099141826522812, 0.0008836765801820575, 0.017520733964661212, 0.9461196340917054, 0.017520733964661212, 0.9043688758080076, 0.09542501327470584, 0.9996682098322346, 0.9875164091710015, 0.09335843761567117, 0.9055768448720103, 0.32666659614829446, 0.15239885134114992, 0.05398883074417419, 0.42370981090364557, 0.03417014604061658, 0.00888423797056031, 0.9880312607506021, 0.25089095428518177, 0.060811015044448996, 0.25526060806083084, 0.05498481001025029, 0.07501238981530835, 0.15694339810872765, 0.14601926366960508, 0.00649550179975966, 0.9916466080966415, 0.954409949587055, 0.03718480323066448, 0.9994504494476596, 0.08574453270741297, 0.13887458058133142, 0.5949513281722949, 0.08995285333108968, 0.017885362650626017, 0.07259353075842324, 0.009161960868528763, 0.14830924155930933, 0.021473345785614287, 0.690869111742497, 0.041515135185520954, 0.011166139808519429, 0.03464366453412438, 0.042946691571228574, 0.995151013428486, 0.9983261690964458, 0.034174784021737334, 0.9295541253912556, 0.02733982721738987, 0.9990985122523657, 0.3290402486156419, 0.34174550644240825, 0.022299023940855166, 0.0622298342535493, 0.13742421730992138, 0.09853057090145306, 0.008556602209863028, 0.04837683930949723, 0.5619155950564679, 0.12032188238515978, 0.06760353185557946, 0.05643964586108011, 0.04093424864649766, 0.10109518983907756, 0.0031010794429164896, 0.9994200731667245, 0.9778891634122966, 0.011312342507340192, 0.01005541556208017, 0.060902257014482195, 0.120112784667451, 0.022838346380430825, 0.7832706943807016, 0.006766917446053578, 0.005075188084540183, 0.056617832305915924, 0.2694227882143585, 0.07711739227874755, 0.014642542837736876, 0.12006885126944239, 0.41877672515927467, 0.007809356180126334, 0.0351421028105685, 0.9973168563959867, 0.0011105978356302747, 0.9944857390276234, 0.9917401190632771, 0.004942226506959853, 0.0016474088356532844, 0.042434827148182364, 0.5034834896770827, 0.2207757898925704, 0.20758658686002726, 0.02007052635387004, 0.006307879711216297, 0.9929494735282861, 0.030150851243485755, 0.944726672295887, 0.024567360272469873, 0.001116698194203176, 0.011232223809427165, 0.27864555219540466, 0.39571988497828015, 0.2821016210598438, 0.032400645604116825, 0.9985305051318186, 0.9876912047030827, 0.9905138485847573, 0.9939610225569261, 0.9918673360188709, 0.3082471790956023, 0.6238335767410998, 0.030274276518318082, 0.03761349506821337, 0.11160541817861679, 0.22321083635723357, 0.6510316060419312, 0.9912481755409088, 0.9996899477772351, 0.996035922996716, 0.9977992152766569, 0.08667282855841227, 0.17240013535073276, 0.33313883558633367, 0.15695661317123383, 0.03592982711148726, 0.18910435321835403, 0.022377348464171894, 0.0037820870643670806, 0.36475604866644296, 0.11211763790976731, 0.04634195700270382, 0.07175528826225108, 0.3191615425831376, 0.0612909753906728, 0.025413331259547256, 0.09995811845176625, 0.12105480676821943, 0.6725825156126383, 0.0095437399526812, 0.03717035560517941, 0.015571365185953536, 0.044202585043997135, 0.07926111949338102, 0.8877245383258675, 0.03170444779735241, 0.18951892409187748, 0.8089222369775257, 0.06683516805396231, 0.2944678874848104, 0.013760181658168709, 0.022802586747822432, 0.5940466995855119, 0.007862960947524977, 0.14406596237285735, 0.43171281653819205, 0.06063382254749888, 0.07179044589623867, 0.03977578759115927, 0.09313355143295829, 0.15910315036463707, 0.9510466427017228, 0.017382053521963602, 0.02731465553451423, 0.11031572408224424, 0.04262198430450346, 0.844918159448098, 0.9980837598852095, 0.9903531590593673, 0.9986861948800471, 0.03722225905849427, 0.18611129529247136, 0.18611129529247136, 0.5955561449359084, 0.11789819441585435, 0.12982264474828986, 0.6102625952389982, 0.004423586413645432, 0.030965104895518028, 0.057891283065533705, 0.04635149242124127, 0.0025002879729300273, 0.07351828621103858, 0.8642881452126974, 0.05020761009534342, 0.007172515727906202, 0.005379386795929652, 0.9996999223743, 0.04575889579454496, 0.010459176181610276, 0.937403665276821, 0.005229588090805138, 0.9847622269143403, 0.010152187906333406, 0.03054279969855267, 0.9689439904368433, 0.9978321692715517, 0.9876399561330284, 0.005143958104859523, 0.12026476483461766, 0.011453787127106445, 0.8475802474058769, 0.020998609733028483, 0.6474042361165634, 0.04617067828104018, 0.031456945642027376, 0.26332507722922915, 0.011669512093010155, 0.0033042564121675815, 0.346946923277596, 0.0615417756766212, 0.06071571157357931, 0.47333473104300605, 0.05410719874924414, 0.9848872403870288, 0.00994835596350534, 0.9996413067380856, 0.9793312620247194, 0.018408482368885704, 0.9917794824443805, 0.9977235445643226, 0.9877413273341177, 0.008664397608194015, 0.019426304270443673, 0.038852608540887346, 0.893609996440409, 0.0038852608540887344, 0.04273786939497608, 0.9291249052809584, 0.002514546428365246, 0.06789275356586164, 0.16813179907415998, 0.7877000494850553, 0.01748159378752734, 0.0071983033242759625, 0.019024087357015045, 0.013986557468663043, 0.890477492171547, 0.09324371645775362, 0.14439158930285434, 0.10808741827813668, 0.1897718030837514, 0.11798855583033241, 0.4084219240280737, 0.008250947960163105, 0.023102654288456696, 0.9898374728573052, 0.32439150227634145, 0.6739947793409996, 0.027309355655498413, 0.11834054117382646, 0.8465900253204508, 0.01151676180315209, 0.77162304081119, 0.21625919385918926, 0.07659500890806431, 0.07846317985704149, 0.6164964131624688, 0.014011282117328838, 0.049506530147895224, 0.1559922742395944, 0.008406769270397303, 0.9917460770873138, 0.9984216135285123, 0.9812250791462175, 0.017264957404332273, 0.04411035059606802, 0.9508231128485772, 0.9971783830476662, 0.99514904126631, 0.9861573111439805, 0.4935083009619448, 0.10333831989158164, 0.11221702995386551, 0.2735135960853557, 0.009371971732410746, 0.007892186722030101, 0.2014752083040383, 0.002518440103800479, 0.7530135910363432, 0.04029504166080766, 0.9972277669495379, 0.995331045002542, 0.0028601466810417872, 0.9997528717583299, 0.9916500750750791, 0.9994423059486557, 0.0894676525358271, 0.010015035731622437, 0.8913381801143968, 0.004006014292648974, 0.004006014292648974, 0.9508781979964233, 0.044341470372890206, 0.24666654170970084, 0.061422893785419576, 0.5947296064937452, 0.012918307026298563, 0.08287215828191531, 0.0014624498520337994, 0.9934832766455376, 0.9769984824216177, 0.022925475985882435, 0.2862244333171321, 0.13357140221466163, 0.5788094095968671, 0.07453118349349704, 0.7563221234056006, 0.06267394975589524, 0.041500318081606306, 0.00931639793668713, 0.03980642754766319, 0.005081671601829344, 0.011010288470630245, 0.7313559438494572, 0.11720447818100276, 0.04219361214516099, 0.037505433017920885, 0.019422456384280456, 0.05223971027496123, 0.9927866922077655, 0.0038480104349138196, 0.24582304085796924, 0.4952347392467118, 0.037232321881772716, 0.004485821913466592, 0.06280150678853229, 0.023774856141372938, 0.07222173280681214, 0.05786710268371904, 0.1599152768868277, 0.1650890064331662, 0.2803220736016156, 0.0009406780993342804, 0.09594916613209661, 0.2666822411612685, 0.01740254483768419, 0.013639832440347067, 0.11684741160663197, 0.2989212892813496, 0.24249839532061293, 0.017607144214697966, 0.07082873922730773, 0.1436582902971948, 0.008803572107348983, 0.10044075449748159, 0.08687231731862945, 0.25933941787767323, 0.3181059854755696, 0.039603556424669305, 0.028105749720733056, 0.05493396536325097, 0.2133481910619282, 0.07833649370504348, 0.5125308399116564, 0.13613354088986213, 0.032480985194774124, 0.12419200221531282, 0.0033436308288738066, 0.11272812508774548, 0.2838170561475026, 0.3002937617783061, 0.24111686127330761, 0.004873391806293994, 0.03968333327982252, 0.01137124754801932, 0.029008284561273773, 0.0898096490017036, 0.21585892710016724, 0.2373651670695566, 0.2232931828920549, 0.05124326313694007, 0.13992331436874308, 0.058943028064252306, 0.07354603051260311, 0.0547282055627663, 0.13318095343845593, 0.034508425182433924, 0.03531721639764722, 0.6847765622139231, 0.057693773351881715, 0.9949747399870454, 0.011449948116800307, 0.03912065606573438, 0.031487357321200844, 0.8148546409789551, 0.01717492217520046, 0.0858746108760023, 0.9945217224202165, 0.0036033395739862918, 0.0018016697869931459, 0.0649781119443006, 0.8733058245314, 0.04678424059989643, 0.015594746866632144, 0.13097397246673984, 0.17334790473539094, 0.046226107929437585, 0.017334790473539096, 0.019260878303932326, 0.012519570897556013, 0.4285545422624943, 0.17238486082019433, 0.9958762157155311, 0.9996591446741989, 0.9958965386911921, 0.004122372702412662, 0.9762612159279541, 0.009787079858926857, 0.007340309894195143, 0.0024467699647317143, 0.9942777756462403, 0.003993083436330282, 0.9998074093872975, 0.9922060539887649, 0.7836334056208633, 0.1222898573066266, 0.06456904465789884, 0.008804869726077115, 0.019566377169060255, 0.9962538114389036, 0.9976758529121221, 0.9504290576731057, 0.03448581486477161, 0.014713947675635889, 0.05301076601210182, 0.8778582851604061, 0.012722583842904436, 0.008481722561936291, 0.010602153202420363, 0.036047320888229234, 0.6625405039728568, 0.06608196195469533, 0.13113139325384854, 0.11770849473180106, 0.004474299507349163, 0.01789719802939665, 0.9998935260781779, 0.9967561146989149, 0.999718574354875, 0.9262800960717954, 0.049598210643497, 0.017202982970942655, 0.005808799444733883, 0.000893661453035982, 0.8100641733710053, 0.030963196709428924, 0.03235793529994374, 0.004184215771544449, 0.12245804824720087, 0.999733752270898, 0.028022402396905333, 0.44478111038492296, 0.5103654564202333, 0.016097975845030723, 0.1468486942942962, 0.014365633137485498, 0.08619379882491299, 0.6320878580493619, 0.02075035897636794, 0.09896325050267787, 0.03296411713156774, 0.875609361307268, 0.07828977818747337, 0.010301286603614918, 0.08098590539331955, 0.564741713609415, 0.011877932791020201, 0.342300426795764, 0.7980807017745185, 0.045847189250876594, 0.100184598733397, 0.005094132138986288, 0.052639365436191646, 0.9963689570554785, 0.9955432591915127, 0.9829837671345639, 0.01679243049413967, 0.9995759558624318, 0.9998071087706996, 0.29611127273034993, 0.2418026550243469, 0.09482457059778308, 0.0452571814216692, 0.012930623263334058, 0.26593981844923714, 0.04267105676900239, 0.786704174015996, 0.21283356549157967, 0.04355657994072642, 0.2076715507888206, 0.2123383272110413, 0.03344523102591493, 0.08711315988145284, 0.012833635161106891, 0.40328726248690444, 0.034399663927009244, 0.17199831963504622, 0.19110924403894025, 0.025481232538525366, 0.5784239786245258, 0.44202174134743283, 0.08450415643406804, 0.468023020250223, 0.05518634040729485, 0.8277951061094228, 0.1241692659164134, 0.05127961447137344, 0.9401262653085131, 0.0035439370921682358, 0.1565238882374304, 0.01594771691475706, 0.22267738062457082, 0.5587607481985252, 0.028942152919373923, 0.013585092186644903, 0.9482820883981234, 0.04214587059547215, 0.9929494792725703, 0.05997099731960201, 0.12708139908201377, 0.7325028958322817, 0.008567285331371716, 0.07139404442809763, 0.4869997926731204, 0.0038701970278658573, 0.1564204632095784, 0.343157469804106, 0.009675492569664643, 0.9955927437143101, 0.9948584667906761, 0.9990897485101834, 0.9996198800889159, 0.9896225393797627, 0.9971497538109309, 0.9985109478871164, 0.9849095810515646, 0.015074044738045422, 0.5041430517946303, 0.47901964389788787, 0.9894200595073233, 0.5283439766977236, 0.22067495404334117, 0.12092175536934727, 0.10526282301935985, 0.013919050977766592, 0.011019248690731885, 0.024343452416157468, 0.44952397927563503, 0.030982575802382232, 0.007745643950595558, 0.02544997298052826, 0.07828632992923368, 0.003872821975297779, 0.37981318372027506, 0.07962819851914905, 0.014598503061843994, 0.9057707581553206, 0.9990945069634752, 0.9959537616716776, 0.9864878668184072, 0.2681932106155571, 0.09107888900329207, 0.3895328379847638, 0.099089084453093, 0.012756977938571853, 0.045984455359968306, 0.08306869355349114, 0.01008691278863821, 0.9414006601769587, 0.015055325703676667, 0.004428036971669608, 0.01682654049234451, 0.02214018485834804, 0.026191425838035777, 0.35675896709688126, 0.22381763897957846, 0.009127315064770044, 0.3214402261940755, 0.009920994635619613, 0.01666727098784095, 0.03611242047365539, 0.9969438357650755, 0.05478957080677317, 0.021261624492180633, 0.7474278763789654, 0.07114466656998904, 0.08259323360424015, 0.02289713406850222, 0.1277988705304378, 0.07632432545567815, 0.7951929721893909, 0.9911000596065254, 0.20717249809531016, 0.06561056738302422, 0.030665808668152623, 0.4746068760152458, 0.15974746841084156, 0.001782895852799571, 0.058122404801266014, 0.0021394750233594854, 0.9878217221191121, 0.02251823639824384, 0.9495189681259487, 0.02251823639824384, 0.04307061155760349, 0.7791865181784632, 0.1775031264192144, 0.885794209820315, 0.012621420088113297, 0.029832447480995065, 0.05507528765722166, 0.01721102739288177, 0.03735095363319413, 0.959919508373089, 0.9882797148363945, 0.8641763553355777, 0.0072261040735015995, 0.06175034390083185, 0.0307109423123818, 0.018065260183753998, 0.01379528959486669, 0.004269970588887309, 0.2998281943057215, 0.05504204651468318, 0.28775145836157157, 0.12076735944149897, 0.18231226377226287, 0.0387849019744814, 0.015792654696196017, 0.01957861443193917, 0.9593521071650193, 0.9523077153229023, 0.045658589090824084, 0.8219830886443892, 0.0429299923408744, 0.07702145684686289, 0.056819107509980815, 0.9994059882920624, 0.3954750900395049, 0.07003204719449566, 0.5355391844284962, 0.0779911585268157, 0.90377989586957, 0.01720393202797405, 0.9932778564164995, 0.33460078948339256, 0.08258232251079477, 0.0526818264293001, 0.5282420974397389, 0.589202509089423, 0.02095658719059507, 0.37641254684645764, 0.012896361348058506, 0.992573309804679, 0.992718194522602, 0.6349414221298191, 0.17897677670773426, 0.07244298104836863, 0.011870908659186455, 0.04352666508368367, 0.004261351826374625, 0.012175290932498929, 0.04170037144380883, 0.07753858380228223, 0.903096446638346, 0.01976473704764057, 0.9995663246135656, 0.9920002375478296, 0.9964881700541247, 0.035389416180349825, 0.9647462584817106, 0.04269940418502381, 0.44283414340274696, 0.05302990519752957, 0.05096380499502842, 0.07644570749254263, 0.23484672301763096, 0.10055020985505607, 0.9985092982267624, 0.01653573364071232, 0.004409528970856618, 0.9193867904236049, 0.05291434765027942, 0.007716675698999083, 0.3614745429993068, 0.019024975947331938, 0.07609990378932775, 0.5326993265252943, 0.9994115459499451, 0.992700755323528, 0.24025313441170962, 0.2882422368038053, 0.3713002986362785, 0.09874680684527375, 0.0012304898049255294, 0.2898176795601886, 0.6698616178513793, 0.03827780673436453, 0.07733422558346974, 0.03583781185575427, 0.7922042620745681, 0.0037724012479741337, 0.01886200623987067, 0.07167562371150854, 0.008548340589082718, 0.9659624865663471, 0.025645021767248154, 0.9518157745466993, 0.021841183232351794, 0.025289791111144183, 0.9933625875588964, 0.13600536700563146, 0.14618263936659706, 0.03885867628732327, 0.04533512233521048, 0.03423264339597527, 0.23500247088047885, 0.3284483352857086, 0.03608305655251447, 0.9996816973780256, 0.9997797209261753, 0.9842917897353474, 0.014498569734235887, 0.01992371349869256, 0.033717053613172024, 0.9456100945148699, 0.9792469280591822, 0.008589885333852475, 0.9924926257674224, 0.15711610013443442, 0.08165088668403679, 0.17938452377553538, 0.016082750407461793, 0.5529991870873401, 0.011134211820550471, 0.9235965129900668, 0.02716460332323726, 0.03056017873864192, 0.019241594020626394, 0.030541191491628702, 0.6574393326355863, 0.010716207540922352, 0.2673693781460127, 0.033756053753905405, 0.9853060832690796, 0.9775394912540488, 0.986960260998328, 0.9444744414600519, 0.008189084752543803, 0.02729694917514601, 0.020472711881359507, 0.9977938164221436, 0.9963816683638718, 0.49269630577826373, 0.5074036581895552, 0.17492598564938422, 0.6465911020703311, 0.17781255306934104, 0.9913202139425332, 0.035032896992468227, 0.06714638590223077, 0.06568668186087792, 0.8320313035711203, 0.07703486752189337, 0.031150711456940487, 0.026520200294422307, 0.8650636762704419, 0.06513054442552603, 0.032565272212763015, 0.8304144414254568, 0.08141318053190753, 0.9930638542861802, 0.006645742810710575, 0.19265456478746365, 0.5524573153623605, 0.027405790202160323, 0.013838567329803727, 0.06485132532986453, 0.03771687958515134, 0.07082090339370142, 0.04043032415962265, 0.17854803648796222, 0.2669496701888426, 0.18727188191897015, 0.027334715683824836, 0.24194131328661983, 0.07851460887907134, 0.01512133208041374, 0.004652717563204228, 0.997293563998096, 0.9893801888967785, 0.014559345567304953, 0.06967686807210227, 0.03431845740864738, 0.8309226505911897, 0.03327850415383989, 0.008319626038459973, 0.008319626038459973, 0.9922726805725098, 0.21372231649583626, 0.047583459144355994, 0.6379409522573829, 0.05000295706695037, 0.05000295706695037, 0.27667037005566625, 0.01844469133704442, 0.6824535794706434, 0.19975446995648297, 0.07220041082764445, 0.719597427915523, 0.009626721443685926, 0.005471491033753683, 0.9930756226262936, 0.07920926858700901, 0.23119456567274205, 0.41635125696369457, 0.001407271269312343, 0.111375469028434, 0.1377115456398507, 0.010454015143463119, 0.012263363918293274, 0.23998490968313352, 0.0693289739084608, 0.5820586380886159, 0.10132696186621193, 0.006856711705232386, 0.8601553912474206, 0.05949201553296281, 0.02478833980540117, 0.05205551359134245, 0.6675892542657632, 0.33129241853368463, 0.1792568720423376, 0.5229372985246314, 0.06762987178034896, 0.0037711824412833997, 0.030923696018523874, 0.11791230433079429, 0.04274006766787853, 0.034694878459807274, 0.9979607649941068, 0.904065366193927, 0.042361092448814364, 0.03113721325297466, 0.005068848669088898, 0.010137697338177796, 0.0068791517651920755, 0.9968426617033594, 0.9787620724877302, 0.020643066041179305, 0.9753848704457222, 0.997033785121369, 0.01770445936797416, 0.9781713800805724, 0.8640001305536679, 0.030857147519773854, 0.10285715839924618, 0.9895760673309053, 0.04341312707263985, 0.9164993493112857, 0.02411840392924436, 0.014471042357546616, 0.038089962745356414, 0.07255230999115507, 0.8397929881476199, 0.007255230999115507, 0.01088284649867326, 0.030834731746240907, 0.03239532174828318, 0.0819411079515398, 0.09051634017902653, 0.018103268035805305, 0.02191448235913274, 0.7308003464980353, 0.023820089520796457, 0.9936497440772646, 0.3172255120794109, 0.10866288259073743, 0.008763135692801407, 0.5643459386164106, 0.9618426846846256, 0.0070723726815046005, 0.028289490726018402, 0.37771452147704976, 0.33188555891877175, 0.0320503202204951, 0.10573610315733431, 0.039538712795377136, 0.02905496319054229, 0.0673955331739383, 0.016474463664740473, 0.9941110000432246, 0.3756589821076672, 0.026950644397182564, 0.5511131772648353, 0.024750591793330928, 0.00990023671733237, 0.008250197264443643, 0.003300078905777457, 0.9632078568220281, 0.027919068313681975, 0.9860230466974005, 0.013217757511632509, 0.019826636267448763, 0.958287419593357, 0.020583145314601838, 0.9468246844716846, 0.010291572657300919, 0.010291572657300919, 0.9987971488664816, 0.1158769252974319, 0.12553333573888456, 0.048282052207263296, 0.6952615517845915, 0.045381892508659046, 0.8471286601616356, 0.02117821650404089, 0.07563648751443175, 0.00907637850173181, 0.9853061077750402, 0.9931132944496903, 0.9962526862440246, 0.14334345807558674, 0.856385275169531, 0.9910362863696397, 0.9975610650438028, 0.9777765535090956, 0.016095087300561244, 0.006035657737710467, 0.9260426037963001, 0.006710453650697827, 0.0033552268253489136, 0.06374930968162935, 0.1301251127668151, 0.6329386553026638, 0.019581934445491594, 0.09727799692276469, 0.1193866325870294, 0.1814485685798047, 0.015778136398243887, 0.7399945970776384, 0.036289713715960945, 0.02524501823719022, 0.9908563580148328, 0.02970351579527468, 0.06480767082605385, 0.2538300440687109, 0.6129725532297593, 0.03510415503077917, 0.6237071039801406, 0.2188841862737831, 0.07897883009878771, 0.053254296866611144, 0.024821918031047566, 0.06749237778533913, 0.9324331577113008, 0.8637220127564657, 0.13361445501204994, 0.8008819005697256, 0.19786494014075573, 0.9888233049130432, 0.006943776590370608, 0.9870082439169651, 0.005455824463862621, 0.9367867849991366, 0.01983234787669405, 0.011666086986290618, 0.030331826164355605, 0.17958127951506456, 0.20271160532022495, 0.3167569617206685, 0.09766137562178824, 0.10344395707307834, 0.05204323306161084, 0.03630176133309892, 0.011565162902580186, 0.39414573058054536, 0.0870891477592549, 0.37110152309785593, 0.03381812266940139, 0.1056442239141477, 0.006883334702621521, 0.0011971016874124385, 0.20178499486344575, 0.46691515267136563, 0.127712022065472, 0.10268046574063948, 0.008173569412190208, 0.019412227353951744, 0.07305127662144999, 0.9993009978173393, 0.9970555704366213], \"Term\": [\"abandon\", \"abandon\", \"abandon\", \"acid\", \"active\", \"active\", \"activity\", \"activity\", \"activity\", \"activity\", \"activity\", \"addict\", \"addicted\", \"addiction\", \"admiration\", \"affect\", \"affect\", \"affect\", \"affect\", \"age\", \"age\", \"age\", \"age\", \"age\", \"age\", \"age\", \"ahead\", \"ahead\", \"ahead\", \"ahead\", \"ahead\", \"air\", \"air\", \"air\", \"air\", \"air\", \"air\", \"air\", \"airship\", \"allow\", \"allow\", \"allow\", \"allow\", \"allow\", \"allow\", \"allow\", \"application\", \"application\", \"application\", \"appropriate\", \"appropriate\", \"appropriate\", \"architect\", \"area\", \"area\", \"area\", \"area\", \"area\", \"area\", \"ask\", \"ask\", \"ask\", \"ask\", \"ask\", \"ask\", \"asleep\", \"assignment\", \"assignment\", \"attempt\", \"attempt\", \"attempt\", \"attempt\", \"away\", \"away\", \"away\", \"away\", \"away\", \"away\", \"away\", \"away\", \"backdrop\", \"bad\", \"bad\", \"bad\", \"bad\", \"bad\", \"bad\", \"bad\", \"bad\", \"battery\", \"begin\", \"begin\", \"begin\", \"begin\", \"begin\", \"begin\", \"begin\", \"begin\", \"believe\", \"believe\", \"believe\", \"believe\", \"believe\", \"believe\", \"believe\", \"benefit\", \"benefit\", \"benefit\", \"benefit\", \"bike\", \"bike\", \"bike\", \"biker\", \"blimp\", \"blood\", \"blood\", \"blood\", \"blood\", \"blood\", \"blueprints\", \"book\", \"book\", \"boyfriend\", \"brother\", \"brother\", \"brother\", \"bud\", \"builder\", \"building\", \"building\", \"bully\", \"bully\", \"bully\", \"bullying\", \"bullying\", \"bumpy\", \"bus\", \"business\", \"business\", \"business\", \"business\", \"california\", \"cam\", \"cam\", \"caps10\", \"caps11\", \"caps12\", \"caps13\", \"caps14\", \"caps15\", \"caps16\", \"caps17\", \"caps18\", \"caps19\", \"caps20\", \"caps21\", \"caps22\", \"caps24\", \"caps3\", \"caps3\", \"caps3\", \"caps3\", \"caps3\", \"caps4\", \"caps4\", \"caps4\", \"caps5\", \"caps6\", \"caps7\", \"caps8\", \"caps9\", \"car\", \"car\", \"car\", \"car\", \"care\", \"care\", \"care\", \"care\", \"care\", \"care\", \"caring\", \"celebration\", \"censor\", \"censorship\", \"certain\", \"certain\", \"certain\", \"certain\", \"certain\", \"certain\", \"chat\", \"chat\", \"chat\", \"child\", \"child\", \"child\", \"child\", \"child\", \"child\", \"childhood\", \"childhood\", \"childhood\", \"childhood\", \"childhood\", \"clock\", \"clock\", \"coaster\", \"communicate\", \"communicate\", \"communicate\", \"communication\", \"communication\", \"communication\", \"computer\", \"computer\", \"conclude\", \"conclude\", \"conclude\", \"conclude\", \"conclude\", \"congregation\", \"convenient\", \"convenient\", \"cooking\", \"coordination\", \"courageous\", \"cream\", \"create\", \"create\", \"create\", \"create\", \"create\", \"create\", \"cry\", \"cry\", \"cry\", \"cuba\", \"cuban\", \"cuber\", \"cuber\", \"culinary\", \"culture\", \"culture\", \"culture\", \"current\", \"current\", \"current\", \"current\", \"current\", \"cycle\", \"cyclist\", \"dad\", \"dad\", \"dad\", \"dance\", \"dance\", \"dance\", \"date1\", \"date1\", \"date1\", \"date1\", \"date1\", \"date1\", \"date1\", \"dear\", \"dear\", \"dear\", \"deer\", \"dehydrate\", \"dehydrated\", \"dehydration\", \"densely\", \"dentist\", \"depression\", \"depression\", \"depression\", \"depression\", \"describe\", \"describe\", \"describe\", \"describe\", \"desert\", \"dictionary\", \"dictionary\", \"didn\", \"didn\", \"didn\", \"different\", \"different\", \"different\", \"different\", \"different\", \"different\", \"different\", \"direction\", \"direction\", \"direction\", \"direction\", \"direction\", \"direction\", \"dirigible\", \"distract\", \"distract\", \"distract\", \"distraction\", \"distraction\", \"distraction\", \"dock\", \"docking\", \"drink\", \"drink\", \"drink\", \"drink\", \"drink\", \"dry\", \"dry\", \"earthquake\", \"easier\", \"easier\", \"easier\", \"easy\", \"easy\", \"easy\", \"easy\", \"easy\", \"easy\", \"easy\", \"effect\", \"effect\", \"effect\", \"effect\", \"effect\", \"effect\", \"efficient\", \"efficient\", \"efficient\", \"email\", \"email\", \"empire\", \"encompass\", \"engine\", \"engine\", \"enjoy\", \"enjoy\", \"enjoy\", \"enjoy\", \"enjoy\", \"enjoy\", \"eternally\", \"example\", \"example\", \"example\", \"example\", \"example\", \"example\", \"example\", \"exceed\", \"exceed\", \"excessive\", \"excessive\", \"exercise\", \"eye\", \"eye\", \"eye\", \"eye\", \"eye\", \"eye\", \"face\", \"face\", \"face\", \"face\", \"face\", \"face\", \"face\", \"face\", \"factory\", \"faraway\", \"fat\", \"fat\", \"fat\", \"feature\", \"feel\", \"feel\", \"feel\", \"feel\", \"feel\", \"feel\", \"feel\", \"finally\", \"finally\", \"finally\", \"finally\", \"finally\", \"finally\", \"finally\", \"finally\", \"flammable\", \"flower\", \"flower\", \"flower\", \"fly\", \"fly\", \"fly\", \"fly\", \"fly\", \"fly\", \"forget\", \"forget\", \"forget\", \"forget\", \"forget\", \"forget\", \"forget\", \"forget\", \"frame\", \"frame\", \"framework\", \"freedom\", \"freedom\", \"freedom\", \"fun\", \"fun\", \"fun\", \"fun\", \"fun\", \"fun\", \"funniest\", \"funny\", \"funny\", \"funny\", \"funny\", \"game\", \"game\", \"game\", \"game\", \"game\", \"geese\", \"generosity\", \"genre\", \"german\", \"giggle\", \"girl\", \"girl\", \"girl\", \"girl\", \"glance\", \"glance\", \"glance\", \"google\", \"grateful\", \"gratefulness\", \"gratitude\", \"great\", \"great\", \"great\", \"great\", \"great\", \"great\", \"great\", \"great\", \"grow\", \"grow\", \"grow\", \"grow\", \"grow\", \"grow\", \"grow\", \"hand\", \"hand\", \"hand\", \"hand\", \"hand\", \"hand\", \"hand\", \"handy\", \"handy\", \"handy\", \"happiness\", \"happiness\", \"happy\", \"happy\", \"happy\", \"happy\", \"happy\", \"happy\", \"hard\", \"hard\", \"hard\", \"hard\", \"hard\", \"hard\", \"hard\", \"health\", \"health\", \"health\", \"healthy\", \"healthy\", \"healthy\", \"heat\", \"heatstroke\", \"helium\", \"hello\", \"hello\", \"hello\", \"hello\", \"help\", \"help\", \"help\", \"help\", \"help\", \"help\", \"help\", \"help\", \"helpful\", \"helpful\", \"helpful\", \"helpful\", \"helpful\", \"hibiscus\", \"highly\", \"highly\", \"highly\", \"highly\", \"hilarious\", \"hilarious\", \"hill\", \"hill\", \"hindenburg\", \"ho\", \"ho\", \"hot\", \"hot\", \"hot\", \"hot\", \"hour\", \"hour\", \"hour\", \"hour\", \"hour\", \"house\", \"house\", \"house\", \"house\", \"house\", \"house\", \"hug\", \"hug\", \"hydrogen\", \"ibm\", \"ibm\", \"ice\", \"immigrant\", \"impatient\", \"impatient\", \"improve\", \"improve\", \"improve\", \"improve\", \"improve\", \"inappropriate\", \"inappropriate\", \"inappropriate\", \"information\", \"information\", \"information\", \"information\", \"information\", \"instant\", \"instant\", \"instant\", \"instead\", \"instead\", \"instead\", \"instead\", \"instead\", \"instead\", \"instead\", \"instill\", \"interact\", \"interact\", \"interaction\", \"interaction\", \"interaction\", \"internet\", \"internet\", \"internet\", \"job\", \"job\", \"job\", \"job\", \"job\", \"job\", \"job\", \"joe\", \"joke\", \"journey\", \"journey\", \"juice\", \"juice\", \"june\", \"katherine\", \"keyboard\", \"kid\", \"kid\", \"kid\", \"kid\", \"kid\", \"kid\", \"kitchen\", \"kitchen\", \"kitchen\", \"kitchen\", \"kurmaskie\", \"landing\", \"landing\", \"laugh\", \"laughing\", \"laughter\", \"law\", \"law\", \"law\", \"law\", \"law\", \"lazy\", \"lazy\", \"learn\", \"learn\", \"learn\", \"learn\", \"learn\", \"learn\", \"librarian\", \"library\", \"library\", \"limit\", \"limit\", \"limit\", \"line\", \"line\", \"line\", \"line\", \"line\", \"line\", \"line\", \"line\", \"listen\", \"listen\", \"listen\", \"listen\", \"listen\", \"listen\", \"literature\", \"literature\", \"little\", \"little\", \"little\", \"little\", \"little\", \"little\", \"little\", \"little\", \"live\", \"live\", \"live\", \"live\", \"live\", \"live\", \"live\", \"live\", \"location1\", \"location1\", \"location1\", \"location1\", \"location1\", \"location1\", \"location1\", \"location1\", \"location2\", \"location2\", \"location2\", \"location2\", \"location2\", \"location2\", \"location2\", \"long\", \"long\", \"long\", \"long\", \"long\", \"long\", \"long\", \"look\", \"look\", \"look\", \"look\", \"look\", \"look\", \"look\", \"look\", \"lot\", \"lot\", \"lot\", \"lot\", \"lot\", \"lot\", \"lot\", \"love\", \"love\", \"love\", \"love\", \"love\", \"love\", \"loving\", \"low\", \"low\", \"low\", \"low\", \"low\", \"low\", \"magazine\", \"magazine\", \"magazine\", \"mail\", \"mail\", \"mail\", \"mail\", \"man\", \"man\", \"man\", \"man\", \"man\", \"man\", \"man\", \"man\", \"marcia\", \"mast\", \"material\", \"material\", \"medium\", \"medium\", \"medium\", \"medium\", \"melt\", \"melt\", \"memoir\", \"mining\", \"minute\", \"minute\", \"minute\", \"minute\", \"minute\", \"modification\", \"modify\", \"mom\", \"mom\", \"mom\", \"moment\", \"moment\", \"moment\", \"moment\", \"moment\", \"moment\", \"month1\", \"month1\", \"month1\", \"month1\", \"month1\", \"month1\", \"mood\", \"moor\", \"mooring\", \"movie\", \"movie\", \"movie\", \"movie\", \"movie\", \"music\", \"music\", \"music\", \"music\", \"music\", \"narcs\", \"nature\", \"nature\", \"nature\", \"nature\", \"negative\", \"negative\", \"negative\", \"negative\", \"negative\", \"negative\", \"neighborhood\", \"neighborhood\", \"neighborhood\", \"neighborhood\", \"newspaper\", \"newspaper\", \"newspaper\", \"newspaper\", \"night\", \"night\", \"night\", \"night\", \"night\", \"obese\", \"obesity\", \"obstacle\", \"obstacle\", \"offend\", \"offensive\", \"old\", \"old\", \"old\", \"old\", \"old\", \"old\", \"old\", \"online\", \"online\", \"organization1\", \"organization1\", \"organization1\", \"organization1\", \"organization1\", \"organization1\", \"organization1\", \"organization2\", \"organization2\", \"organization2\", \"organization2\", \"organization2\", \"organization3\", \"organization3\", \"organization3\", \"outdoor\", \"outdoor\", \"outdoor\", \"outdoors\", \"outdoors\", \"outside\", \"outside\", \"outside\", \"outside\", \"outside\", \"outside\", \"outside\", \"overweight\", \"overweight\", \"pant\", \"paper\", \"paper\", \"paper\", \"paper\", \"paper\", \"paragraph\", \"paragraph\", \"paragraph\", \"paragraph\", \"paragraph\", \"passionate\", \"paterson\", \"patience\", \"patient\", \"patiently\", \"pedal\", \"pedestrian\", \"pee\", \"percent1\", \"percent1\", \"percent1\", \"perseverance\", \"person\", \"person\", \"person\", \"person\", \"person\", \"person\", \"person1\", \"person1\", \"person1\", \"person1\", \"person1\", \"person1\", \"person1\", \"person1\", \"person2\", \"person2\", \"person2\", \"person3\", \"person4\", \"person5\", \"place\", \"place\", \"place\", \"place\", \"place\", \"place\", \"place\", \"place\", \"plant\", \"plant\", \"plant\", \"plant\", \"plant\", \"play\", \"play\", \"play\", \"play\", \"play\", \"play\", \"play\", \"play\", \"populated\", \"positive\", \"positive\", \"positive\", \"positive\", \"positive\", \"positive\", \"post\", \"post\", \"post\", \"predator\", \"problem\", \"problem\", \"problem\", \"problem\", \"problem\", \"problem\", \"problem\", \"problem\", \"profanity\", \"program\", \"program\", \"program\", \"project\", \"project\", \"project\", \"public\", \"public\", \"public\", \"public\", \"public\", \"pump\", \"pump\", \"puppy\", \"read\", \"read\", \"read\", \"read\", \"read\", \"read\", \"read\", \"reason\", \"reason\", \"reason\", \"reason\", \"reason\", \"reason\", \"reason\", \"reconnect\", \"reconnect\", \"red\", \"red\", \"relationship\", \"relationship\", \"relationship\", \"relationship\", \"remove\", \"reply\", \"reply\", \"reply\", \"research\", \"research\", \"research\", \"retake\", \"rid\", \"rid\", \"rid\", \"rid\", \"ride\", \"ride\", \"ride\", \"ride\", \"rider\", \"ridiculously\", \"right\", \"right\", \"right\", \"right\", \"right\", \"right\", \"right\", \"right\", \"road\", \"road\", \"road\", \"rodriguez\", \"roller\", \"rolling\", \"rough\", \"rough\", \"run\", \"run\", \"run\", \"run\", \"run\", \"run\", \"run\", \"sacrifice\", \"safety\", \"safety\", \"safety\", \"safety\", \"safety\", \"said\", \"said\", \"said\", \"said\", \"sang\", \"sangs\", \"school\", \"school\", \"school\", \"school\", \"school\", \"screen\", \"screen\", \"screen\", \"search\", \"search\", \"search\", \"search\", \"search\", \"search\", \"seat\", \"seat\", \"seat\", \"section\", \"section\", \"section\", \"selflessly\", \"set\", \"set\", \"set\", \"set\", \"set\", \"set\", \"set\", \"set\", \"setting\", \"shelf\", \"shift\", \"shift\", \"ship\", \"ship\", \"ship\", \"shot\", \"shot\", \"shower\", \"simple\", \"simple\", \"simple\", \"simple\", \"simple\", \"simple\", \"sister\", \"sister\", \"sister\", \"sister\", \"sit\", \"sit\", \"sit\", \"sit\", \"sit\", \"skate\", \"skype\", \"slavery\", \"smile\", \"smile\", \"smile\", \"smile\", \"smith\", \"snake\", \"social\", \"social\", \"society\", \"society\", \"society\", \"solidarity\", \"speed\", \"speed\", \"speed\", \"speed\", \"spend\", \"spend\", \"spend\", \"spend\", \"spending\", \"spending\", \"spending\", \"spending\", \"spring\", \"spring\", \"start\", \"start\", \"start\", \"start\", \"start\", \"start\", \"start\", \"start\", \"stay\", \"stay\", \"stay\", \"stay\", \"stay\", \"stay\", \"stay\", \"stay\", \"steel\", \"stomach\", \"stress\", \"stress\", \"stress\", \"stress\", \"stress\", \"stress\", \"stress\", \"stroke\", \"student\", \"student\", \"student\", \"student\", \"student\", \"suicide\", \"suicide\", \"suicide\", \"sun\", \"sun\", \"sun\", \"sun\", \"swivel\", \"swivel\", \"talk\", \"talk\", \"talk\", \"talk\", \"talk\", \"talk\", \"talk\", \"talk\", \"teach\", \"teach\", \"teach\", \"teach\", \"teach\", \"team\", \"team\", \"team\", \"team\", \"technology\", \"technology\", \"tell\", \"tell\", \"tell\", \"tell\", \"tell\", \"tell\", \"tell\", \"tell\", \"terrain\", \"test\", \"test\", \"test\", \"test\", \"test\", \"test\", \"tether\", \"textbook\", \"thankful\", \"thankful\", \"thirst\", \"thirsty\", \"thirsty\", \"time1\", \"time1\", \"time1\", \"tolerant\", \"tool\", \"tool\", \"tool\", \"tool\", \"touch\", \"touch\", \"touch\", \"touch\", \"touch\", \"touch\", \"town\", \"town\", \"town\", \"town\", \"town\", \"town\", \"town\", \"transmit\", \"travel\", \"travel\", \"travel\", \"travel\", \"truck\", \"truck\", \"truck\", \"try\", \"try\", \"try\", \"try\", \"try\", \"try\", \"try\", \"try\", \"tumbleweed\", \"type\", \"type\", \"type\", \"type\", \"type\", \"type\", \"type\", \"typing\", \"typing\", \"uncensored\", \"unhealthy\", \"unhealthy\", \"unhealthy\", \"update\", \"update\", \"update\", \"update\", \"urban\", \"usage\", \"usage\", \"usage\", \"usage\", \"useful\", \"useful\", \"useful\", \"useful\", \"useful\", \"van\", \"vietnam\", \"violence\", \"violent\", \"violent\", \"virus\", \"vow\", \"wait\", \"wait\", \"wait\", \"wake\", \"wake\", \"wake\", \"wake\", \"walk\", \"walk\", \"walk\", \"walk\", \"walk\", \"warm\", \"warm\", \"warm\", \"warm\", \"warm\", \"warmth\", \"waste\", \"waste\", \"waste\", \"waste\", \"waste\", \"watch\", \"watch\", \"watch\", \"watch\", \"watch\", \"water\", \"water\", \"web\", \"web\", \"website\", \"website\", \"wet\", \"wind\", \"wind\", \"wind\", \"winter\", \"winter\", \"winter\", \"winter\", \"work\", \"work\", \"work\", \"work\", \"work\", \"work\", \"work\", \"work\", \"world\", \"world\", \"world\", \"world\", \"world\", \"world\", \"world\", \"year\", \"year\", \"year\", \"year\", \"year\", \"year\", \"year\", \"york\", \"yosemite\"]}, \"R\": 30, \"lambda.step\": 0.01, \"plot.opts\": {\"xlab\": \"PC1\", \"ylab\": \"PC2\"}, \"topic.order\": [1, 4, 7, 3, 5, 6, 8, 2]};\n",
       "\n",
       "function LDAvis_load_lib(url, callback){\n",
       "  var s = document.createElement('script');\n",
       "  s.src = url;\n",
       "  s.async = true;\n",
       "  s.onreadystatechange = s.onload = callback;\n",
       "  s.onerror = function(){console.warn(\"failed to load library \" + url);};\n",
       "  document.getElementsByTagName(\"head\")[0].appendChild(s);\n",
       "}\n",
       "\n",
       "if(typeof(LDAvis) !== \"undefined\"){\n",
       "   // already loaded: just create the visualization\n",
       "   !function(LDAvis){\n",
       "       new LDAvis(\"#\" + \"ldavis_el1146022842250645362787206531\", ldavis_el1146022842250645362787206531_data);\n",
       "   }(LDAvis);\n",
       "}else if(typeof define === \"function\" && define.amd){\n",
       "   // require.js is available: use it to load d3/LDAvis\n",
       "   require.config({paths: {d3: \"https://cdnjs.cloudflare.com/ajax/libs/d3/3.5.5/d3.min\"}});\n",
       "   require([\"d3\"], function(d3){\n",
       "      window.d3 = d3;\n",
       "      LDAvis_load_lib(\"https://cdn.rawgit.com/bmabey/pyLDAvis/files/ldavis.v1.0.0.js\", function(){\n",
       "        new LDAvis(\"#\" + \"ldavis_el1146022842250645362787206531\", ldavis_el1146022842250645362787206531_data);\n",
       "      });\n",
       "    });\n",
       "}else{\n",
       "    // require.js not available: dynamically load d3 & LDAvis\n",
       "    LDAvis_load_lib(\"https://cdnjs.cloudflare.com/ajax/libs/d3/3.5.5/d3.min.js\", function(){\n",
       "         LDAvis_load_lib(\"https://cdn.rawgit.com/bmabey/pyLDAvis/files/ldavis.v1.0.0.js\", function(){\n",
       "                 new LDAvis(\"#\" + \"ldavis_el1146022842250645362787206531\", ldavis_el1146022842250645362787206531_data);\n",
       "            })\n",
       "         });\n",
       "}\n",
       "</script>"
      ],
      "text/plain": [
       "PreparedData(topic_coordinates=              x         y  topics  cluster       Freq\n",
       "topic                                                \n",
       "0      0.071756 -0.123076       1        1  25.791481\n",
       "3      0.080689  0.073752       2        1  19.300420\n",
       "6      0.107037 -0.126833       3        1  14.366069\n",
       "2     -0.375785 -0.015805       4        1  12.629813\n",
       "4      0.074849 -0.098745       5        1   8.237609\n",
       "5      0.056780 -0.088600       6        1   7.519359\n",
       "7     -0.103776  0.055201       7        1   7.491409\n",
       "1      0.088450  0.324106       8        1   4.663839, topic_info=     Category          Freq           Term         Total  loglift  logprob\n",
       "term                                                                      \n",
       "360   Default  20459.000000       computer  20459.000000  30.0000  30.0000\n",
       "202   Default  13892.000000           book  13892.000000  29.0000  29.0000\n",
       "229   Default   7777.000000       building   7777.000000  28.0000  28.0000\n",
       "520   Default   6250.000000      dirigible   6250.000000  27.0000  27.0000\n",
       "268   Default   4649.000000          caps3   4649.000000  26.0000  26.0000\n",
       "1160  Default   4307.000000           mood   4307.000000  25.0000  25.0000\n",
       "269   Default   3450.000000          caps4   3450.000000  24.0000  24.0000\n",
       "446   Default   3904.000000        cyclist   3904.000000  23.0000  23.0000\n",
       "1039  Default   7502.000000        library   7502.000000  22.0000  22.0000\n",
       "1228  Default   3989.000000       obstacle   3989.000000  21.0000  21.0000\n",
       "1017  Default   4648.000000          laugh   4648.000000  20.0000  20.0000\n",
       "270   Default   2464.000000          caps5   2464.000000  19.0000  19.0000\n",
       "601   Default   3443.000000         empire   3443.000000  18.0000  18.0000\n",
       "1942  Default   2889.000000          water   2889.000000  17.0000  17.0000\n",
       "1236  Default   5073.000000      offensive   5073.000000  16.0000  16.0000\n",
       "1128  Default   2581.000000         memoir   2581.000000  15.0000  15.0000\n",
       "1078  Default   3709.000000           love   3709.000000  14.0000  14.0000\n",
       "1111  Default   2978.000000           mast   2978.000000  13.0000  13.0000\n",
       "1308  Default   3614.000000        person1   3614.000000  12.0000  12.0000\n",
       "271   Default   1905.000000          caps6   1905.000000  11.0000  11.0000\n",
       "1584  Default   2172.000000        setting   2172.000000  10.0000  10.0000\n",
       "1287  Default   3266.000000        patient   3266.000000   9.0000   9.0000\n",
       "228   Default   2571.000000        builder   2571.000000   8.0000   8.0000\n",
       "47    Default   2427.000000         affect   2427.000000   7.0000   7.0000\n",
       "1178  Default   1952.000000          narcs   1952.000000   6.0000   6.0000\n",
       "1689  Default   2375.000000          spend   2375.000000   5.0000   5.0000\n",
       "541   Default   2388.000000           dock   2388.000000   4.0000   4.0000\n",
       "421   Default   2376.000000         create   2376.000000   3.0000   3.0000\n",
       "1162  Default   2322.000000        mooring   2322.000000   2.0000   2.0000\n",
       "1595  Default   3293.000000          shelf   3293.000000   1.0000   1.0000\n",
       "...       ...           ...            ...           ...      ...      ...\n",
       "262    Topic8    124.196788         caps18    125.069671   3.0583  -5.7750\n",
       "261    Topic8    123.171418         caps17    124.044309   3.0583  -5.7833\n",
       "265    Topic8    108.816368         caps21    109.689253   3.0573  -5.9072\n",
       "263    Topic8     94.461324         caps19     95.334197   3.0561  -6.0487\n",
       "1312   Topic8     90.359737        person5     91.232749   3.0557  -6.0931\n",
       "264    Topic8     80.106266         caps20     80.979141   3.0545  -6.2135\n",
       "266    Topic8     67.801938         caps22     68.674807   3.0525  -6.3803\n",
       "267    Topic8     50.370793         caps24     51.243667   3.0482  -6.6775\n",
       "269    Topic8   3283.128967          caps4   3450.854224   3.0155  -2.5003\n",
       "1309   Topic8   1364.864900        person2   1507.003828   2.9663  -3.3781\n",
       "268    Topic8   3705.180065          caps3   4649.370656   2.8383  -2.3794\n",
       "787    Topic8     35.247244         glance     53.760831   2.6432  -7.0345\n",
       "860    Topic8     31.588467          hello     53.731290   2.5341  -7.1441\n",
       "1252   Topic8    454.040469  organization2    784.891389   2.5180  -4.4787\n",
       "1481   Topic8    129.840546          reply    242.746010   2.4396  -5.7306\n",
       "1539   Topic8     27.797402           said     52.562484   2.4283  -7.2719\n",
       "1251   Topic8   1037.218586  organization1   2571.368095   2.1574  -3.6526\n",
       "1308   Topic8   1373.286565        person1   3614.935076   2.0975  -3.3719\n",
       "1253   Topic8     71.871211  organization3    153.838587   2.3043  -6.3220\n",
       "112    Topic8    327.842006            ask   1807.944101   1.3579  -4.8043\n",
       "1072   Topic8    387.279820           look   4309.113824   0.6560  -4.6377\n",
       "1064   Topic8    167.369961      location2    782.757984   1.5227  -5.4767\n",
       "456    Topic8    212.283040          date1   1494.951397   1.1134  -5.2390\n",
       "1063   Topic8    250.757531      location1   2498.985609   0.7662  -5.0724\n",
       "1099   Topic8    178.872608            man   1038.374247   1.3066  -5.4102\n",
       "1931   Topic8    188.808638           walk   1583.091808   0.9389  -5.3561\n",
       "1078   Topic8    214.099151           love   3709.239101   0.2132  -5.2304\n",
       "164    Topic8    154.070380          begin   1026.931110   1.1684  -5.5595\n",
       "1530   Topic8    145.755463            run   1452.010893   0.7666  -5.6149\n",
       "671    Topic8    150.400432           face   3492.702104  -0.0798  -5.5836\n",
       "\n",
       "[514 rows x 6 columns], token_table=      Topic      Freq        Term\n",
       "term                             \n",
       "3         4  0.033762     abandon\n",
       "3         7  0.949545     abandon\n",
       "3         8  0.008440     abandon\n",
       "18        7  0.995884        acid\n",
       "21        3  0.067740      active\n",
       "21        5  0.929887      active\n",
       "22        1  0.104593    activity\n",
       "22        2  0.077698    activity\n",
       "22        3  0.140454    activity\n",
       "22        5  0.657445    activity\n",
       "22        7  0.017930    activity\n",
       "29        5  0.997873      addict\n",
       "30        5  0.991048    addicted\n",
       "31        5  0.991024   addiction\n",
       "35        6  0.989562  admiration\n",
       "47        1  0.040774      affect\n",
       "47        3  0.025536      affect\n",
       "47        5  0.073312      affect\n",
       "47        7  0.859971      affect\n",
       "51        1  0.737413         age\n",
       "51        2  0.038666         age\n",
       "51        3  0.069046         age\n",
       "51        4  0.013119         age\n",
       "51        5  0.075951         age\n",
       "51        6  0.044190         age\n",
       "51        8  0.021404         age\n",
       "54        1  0.036511       ahead\n",
       "54        2  0.070740       ahead\n",
       "54        3  0.031947       ahead\n",
       "54        7  0.839747       ahead\n",
       "...     ...       ...         ...\n",
       "1964      4  0.987008        wind\n",
       "1964      8  0.005456        wind\n",
       "1966      1  0.936787      winter\n",
       "1966      2  0.019832      winter\n",
       "1966      5  0.011666      winter\n",
       "1966      6  0.030332      winter\n",
       "1976      1  0.179581        work\n",
       "1976      2  0.202712        work\n",
       "1976      3  0.316757        work\n",
       "1976      4  0.097661        work\n",
       "1976      5  0.103444        work\n",
       "1976      6  0.052043        work\n",
       "1976      7  0.036302        work\n",
       "1976      8  0.011565        work\n",
       "1978      1  0.394146       world\n",
       "1978      2  0.087089       world\n",
       "1978      3  0.371102       world\n",
       "1978      4  0.033818       world\n",
       "1978      5  0.105644       world\n",
       "1978      7  0.006883       world\n",
       "1978      8  0.001197       world\n",
       "1991      1  0.201785        year\n",
       "1991      2  0.466915        year\n",
       "1991      3  0.127712        year\n",
       "1991      5  0.102680        year\n",
       "1991      6  0.008174        year\n",
       "1991      7  0.019412        year\n",
       "1991      8  0.073051        year\n",
       "1994      4  0.999301        york\n",
       "1995      7  0.997056    yosemite\n",
       "\n",
       "[1325 rows x 3 columns], R=30, lambda_step=0.01, plot_opts={'xlab': 'PC1', 'ylab': 'PC2'}, topic_order=[1, 4, 7, 3, 5, 6, 8, 2])"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "prepare(lda_base, doc_term_matrix, vectorizer)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### \"Score Allocation\"\n",
    "\n",
    "Can we take this to the next level and assign target scores based on word probabilities? To keep it simple, we'll limit the essays to topic number 4, \"hibiscus\". This topic has only four target scores and a reasonably balanced distribution."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "hibiscus = training_set[training_set.topic == 'hibiscus']\n",
    "\n",
    "# Split essays into training and test sets\n",
    "train_essays, test_essays = train_test_split(hibiscus, \n",
    "                                         stratify=hibiscus.target_score, \n",
    "                                         test_size=0.2, \n",
    "                                         random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "vectorizer = CountVectorizer(max_df=.2, \n",
    "                             min_df=3, \n",
    "                             stop_words=STOP_WORDS, \n",
    "                             max_features=400) # limit to account for smaller set of essays\n",
    "\n",
    "# Train and test doc-term matrices\n",
    "train_dtm = vectorizer.fit_transform(train_essays.l_essay)\n",
    "test_dtm = vectorizer.fit_transform(test_essays.l_essay)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "iteration: 1 of max_iter: 500\n",
      "iteration: 2 of max_iter: 500\n",
      "iteration: 3 of max_iter: 500\n",
      "iteration: 4 of max_iter: 500\n",
      "iteration: 5 of max_iter: 500, perplexity: 289.0970\n",
      "iteration: 6 of max_iter: 500\n",
      "iteration: 7 of max_iter: 500\n",
      "iteration: 8 of max_iter: 500\n",
      "iteration: 9 of max_iter: 500\n",
      "iteration: 10 of max_iter: 500, perplexity: 279.5138\n",
      "iteration: 11 of max_iter: 500\n",
      "iteration: 12 of max_iter: 500\n",
      "iteration: 13 of max_iter: 500\n",
      "iteration: 14 of max_iter: 500\n",
      "iteration: 15 of max_iter: 500, perplexity: 273.1729\n",
      "iteration: 16 of max_iter: 500\n",
      "iteration: 17 of max_iter: 500\n",
      "iteration: 18 of max_iter: 500\n",
      "iteration: 19 of max_iter: 500\n",
      "iteration: 20 of max_iter: 500, perplexity: 270.1631\n",
      "iteration: 21 of max_iter: 500\n",
      "iteration: 22 of max_iter: 500\n",
      "iteration: 23 of max_iter: 500\n",
      "iteration: 24 of max_iter: 500\n",
      "iteration: 25 of max_iter: 500, perplexity: 268.8611\n",
      "iteration: 26 of max_iter: 500\n",
      "iteration: 27 of max_iter: 500\n",
      "iteration: 28 of max_iter: 500\n",
      "iteration: 29 of max_iter: 500\n",
      "iteration: 30 of max_iter: 500, perplexity: 268.0383\n",
      "iteration: 31 of max_iter: 500\n",
      "iteration: 32 of max_iter: 500\n",
      "iteration: 33 of max_iter: 500\n",
      "iteration: 34 of max_iter: 500\n",
      "iteration: 35 of max_iter: 500, perplexity: 267.3695\n",
      "iteration: 36 of max_iter: 500\n",
      "iteration: 37 of max_iter: 500\n",
      "iteration: 38 of max_iter: 500\n",
      "iteration: 39 of max_iter: 500\n",
      "iteration: 40 of max_iter: 500, perplexity: 266.9789\n",
      "iteration: 41 of max_iter: 500\n",
      "iteration: 42 of max_iter: 500\n",
      "iteration: 43 of max_iter: 500\n",
      "iteration: 44 of max_iter: 500\n",
      "iteration: 45 of max_iter: 500, perplexity: 266.7623\n",
      "iteration: 46 of max_iter: 500\n",
      "iteration: 47 of max_iter: 500\n",
      "iteration: 48 of max_iter: 500\n",
      "iteration: 49 of max_iter: 500\n",
      "iteration: 50 of max_iter: 500, perplexity: 266.5564\n",
      "iteration: 51 of max_iter: 500\n",
      "iteration: 52 of max_iter: 500\n",
      "iteration: 53 of max_iter: 500\n",
      "iteration: 54 of max_iter: 500\n",
      "iteration: 55 of max_iter: 500, perplexity: 266.4294\n",
      "iteration: 56 of max_iter: 500\n",
      "iteration: 57 of max_iter: 500\n",
      "iteration: 58 of max_iter: 500\n",
      "iteration: 59 of max_iter: 500\n",
      "iteration: 60 of max_iter: 500, perplexity: 266.3415\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "['lda_score.pkl']"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lda_score = LatentDirichletAllocation(n_components=4,\n",
    "                                     n_jobs=-1,\n",
    "                                     learning_method='batch',\n",
    "                                     evaluate_every=5,\n",
    "                                     verbose=1, \n",
    "                                     max_iter=500)\n",
    "lda_score.fit(train_dtm)\n",
    "\n",
    "# save training model\n",
    "joblib.dump(lda_score, 'lda_score.pkl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Score A</th>\n",
       "      <th>Score B</th>\n",
       "      <th>Score C</th>\n",
       "      <th>Score D</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>target_score</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.660247</td>\n",
       "      <td>0.025972</td>\n",
       "      <td>0.026131</td>\n",
       "      <td>0.287649</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.014167</td>\n",
       "      <td>0.122066</td>\n",
       "      <td>0.566254</td>\n",
       "      <td>0.297513</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.042368</td>\n",
       "      <td>0.043872</td>\n",
       "      <td>0.559395</td>\n",
       "      <td>0.354365</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.015324</td>\n",
       "      <td>0.535914</td>\n",
       "      <td>0.016749</td>\n",
       "      <td>0.432014</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.723890</td>\n",
       "      <td>0.014013</td>\n",
       "      <td>0.013799</td>\n",
       "      <td>0.248298</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               Score A   Score B   Score C   Score D\n",
       "target_score                                        \n",
       "1             0.660247  0.025972  0.026131  0.287649\n",
       "2             0.014167  0.122066  0.566254  0.297513\n",
       "0             0.042368  0.043872  0.559395  0.354365\n",
       "1             0.015324  0.535914  0.016749  0.432014\n",
       "2             0.723890  0.014013  0.013799  0.248298"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "topic_labels = ['Score A', 'Score B', 'Score C', 'Score D'] #.format(i) for i in range(1, 5)]\n",
    "\n",
    "train_preds = lda_score.transform(train_dtm)\n",
    "train_eval = pd.DataFrame(train_preds, columns=topic_labels, index=train_essays.target_score)\n",
    "train_eval.sample(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "train_eval.groupby(level='target_score')\\\n",
    "            .mean()\\\n",
    "            .plot\\\n",
    "            .bar(title='Avg. Topic Probabilities', rot=0, colormap='tab10', figsize=(8,5));"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>target_score</th>\n",
       "      <th>assignment</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>Score A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>Score D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>Score C</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>Score C</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   target_score assignment\n",
       "0             0    Score A\n",
       "1             1    Score D\n",
       "2             2    Score C\n",
       "3             3    Score C"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df = train_eval.groupby(level='target_score').agg('median')\n",
    "sns.heatmap(df, annot=True, fmt='.1%', cmap='Blues', square=True)\n",
    "plt.title('Train Data: Score Assignments');\n",
    "\n",
    "df = train_eval\\\n",
    "            .idxmax(axis=1)\\\n",
    "            .reset_index()\\\n",
    "            .groupby('target_score', as_index=False)\\\n",
    "            .agg(lambda x:x.value_counts().index[0])\\\n",
    "            .rename(columns={0:'assignment'})\n",
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Score A</th>\n",
       "      <th>Score B</th>\n",
       "      <th>Score C</th>\n",
       "      <th>Score D</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>target_score</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.011977</td>\n",
       "      <td>0.254760</td>\n",
       "      <td>0.347584</td>\n",
       "      <td>0.385679</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.286519</td>\n",
       "      <td>0.353608</td>\n",
       "      <td>0.170209</td>\n",
       "      <td>0.189664</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.607361</td>\n",
       "      <td>0.133645</td>\n",
       "      <td>0.133992</td>\n",
       "      <td>0.125002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.202250</td>\n",
       "      <td>0.137793</td>\n",
       "      <td>0.318266</td>\n",
       "      <td>0.341692</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.620999</td>\n",
       "      <td>0.032699</td>\n",
       "      <td>0.313826</td>\n",
       "      <td>0.032476</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               Score A   Score B   Score C   Score D\n",
       "target_score                                        \n",
       "2             0.011977  0.254760  0.347584  0.385679\n",
       "3             0.286519  0.353608  0.170209  0.189664\n",
       "0             0.607361  0.133645  0.133992  0.125002\n",
       "1             0.202250  0.137793  0.318266  0.341692\n",
       "1             0.620999  0.032699  0.313826  0.032476"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_preds = lda_score.transform(test_dtm)\n",
    "test_eval = pd.DataFrame(test_preds, columns=topic_labels, index=test_essays.target_score)\n",
    "test_eval.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "test_eval.groupby(level='target_score')\\\n",
    "            .mean()\\\n",
    "            .plot\\\n",
    "            .bar(title='Avg. Topic Probabilities', rot=0);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df = test_eval.groupby(level='target_score').agg('median')\n",
    "sns.heatmap(df, annot=True, fmt='.1%', cmap='Blues', square=True)\n",
    "plt.title('Topic Assignments')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes = plt.subplots(ncols=2, figsize=(8,6), sharey=True)\n",
    "source = ['Train', 'Test']\n",
    "for i, df in enumerate([train_eval, test_eval]):\n",
    "    df = df.groupby(level='target_score').agg('median')\n",
    "    sns.heatmap(df, annot=True, fmt='.1%', cmap='Blues', square=True, ax=axes[i], cbar=False)\n",
    "    axes[i].set_title('{} Data: Topic Assignments'.format(source[i]))\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "While the charts above are very similar to a confusion matrix, the ordering of the LDA derived topics (A,B,C,D) doesn't necessarily match the human-labeled topic ordering (0,1,2,3). Thus, the high percentages, shown here as deep blues, are not expected to be found along the diagonal. Instead the goal is to find topic distinction, indicated by a single dark square in each column, and model accuracy, indicated by identical color patterns between train and test data sets.\n",
    "\n",
    "As seen above, there is some agreement between train and test data that essays with highest and lowest scores are distinct and assigned \"Score A\" and \"Score B\" respectively.  Overall, both topic distinction and model accuracy are rather poor. It is highly improbable this approach could be extended to any of the other topics due to the larger range of scores and class imbalance. \n",
    "\n",
    "Furthermore, repeated LDA runs show a lack of reproducability, which is a sign of poor distinction of topics. In summary, topic modelling, or more specifically using word frequencies and probabilities is not a useful tool to grade student essays. In the next notebook we'll continue with machine learning algorithms."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:capstone2]",
   "language": "python",
   "name": "conda-env-capstone2-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}