scripts/05_nc_voter_name_lstm_hispanic.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Ethnicolr: NC Voter Registration Name Model For Hispanic/Non-Hispanic"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We build new LSTM models based on NC data. We start by assuming y = concatenation of ethnic code and race code. We remove U and also UN --- assuming they are 'missing at random.' This gives us 12 categories.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from __future__ import print_function\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import classification_report,confusion_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "NGRAMS = (2, 3)\n",
    "SAMPLE = 1000000\n",
    "EPOCHS = 15\n",
    "\n",
    "# Florida voter\n",
    "df = pd.read_csv('/opt/data/nc_voter/nc_voter_name_gender_race_ethnic_age.csv', usecols=['last_name', 'first_name', 'race_code', 'ethnic_code'], nrows=SAMPLE * 2)\n",
    "df.dropna(subset=['first_name', 'last_name'], inplace=True)"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "https://s3.amazonaws.com/dl.ncsbe.gov/data/layout_ncvoter.txt\n",
    "\n",
    "/* ****************************************************************************\n",
    "Race codes\n",
    "race               description\n",
    "*******************************************************************************\n",
    "A                  ASIAN\n",
    "B                  BLACK or AFRICAN AMERICAN\n",
    "I                  AMERICAN INDIAN or ALASKA NATIVE\n",
    "M                  TWO or MORE RACES \n",
    "O                  OTHER\n",
    "P                  NATIVE HAWAIIAN or PACIFIC ISLANDER\n",
    "U                  UNDESIGNATED\n",
    "W                  WHITE\n",
    "**************************************************************************** */\n",
    "\n",
    "\n",
    "/* ****************************************************************************\n",
    "Ethnic codes\n",
    "ethnicity          description\n",
    "*******************************************************************************\n",
    "HL                 HISPANIC or LATINO\n",
    "NL                 NOT HISPANIC or NOT LATINO\n",
    "UN                 UNDESIGNATED\n",
    "**************************************************************************** */\n"
   ]
  },
  {
   "attachments": {
    "image.png": {
     "image/png": "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"
    }
   },
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![image.png](attachment:image.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Drop U and UN from race and ethnic."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "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>last_name</th>\n",
       "      <th>first_name</th>\n",
       "      <th>race_code</th>\n",
       "      <th>ethnic_code</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>AABEL</td>\n",
       "      <td>EVELYN</td>\n",
       "      <td>W</td>\n",
       "      <td>NL</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>AARON</td>\n",
       "      <td>CLAUDIA</td>\n",
       "      <td>W</td>\n",
       "      <td>NL</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>AARON</td>\n",
       "      <td>SANDRA</td>\n",
       "      <td>W</td>\n",
       "      <td>HL</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>AARON</td>\n",
       "      <td>WILLIE</td>\n",
       "      <td>W</td>\n",
       "      <td>NL</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>AARONSON</td>\n",
       "      <td>GENA</td>\n",
       "      <td>W</td>\n",
       "      <td>NL</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  last_name first_name race_code ethnic_code\n",
       "0     AABEL     EVELYN         W          NL\n",
       "2     AARON    CLAUDIA         W          NL\n",
       "6     AARON     SANDRA         W          HL\n",
       "7     AARON     WILLIE         W          NL\n",
       "8  AARONSON       GENA         W          NL"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.drop(df[(df.race_code=='U')|(df.ethnic_code=='UN')].index, inplace=True)\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "race_code  ethnic_code\n",
       "A          HL                 149\n",
       "           NL               10142\n",
       "B          HL                1497\n",
       "           NL              266127\n",
       "I          HL                 157\n",
       "           NL                5284\n",
       "M          HL                2634\n",
       "           NL                5054\n",
       "O          HL               18580\n",
       "           NL               17627\n",
       "W          HL               10598\n",
       "           NL             1210076\n",
       "dtype: int64"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.groupby(['race_code', 'ethnic_code']).size()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.drop_duplicates(inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>last_name</th>\n",
       "      <th>first_name</th>\n",
       "      <th>race_code</th>\n",
       "      <th>ethnic_code</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>61254</th>\n",
       "      <td>MADARASZ</td>\n",
       "      <td>AGNES</td>\n",
       "      <td>W</td>\n",
       "      <td>NL</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1205421</th>\n",
       "      <td>ESTEP</td>\n",
       "      <td>DONNA</td>\n",
       "      <td>W</td>\n",
       "      <td>NL</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1348373</th>\n",
       "      <td>MINAHAN</td>\n",
       "      <td>SHELBY</td>\n",
       "      <td>W</td>\n",
       "      <td>NL</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13816</th>\n",
       "      <td>BUSH</td>\n",
       "      <td>DAVID</td>\n",
       "      <td>W</td>\n",
       "      <td>NL</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1051654</th>\n",
       "      <td>RODRIGUEZ</td>\n",
       "      <td>ROBERT</td>\n",
       "      <td>O</td>\n",
       "      <td>NL</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         last_name first_name race_code ethnic_code\n",
       "61254     MADARASZ      AGNES         W          NL\n",
       "1205421      ESTEP      DONNA         W          NL\n",
       "1348373    MINAHAN     SHELBY         W          NL\n",
       "13816         BUSH      DAVID         W          NL\n",
       "1051654  RODRIGUEZ     ROBERT         O          NL"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sdf = df.sample(SAMPLE, random_state=21)\n",
    "sdf.head()"
   ]
  },
  {
   "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>first_name</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ethnic_code</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>HL</th>\n",
       "      <td>28736</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NL</th>\n",
       "      <td>971264</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             first_name\n",
       "ethnic_code            \n",
       "HL                28736\n",
       "NL               971264"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Additional features\n",
    "sdf['first_name'] = sdf.first_name.str.title()\n",
    "sdf['last_name'] = sdf.last_name.str.title()\n",
    "\n",
    "sdf.groupby('ethnic_code').agg({'first_name': 'count'})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Preprocessing the input data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "num_words = 11666\n",
      "CPU times: user 47min 12s, sys: 540 ms, total: 47min 13s\n",
      "Wall time: 47min 14s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# concat last name and first name\n",
    "sdf['last_name_first_name'] = sdf['last_name'] + ' ' + sdf['first_name']\n",
    "\n",
    "# build n-gram list\n",
    "vect = CountVectorizer(analyzer='char', max_df=0.3, min_df=10, ngram_range=NGRAMS, lowercase=False) \n",
    "a = vect.fit_transform(sdf.last_name_first_name)\n",
    "vocab = vect.vocabulary_\n",
    "\n",
    "# sort n-gram by freq (highest -> lowest)\n",
    "words = []\n",
    "for b in vocab:\n",
    "    c = vocab[b]\n",
    "    #print(b, c, a[:, c].sum())\n",
    "    words.append((a[:, c].sum(), b))\n",
    "    #break\n",
    "words = sorted(words, reverse=True)\n",
    "words_list = [w[1] for w in words]\n",
    "num_words = len(words_list)\n",
    "print(\"num_words = %d\" % num_words)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'Ma': 2393,\n",
       " 'ad': 4084,\n",
       " 'da': 5050,\n",
       " 'ar': 4350,\n",
       " 'ra': 9252,\n",
       " 'as': 4379,\n",
       " 'sz': 9987,\n",
       " 'z ': 11438,\n",
       " ' A': 2,\n",
       " 'Ag': 593,\n",
       " 'gn': 6202,\n",
       " 'ne': 8171,\n",
       " 'es': 5717,\n",
       " 'Mad': 2398,\n",
       " 'ada': 4087,\n",
       " 'dar': 5066,\n",
       " 'ara': 4353,\n",
       " 'ras': 9273,\n",
       " 'asz': 4405,\n",
       " 'sz ': 9988,\n",
       " 'z A': 11439,\n",
       " ' Ag': 11,\n",
       " 'Agn': 600,\n",
       " 'gne': 6205,\n",
       " 'nes': 8193,\n",
       " 'Es': 1408,\n",
       " 'st': 9929,\n",
       " 'te': 10082,\n",
       " 'ep': 5669,\n",
       " 'p ': 8983,\n",
       " ' D': 51,\n",
       " 'Do': 1173,\n",
       " 'on': 8739,\n",
       " 'nn': 8304,\n",
       " 'na': 8096,\n",
       " 'Est': 1420,\n",
       " 'ste': 9935,\n",
       " 'tep': 10101,\n",
       " 'ep ': 5670,\n",
       " 'p D': 8987,\n",
       " ' Do': 60,\n",
       " 'Don': 1186,\n",
       " 'onn': 8756,\n",
       " 'nna': 8307,\n",
       " 'Mi': 2467,\n",
       " 'in': 6857,\n",
       " 'ah': 4166,\n",
       " 'ha': 6333,\n",
       " 'an': 4293,\n",
       " 'n ': 8044,\n",
       " ' S': 238,\n",
       " 'Sh': 3217,\n",
       " 'he': 6381,\n",
       " 'el': 5577,\n",
       " 'lb': 7456,\n",
       " 'by': 4718,\n",
       " 'Min': 2478,\n",
       " 'ina': 6860,\n",
       " 'nah': 8106,\n",
       " 'aha': 4169,\n",
       " 'han': 6350,\n",
       " 'an ': 4294,\n",
       " 'n S': 8063,\n",
       " ' Sh': 242,\n",
       " 'She': 3219,\n",
       " 'hel': 6393,\n",
       " 'elb': 5581,\n",
       " 'lby': 7465,\n",
       " 'Bu': 911,\n",
       " 'us': 10667,\n",
       " 'sh': 9794,\n",
       " 'h ': 6290,\n",
       " 'Da': 1083,\n",
       " 'av': 4451,\n",
       " 'vi': 10808,\n",
       " 'id': 6695,\n",
       " 'Bus': 926,\n",
       " 'ush': 10675,\n",
       " 'sh ': 9795,\n",
       " 'h D': 6294,\n",
       " ' Da': 54,\n",
       " 'Dav': 1107,\n",
       " 'avi': 4454,\n",
       " 'vid': 10812,\n",
       " 'Ro': 3096,\n",
       " 'od': 8558,\n",
       " 'dr': 5237,\n",
       " 'ri': 9387,\n",
       " 'ig': 6755,\n",
       " 'gu': 6262,\n",
       " 'ue': 10434,\n",
       " 'ez': 5853,\n",
       " ' R': 229,\n",
       " 'ob': 8527,\n",
       " 'be': 4605,\n",
       " 'er': 5689,\n",
       " 'rt': 9568,\n",
       " 'Rod': 3100,\n",
       " 'odr': 8576,\n",
       " 'dri': 5240,\n",
       " 'rig': 9396,\n",
       " 'igu': 6774,\n",
       " 'gue': 6267,\n",
       " 'uez': 10454,\n",
       " 'ez ': 5854,\n",
       " 'z R': 11456,\n",
       " ' Ro': 235,\n",
       " 'Rob': 3098,\n",
       " 'obe': 8532,\n",
       " 'ber': 4617,\n",
       " 'ert': 5711,\n",
       " 'Bi': 848,\n",
       " 'is': 6948,\n",
       " 'ss': 9910,\n",
       " 'se': 9758,\n",
       " 'ey': 5830,\n",
       " 'y ': 11150,\n",
       " ' M': 174,\n",
       " 'rg': 9365,\n",
       " 'ga': 6074,\n",
       " 're': 9325,\n",
       " 'et': 5745,\n",
       " 'Bis': 863,\n",
       " 'iss': 6965,\n",
       " 'sse': 9916,\n",
       " 'sey': 9782,\n",
       " 'ey ': 5831,\n",
       " 'y M': 11163,\n",
       " ' Ma': 176,\n",
       " 'Mar': 2411,\n",
       " 'arg': 4359,\n",
       " 'rga': 9368,\n",
       " 'gar': 6090,\n",
       " 'are': 4357,\n",
       " 'ret': 9348,\n",
       " 'Ch': 978,\n",
       " 'dw': 5295,\n",
       " 'wi': 10973,\n",
       " 'ic': 6679,\n",
       " 'ck': 4872,\n",
       " 'k ': 7152,\n",
       " ' V': 276,\n",
       " 'Vi': 3640,\n",
       " 'ir': 6922,\n",
       " 'gi': 6167,\n",
       " 'ni': 8240,\n",
       " 'ia': 6638,\n",
       " 'Cha': 979,\n",
       " 'had': 6340,\n",
       " 'adw': 4106,\n",
       " 'dwi': 5298,\n",
       " 'wic': 10976,\n",
       " 'ick': 6686,\n",
       " 'ck ': 4873,\n",
       " 'k V': 7174,\n",
       " ' Vi': 279,\n",
       " 'Vir': 3653,\n",
       " 'irg': 6931,\n",
       " 'rgi': 9372,\n",
       " 'gin': 6176,\n",
       " 'ini': 6868,\n",
       " 'nia': 8243,\n",
       " 'Fr': 1545,\n",
       " 'dl': 5189,\n",
       " 'le': 7497,\n",
       " ' P': 215,\n",
       " 'Pa': 2867,\n",
       " 'at': 4406,\n",
       " 'tr': 10228,\n",
       " 'ci': 4849,\n",
       " 'Fri': 1548,\n",
       " 'rid': 9393,\n",
       " 'idl': 6706,\n",
       " 'dle': 5192,\n",
       " 'ley': 7523,\n",
       " 'y P': 11166,\n",
       " ' Pa': 216,\n",
       " 'Pat': 2886,\n",
       " 'atr': 4420,\n",
       " 'tri': 10231,\n",
       " 'ric': 9392,\n",
       " 'ici': 6685,\n",
       " 'cia': 4852,\n",
       " 'Pr': 2975,\n",
       " 'ce': 4793,\n",
       " 'e ': 5315,\n",
       " ' E': 65,\n",
       " 'Ev': 1439,\n",
       " 've': 10787,\n",
       " 'tt': 10255,\n",
       " 'Pri': 2978,\n",
       " 'ice': 6683,\n",
       " 'ce ': 4794,\n",
       " 'e E': 5320,\n",
       " ' Ev': 85,\n",
       " 'Eve': 1441,\n",
       " 'ver': 10802,\n",
       " 'ere': 5696,\n",
       " 'ett': 5762,\n",
       " 'bi': 4627,\n",
       " 'ns': 8362,\n",
       " 'so': 9871,\n",
       " 'Ra': 3018,\n",
       " 'ac': 4061,\n",
       " 'ch': 4826,\n",
       " 'obi': 8535,\n",
       " 'bin': 4637,\n",
       " 'ins': 6876,\n",
       " 'nso': 8376,\n",
       " 'son': 9884,\n",
       " 'on ': 8740,\n",
       " 'n R': 8062,\n",
       " ' Ra': 231,\n",
       " 'Rac': 3021,\n",
       " 'ach': 4069,\n",
       " 'che': 4832,\n",
       " 'ui': 10497,\n",
       " 'ie': 6716,\n",
       " 'aq': 4346,\n",
       " 'qu': 9192,\n",
       " 'it': 6972,\n",
       " 'ta': 10043,\n",
       " 'Bui': 919,\n",
       " 'uie': 10503,\n",
       " 'ie ': 6717,\n",
       " 'e S': 5334,\n",
       " 'Sha': 3218,\n",
       " 'haq': 6353,\n",
       " 'aqu': 4348,\n",
       " 'qui': 9196,\n",
       " 'uit': 10514,\n",
       " 'ita': 6974,\n",
       " 'Be': 823,\n",
       " 'ec': 5410,\n",
       " 'Aa': 542,\n",
       " 'ro': 9490,\n",
       " 'Bec': 826,\n",
       " 'eck': 5417,\n",
       " 'k A': 7153,\n",
       " ' Aa': 5,\n",
       " 'Aar': 545,\n",
       " 'aro': 4367,\n",
       " 'ron': 9506,\n",
       " 'Ko': 2165,\n",
       " 'nz': 8453,\n",
       " 'ze': 11520,\n",
       " 'lm': 7619,\n",
       " 'ma': 7822,\n",
       " 'rc': 9292,\n",
       " 'cy': 4990,\n",
       " 'Kon': 2178,\n",
       " 'onz': 8766,\n",
       " 'nze': 8456,\n",
       " 'zel': 11531,\n",
       " 'elm': 5592,\n",
       " 'lma': 7621,\n",
       " 'man': 7837,\n",
       " 'ann': 4309,\n",
       " 'nn ': 8305,\n",
       " 'n D': 8048,\n",
       " 'Dar': 1103,\n",
       " 'arc': 4355,\n",
       " 'rcy': 9304,\n",
       " 'am': 4269,\n",
       " 'mo': 7945,\n",
       " 'os': 8837,\n",
       " 's ': 9661,\n",
       " ' N': 183,\n",
       " 'No': 2629,\n",
       " 'or': 8809,\n",
       " 'Ram': 3031,\n",
       " 'amo': 4285,\n",
       " 'mos': 7960,\n",
       " 'os ': 8838,\n",
       " 's N': 9675,\n",
       " ' No': 190,\n",
       " 'Nor': 2644,\n",
       " 'ori': 8820,\n",
       " 'ris': 9408,\n",
       " 'La': 2241,\n",
       " 'to': 10200,\n",
       " 'ou': 8882,\n",
       " 'ur': 10640,\n",
       " 'r ': 9204,\n",
       " 'Se': 3189,\n",
       " 'ea': 5368,\n",
       " 'Lat': 2264,\n",
       " 'ato': 4419,\n",
       " 'tou': 10221,\n",
       " 'our': 8900,\n",
       " 'ur ': 10641,\n",
       " 'r N': 9218,\n",
       " ' Se': 241,\n",
       " 'Sea': 3190,\n",
       " 'ean': 5382,\n",
       " 'Bo': 877,\n",
       " 'oy': 8952,\n",
       " 'yc': 11232,\n",
       " ' W': 284,\n",
       " 'Wi': 3733,\n",
       " 'il': 6812,\n",
       " 'll': 7597,\n",
       " 'li': 7552,\n",
       " 'ms': 7994,\n",
       " 'Boy': 901,\n",
       " 'oyc': 8956,\n",
       " 'yce': 11234,\n",
       " 'e W': 5338,\n",
       " ' Wi': 289,\n",
       " 'Wil': 3740,\n",
       " 'ill': 6826,\n",
       " 'lli': 7608,\n",
       " 'lia': 7555,\n",
       " 'iam': 6653,\n",
       " 'ams': 4288,\n",
       " 'Li': 2299,\n",
       " 'eb': 5393,\n",
       " 'ge': 6113,\n",
       " 'mi': 7900,\n",
       " 'io': 6883,\n",
       " 'Lin': 2311,\n",
       " 'ine': 6864,\n",
       " 'neb': 8176,\n",
       " 'ebe': 5397,\n",
       " 'erg': 5698,\n",
       " 'rge': 9369,\n",
       " 'ger': 6129,\n",
       " 'er ': 5690,\n",
       " 'r D': 9208,\n",
       " 'Dam': 1098,\n",
       " 'ami': 4280,\n",
       " 'mio': 7912,\n",
       " 'ion': 6895,\n",
       " 'Ho': 1771,\n",
       " 'ol': 8691,\n",
       " 'lg': 7538,\n",
       " ' T': 256,\n",
       " 'Tr': 3489,\n",
       " 'Hol': 1784,\n",
       " 'olg': 8700,\n",
       " 'lgu': 7545,\n",
       " 'gui': 6269,\n",
       " 'uin': 10509,\n",
       " 'in ': 6858,\n",
       " 'n T': 8064,\n",
       " ' Tr': 263,\n",
       " 'Tra': 3490,\n",
       " 'rac': 9257,\n",
       " 'acy': 4082,\n",
       " 'Ol': 2758,\n",
       " 'ld': 7479,\n",
       " 'dh': 5145,\n",
       " 'm ': 7792,\n",
       " ' B': 30,\n",
       " 'ty': 10305,\n",
       " 'Old': 2761,\n",
       " 'ldh': 7486,\n",
       " 'dha': 5146,\n",
       " 'ham': 6349,\n",
       " 'am ': 4270,\n",
       " 'm B': 7794,\n",
       " ' Be': 32,\n",
       " 'Bet': 840,\n",
       " 'tty': 10274,\n",
       " 'Gu': 1678,\n",
       " 'Pe': 2893,\n",
       " 'Gus': 1692,\n",
       " 'uss': 10683,\n",
       " 'ss ': 9911,\n",
       " 's P': 9677,\n",
       " ' Pe': 217,\n",
       " 'Pet': 2908,\n",
       " 'ete': 5750,\n",
       " 'ter': 10102,\n",
       " 'Fi': 1504,\n",
       " 'g ': 6038,\n",
       " ' K': 151,\n",
       " 'Kr': 2189,\n",
       " 'ti': 10143,\n",
       " 'Fig': 1510,\n",
       " 'ig ': 6756,\n",
       " 'g K': 6049,\n",
       " ' Kr': 159,\n",
       " 'Kri': 2192,\n",
       " 'ist': 6966,\n",
       " 'sti': 9939,\n",
       " 'tin': 10157,\n",
       " 'ow': 8919,\n",
       " 'wl': 10994,\n",
       " 'la': 7428,\n",
       " 'nd': 8147,\n",
       " 'd ': 5008,\n",
       " 'Ri': 3077,\n",
       " 'rd': 9306,\n",
       " 'Row': 3119,\n",
       " 'owl': 8931,\n",
       " 'wla': 10996,\n",
       " 'lan': 7444,\n",
       " 'and': 4299,\n",
       " 'nd ': 8148,\n",
       " 'd R': 5026,\n",
       " ' Ri': 234,\n",
       " 'Ric': 3080,\n",
       " 'ich': 6684,\n",
       " 'cha': 4829,\n",
       " 'har': 6354,\n",
       " 'ard': 4356,\n",
       " 'Ke': 2095,\n",
       " 'ei': 5529,\n",
       " 'th': 10121,\n",
       " 'Kei': 2104,\n",
       " 'eit': 5548,\n",
       " 'ith': 6979,\n",
       " 'th ': 10122,\n",
       " 'h R': 6308,\n",
       " 'Ros': 3115,\n",
       " 'ose': 8844,\n",
       " 'Mc': 2421,\n",
       " 'nt': 8384,\n",
       " 'rr': 9533,\n",
       " 'ry': 9630,\n",
       " 'Mci': 2430,\n",
       " 'cin': 4862,\n",
       " 'int': 6877,\n",
       " 'nto': 8399,\n",
       " 'tos': 10219,\n",
       " 'osh': 8847,\n",
       " 'h S': 6309,\n",
       " 'her': 6399,\n",
       " 'err': 5709,\n",
       " 'rry': 9544,\n",
       " 'Vo': 3661,\n",
       " 'oz': 8968,\n",
       " 'zn': 11590,\n",
       " 'ny': 8437,\n",
       " 'yu': 11412,\n",
       " 'uk': 10523,\n",
       " 'Al': 636,\n",
       " 'ex': 5818,\n",
       " 'Voz': 3676,\n",
       " 'ozn': 8978,\n",
       " 'zny': 11594,\n",
       " 'nyu': 8452,\n",
       " 'yuk': 11416,\n",
       " 'uk ': 10524,\n",
       " ' Al': 16,\n",
       " 'Ale': 643,\n",
       " 'lex': 7522,\n",
       " 'ih': 6775,\n",
       " 'hu': 6558,\n",
       " 'ut': 10689,\n",
       " 't ': 9999,\n",
       " ' G': 98,\n",
       " 'Gh': 1610,\n",
       " 'eo': 5649,\n",
       " 'gh': 6150,\n",
       " 'Mih': 2473,\n",
       " 'ihu': 6783,\n",
       " 'hut': 6577,\n",
       " 'ut ': 10690,\n",
       " 't G': 10006,\n",
       " ' Gh': 101,\n",
       " 'Ghe': 1612,\n",
       " 'heo': 6396,\n",
       " 'eor': 5663,\n",
       " 'org': 8818,\n",
       " 'rgh': 9371,\n",
       " 'ghe': 6155,\n",
       " 'Wr': 3764,\n",
       " 'ht': 6541,\n",
       " 'Ed': 1280,\n",
       " 'Wri': 3767,\n",
       " 'igh': 6764,\n",
       " 'ght': 6164,\n",
       " 'ht ': 6542,\n",
       " 't E': 10004,\n",
       " ' Ed': 70,\n",
       " 'Edw': 1293,\n",
       " 'win': 10984,\n",
       " 'Ta': 3380,\n",
       " 'em': 5605,\n",
       " 'De': 1112,\n",
       " 'bo': 4662,\n",
       " 'Tat': 3401,\n",
       " 'ate': 4410,\n",
       " 'tem': 10098,\n",
       " 'em ': 5606,\n",
       " 'm D': 7796,\n",
       " ' De': 55,\n",
       " 'Deb': 1117,\n",
       " 'ebo': 5403,\n",
       " 'bor': 4675,\n",
       " 'ora': 8812,\n",
       " 'rah': 9262,\n",
       " 'Ca': 942,\n",
       " 'ap': 4328,\n",
       " 'pp': 9129,\n",
       " 'ps': 9148,\n",
       " 'ly': 7762,\n",
       " 'Cap': 956,\n",
       " 'app': 4341,\n",
       " 'pps': 9139,\n",
       " 'ps ': 9149,\n",
       " 's K': 9672,\n",
       " ' Ke': 153,\n",
       " 'Kel': 2106,\n",
       " 'ell': 5591,\n",
       " 'lly': 7618,\n",
       " 'Br': 903,\n",
       " 'nc': 8133,\n",
       " 'Bra': 904,\n",
       " 'ran': 9268,\n",
       " 'anc': 4298,\n",
       " 'nch': 8137,\n",
       " 'ch ': 4827,\n",
       " 'h B': 6292,\n",
       " ' Br': 37,\n",
       " 'Bri': 906,\n",
       " 'rit': 9409,\n",
       " 'itt': 6988,\n",
       " 'tta': 10258,\n",
       " 'tan': 10059,\n",
       " 'any': 4319,\n",
       " 'Sm': 3263,\n",
       " ' L': 164,\n",
       " 'Lu': 2349,\n",
       " 'ua': 10364,\n",
       " 'Smi': 3266,\n",
       " 'mit': 7915,\n",
       " 'h L': 6302,\n",
       " ' Lu': 172,\n",
       " 'Lua': 2351,\n",
       " 'uan': 10375,\n",
       " 'Wa': 3689,\n",
       " 'al': 4242,\n",
       " 'l ': 7382,\n",
       " 'lo': 7639,\n",
       " 'Wal': 3699,\n",
       " 'all': 4256,\n",
       " 'll ': 7598,\n",
       " 'l D': 7386,\n",
       " 'Del': 1127,\n",
       " 'elo': 5594,\n",
       " 'lor': 7657,\n",
       " 'Wh': 3727,\n",
       " 'hi': 6418,\n",
       " 'tn': 10193,\n",
       " 'art': 4372,\n",
       " 'rti': 9577,\n",
       " 'nez': 8200,\n",
       " 'z W': 11461,\n",
       " ' Wh': 288,\n",
       " 'Whi': 3730,\n",
       " 'hit': 6438,\n",
       " 'itn': 6984,\n",
       " 'tne': 10195,\n",
       " 'ney': 8199,\n",
       " 'Ha': 1704,\n",
       " 'rm': 9454,\n",
       " ' H': 109,\n",
       " 'rv': 9615,\n",
       " 'Har': 1722,\n",
       " 'arm': 4365,\n",
       " 'rmo': 9463,\n",
       " 'mon': 7957,\n",
       " 'n H': 8052,\n",
       " ' Ha': 110,\n",
       " 'arv': 4374,\n",
       " 'rve': 9617,\n",
       " 'vey': 10805,\n",
       " 'Cu': 1050,\n",
       " 'ul': 10535,\n",
       " 'lp': 7666,\n",
       " 'au': 4428,\n",
       " 'Cul': 1057,\n",
       " 'ulp': 10552,\n",
       " 'lp ': 7667,\n",
       " 'p P': 8999,\n",
       " 'Pau': 2887,\n",
       " 'aul': 4440,\n",
       " 'We': 3709,\n",
       " 'sn': 9864,\n",
       " 'An': 676,\n",
       " 'Wei': 3717,\n",
       " 'eis': 5547,\n",
       " 'isn': 6961,\n",
       " 'sne': 9866,\n",
       " 'ner': 8192,\n",
       " 'r A': 9205,\n",
       " ' An': 18,\n",
       " 'Ani': 683,\n",
       " 'nit': 8262,\n",
       " 'Lo': 2324,\n",
       " 'of': 8606,\n",
       " 'fl': 5983,\n",
       " 'St': 3316,\n",
       " 'ph': 9064,\n",
       " 'en': 5622,\n",
       " 'Lof': 2331,\n",
       " 'ofl': 8612,\n",
       " 'fli': 5986,\n",
       " 'lin': 7568,\n",
       " ' St': 251,\n",
       " 'Ste': 3320,\n",
       " 'eph': 5676,\n",
       " 'phe': 9067,\n",
       " 'hen': 6395,\n",
       " ' Q': 224,\n",
       " 'Qu': 3008,\n",
       " 't Q': 10016,\n",
       " ' Qu': 227,\n",
       " 'Que': 3010,\n",
       " 'uen': 10448,\n",
       " 'ent': 5643,\n",
       " 'ton': 10215,\n",
       " 'eh': 5513,\n",
       " ' J': 139,\n",
       " 'Je': 1992,\n",
       " 'if': 6743,\n",
       " 'fe': 5919,\n",
       " 'Sto': 3323,\n",
       " 'one': 8747,\n",
       " 'neh': 8182,\n",
       " 'eha': 5515,\n",
       " 'm J': 7802,\n",
       " ' Je': 143,\n",
       " 'Jen': 2004,\n",
       " 'enn': 5638,\n",
       " 'nni': 8311,\n",
       " 'nif': 8248,\n",
       " 'ife': 6746,\n",
       " 'fer': 5932,\n",
       " 'Jo': 2027,\n",
       " 'oh': 8637,\n",
       " 'hn': 6474,\n",
       " 'arr': 4370,\n",
       " 'rri': 9538,\n",
       " 'is ': 6949,\n",
       " 's J': 9671,\n",
       " ' Jo': 146,\n",
       " 'Joh': 2036,\n",
       " 'ohn': 8645,\n",
       " 'hnn': 6483,\n",
       " 'nie': 8247,\n",
       " 'de': 5099,\n",
       " 'Hu': 1809,\n",
       " 'ug': 10469,\n",
       " 'ild': 6818,\n",
       " 'lde': 7484,\n",
       " 'der': 5117,\n",
       " 'r H': 9212,\n",
       " ' Hu': 115,\n",
       " 'Hug': 1817,\n",
       " 'ugh': 10476,\n",
       " 'ghi': 6157,\n",
       " 'hie': 6425,\n",
       " 'hm': 6466,\n",
       " 'rj': 9416,\n",
       " 'jo': 7123,\n",
       " 'Str': 3324,\n",
       " 'tro': 10232,\n",
       " 'roh': 9500,\n",
       " 'ohm': 8644,\n",
       " 'hma': 6468,\n",
       " 'n M': 8057,\n",
       " 'arj': 4362,\n",
       " 'rjo': 9420,\n",
       " 'jor': 7132,\n",
       " 'rie': 9394,\n",
       " 'Du': 1206,\n",
       " 'un': 10581,\n",
       " 'np': 8348,\n",
       " 'hy': 6587,\n",
       " 'rl': 9440,\n",
       " 'Dun': 1221,\n",
       " 'unp': 10599,\n",
       " 'nph': 8351,\n",
       " 'phy': 9075,\n",
       " 'hy ': 6588,\n",
       " 'y S': 11169,\n",
       " 'Shi': 3220,\n",
       " 'hir': 6436,\n",
       " 'irl': 6935,\n",
       " 'rle': 9445,\n",
       " 'Co': 1014,\n",
       " 'Ty': 3528,\n",
       " 'yf': 11272,\n",
       " 'ff': 5939,\n",
       " 'fa': 5893,\n",
       " 'Col': 1026,\n",
       " 'ole': 8698,\n",
       " 'lem': 7512,\n",
       " 'ema': 5607,\n",
       " ' Ty': 267,\n",
       " 'yff': 11274,\n",
       " 'ffa': 5942,\n",
       " 'fan': 5905,\n",
       " 'Po': 2952,\n",
       " 'my': 8029,\n",
       " 'Pol': 2962,\n",
       " 'oll': 8704,\n",
       " 'lle': 7604,\n",
       " 'y J': 11160,\n",
       " 'Jer': 2007,\n",
       " 'rem': 9341,\n",
       " 'emy': 5621,\n",
       " 'lq': 7675,\n",
       " 'uh': 10486,\n",
       " 'ho': 6489,\n",
       " 'olq': 8709,\n",
       " 'lqu': 7677,\n",
       " 'quh': 9195,\n",
       " 'uho': 10494,\n",
       " 'hou': 6508,\n",
       " 'oun': 8897,\n",
       " 'un ': 10582,\n",
       " 'no': 8320,\n",
       " 'ino': 6873,\n",
       " 'nor': 8339,\n",
       " 'or ': 8810,\n",
       " 'r J': 9214,\n",
       " 'rh': 9381,\n",
       " 'Dur': 1225,\n",
       " 'urh': 10650,\n",
       " 'rha': 9382,\n",
       " 'm A': 7793,\n",
       " 'Ant': 690,\n",
       " 'nth': 8392,\n",
       " 'tho': 10136,\n",
       " 'hon': 6502,\n",
       " 'ony': 8765,\n",
       " 'Fo': 1527,\n",
       " 'og': 8620,\n",
       " 'aw': 4463,\n",
       " 'wn': 11006,\n",
       " 'Fog': 1532,\n",
       " 'oga': 8622,\n",
       " 'rty': 9588,\n",
       " 'ty ': 10306,\n",
       " 'haw': 6359,\n",
       " 'awn': 4476,\n",
       " 'ay': 4497,\n",
       " 'yb': 11222,\n",
       " 'ray': 9279,\n",
       " 'ayb': 4501,\n",
       " 'ybo': 11228,\n",
       " 'boy': 4681,\n",
       " 'oy ': 8953,\n",
       " 'y L': 11162,\n",
       " ' La': 167,\n",
       " 'Lar': 2262,\n",
       " 'Te': 3408,\n",
       " 'Le': 2271,\n",
       " 'ew': 5799,\n",
       " 'Tel': 3420,\n",
       " 'le ': 7498,\n",
       " 'e L': 5327,\n",
       " ' Le': 168,\n",
       " 'Lew': 2295,\n",
       " 'ewi': 5806,\n",
       " 'wis': 10986,\n",
       " 'Me': 2444,\n",
       " 'lv': 7748,\n",
       " 'Bos': 895,\n",
       " 'ost': 8858,\n",
       " 'st ': 9930,\n",
       " 't M': 10012,\n",
       " ' Me': 178,\n",
       " 'Mel': 2456,\n",
       " 'elv': 5601,\n",
       " 'lvi': 7751,\n",
       " 'vin': 10819,\n",
       " 'ip': 6902,\n",
       " 'To': 3469,\n",
       " 'om': 8718,\n",
       " 'mm': 7931,\n",
       " 'Tri': 3492,\n",
       " 'rip': 9405,\n",
       " 'ipp': 6913,\n",
       " 'pp ': 9130,\n",
       " 'p T': 9002,\n",
       " ' To': 262,\n",
       " 'Tom': 3478,\n",
       " 'omm': 8729,\n",
       " 'mmy': 7940,\n",
       " 'Ba': 798,\n",
       " 'ai': 4185,\n",
       " 'ct': 4959,\n",
       " 'Bai': 806,\n",
       " 'ain': 4200,\n",
       " 'es ': 5718,\n",
       " 's V': 9683,\n",
       " 'Vic': 3643,\n",
       " 'ict': 6691,\n",
       " 'cto': 4964,\n",
       " 'tor': 10218,\n",
       " 'wm': 11003,\n",
       " 'Bow': 899,\n",
       " 'owm': 8932,\n",
       " 'wma': 11004,\n",
       " 'n J': 8054,\n",
       " 'Jes': 2008,\n",
       " 'ess': 5738,\n",
       " 'Dea': 1116,\n",
       " 'ear': 5385,\n",
       " 'rma': 9456,\n",
       " 'n W': 8067,\n",
       " 'wk': 10988,\n",
       " 'ki': 7258,\n",
       " 'Ge': 1592,\n",
       " 'Haw': 1727,\n",
       " 'awk': 4474,\n",
       " 'wki': 10992,\n",
       " 'kin': 7270,\n",
       " 'ns ': 8363,\n",
       " 's G': 9668,\n",
       " ' Ge': 100,\n",
       " 'Ger': 1605,\n",
       " 'era': 5692,\n",
       " 'ral': 9266,\n",
       " 'ald': 4248,\n",
       " 'Gr': 1668,\n",
       " 'Gra': 1669,\n",
       " 'rav': 9276,\n",
       " 'ave': 4453,\n",
       " 'ves': 10803,\n",
       " 's T': 9681,\n",
       " ' Te': 259,\n",
       " 'Ter': 3426,\n",
       " 'Ce': 965,\n",
       " 'ot': 8863,\n",
       " 'Cec': 968,\n",
       " 'eci': 5416,\n",
       " 'cil': 4860,\n",
       " 'il ': 6813,\n",
       " 'Dor': 1189,\n",
       " 'oro': 8826,\n",
       " 'rot': 9512,\n",
       " 'oth': 8868,\n",
       " 'thy': 10142,\n",
       " 'cu': 4965,\n",
       " ' C': 40,\n",
       " 'Cy': 1068,\n",
       " 'yn': 11333,\n",
       " 'Hon': 1786,\n",
       " 'eyc': 5835,\n",
       " 'ycu': 11240,\n",
       " 'cut': 4980,\n",
       " 'utt': 10704,\n",
       " 'tt ': 10256,\n",
       " 't C': 10002,\n",
       " ' Cy': 50,\n",
       " 'Cyn': 1072,\n",
       " 'ynt': 11346,\n",
       " 'thi': 10131,\n",
       " 'hia': 6421,\n",
       " 'Mo': 2487,\n",
       " 'oo': 8767,\n",
       " 'Moo': 2502,\n",
       " 'oon': 8781,\n",
       " 'gie': 6173,\n",
       " 'co': 4913,\n",
       " 'oc': 8543,\n",
       " 'Ka': 2070,\n",
       " 'Han': 1719,\n",
       " 'nco': 8141,\n",
       " 'coc': 4918,\n",
       " 'ock': 8550,\n",
       " 'k K': 7163,\n",
       " ' Ka': 152,\n",
       " 'Kat': 2089,\n",
       " 'ati': 4413,\n",
       " 'tie': 10150,\n",
       " 'Gw': 1696,\n",
       " 'we': 10945,\n",
       " 'do': 5208,\n",
       " 'ail': 4198,\n",
       " 'ile': 6819,\n",
       " 'y G': 11157,\n",
       " ' Gw': 107,\n",
       " 'Gwe': 1698,\n",
       " 'wen': 10957,\n",
       " 'end': 5628,\n",
       " 'ndo': 8161,\n",
       " 'dol': 5220,\n",
       " 'oly': 8716,\n",
       " 'lyn': 7775,\n",
       " 'Ir': 1923,\n",
       " 'ng': 8209,\n",
       " 'Irv': 1930,\n",
       " 'rvi': 9618,\n",
       " 'ing': 6866,\n",
       " 'ng ': 8210,\n",
       " 'g B': 6040,\n",
       " 'Bre': 905,\n",
       " 'ren': 9342,\n",
       " 'uc': 10399,\n",
       " 'Na': 2543,\n",
       " 'Luc': 2353,\n",
       " 'uce': 10403,\n",
       " 'cey': 4813,\n",
       " 'y N': 11164,\n",
       " ' Na': 184,\n",
       " 'Nan': 2557,\n",
       " 'nci': 8138,\n",
       " 'He': 1730,\n",
       " 'nr': 8355,\n",
       " 'ya': 11199,\n",
       " 'Hen': 1741,\n",
       " 'enr': 5641,\n",
       " 'nry': 8361,\n",
       " 'ry ': 9631,\n",
       " 'toy': 10225,\n",
       " 'oya': 8955,\n",
       " 'lt': 7709,\n",
       " 'Cr': 1040,\n",
       " 'ys': 11385,\n",
       " 'alt': 4263,\n",
       " 'lto': 7719,\n",
       " 'n C': 8047,\n",
       " ' Cr': 48,\n",
       " 'Cry': 1046,\n",
       " 'rys': 9648,\n",
       " 'yst': 11397,\n",
       " 'sta': 9932,\n",
       " 'tal': 10057,\n",
       " 'Bar': 815,\n",
       " 'rr ': 9534,\n",
       " ' He': 111,\n",
       " 'Hea': 1731,\n",
       " 'eat': 5387,\n",
       " 'ath': 4412,\n",
       " 'the': 10127,\n",
       " 'ee': 5452,\n",
       " 'Gre': 1670,\n",
       " 'ree': 9333,\n",
       " 'ees': 5472,\n",
       " 'eso': 5734,\n",
       " 'n B': 8046,\n",
       " ' Bi': 34,\n",
       " 'Bil': 858,\n",
       " 'rn': 9468,\n",
       " 'Di': 1145,\n",
       " 'ix': 7023,\n",
       " 'xi': 11116,\n",
       " 'Ker': 2111,\n",
       " 'ern': 5705,\n",
       " 'rno': 9484,\n",
       " 'noh': 8330,\n",
       " 'oha': 8639,\n",
       " ' Di': 57,\n",
       " 'Dix': 1165,\n",
       " 'ixi': 7029,\n",
       " 'xie': 11118,\n",
       " 'ke': 7224,\n",
       " 'Gi': 1615,\n",
       " 'Doc': 1177,\n",
       " 'cke': 4879,\n",
       " 'ker': 7239,\n",
       " 'ery': 5715,\n",
       " ' Gi': 102,\n",
       " 'Gir': 1627,\n",
       " 'irt': 6942,\n",
       " 'rth': 9576,\n",
       " ' F': 90,\n",
       " 'nk': 8274,\n",
       " 'kl': 7283,\n",
       " 'Boo': 892,\n",
       " 'ne ': 8172,\n",
       " 'e F': 5321,\n",
       " ' Fr': 96,\n",
       " 'Fra': 1546,\n",
       " 'ank': 4306,\n",
       " 'nkl': 8282,\n",
       " 'kli': 7287,\n",
       " 'Ja': 1967,\n",
       " 'Jac': 1971,\n",
       " 'ack': 4071,\n",
       " 'k J': 7162,\n",
       " ' Ja': 142,\n",
       " 'Jar': 1984,\n",
       " 'rro': 9541,\n",
       " 'rod': 9496,\n",
       " 'oa': 8509,\n",
       " 'mse': 8000,\n",
       " 'Joa': 2030,\n",
       " 'oan': 8521,\n",
       " 'Ve': 3625,\n",
       " 'si': 9815,\n",
       " 'Vel': 3632,\n",
       " 'ele': 5584,\n",
       " 'lez': 7524,\n",
       " 'z C': 11441,\n",
       " ' Ca': 42,\n",
       " 'Cas': 958,\n",
       " 'asi': 4390,\n",
       " 'sil': 9825,\n",
       " 'lda': 7482,\n",
       " 'ax': 4484,\n",
       " 'x ': 11072,\n",
       " ...}"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vocab"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "from itertools import chain\n",
    "\n",
    "def n_grams(seq, n=1):\n",
    "    \"\"\"Returns an itirator over the n-grams given a listTokens\"\"\"\n",
    "    shiftToken = lambda i: (el for j,el in enumerate(seq) if j>=i)\n",
    "    shiftedTokens = (shiftToken(i) for i in range(n))\n",
    "    tupleNGrams = zip(*shiftedTokens)\n",
    "    return tupleNGrams # if join in generator : (\" \".join(i) for i in tupleNGrams)\n",
    "\n",
    "def range_ngrams(listTokens, ngramRange=(1,2)):\n",
    "    \"\"\"Returns an itirator over all n-grams for n in range(ngramRange) given a listTokens.\"\"\"\n",
    "    ngrams = (ngramRange[0], ngramRange[1] + 1)\n",
    "    return chain(*(n_grams(listTokens, i) for i in range(*ngramRange)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Max feature len = 37, Avg. feature len = 12\n",
      "CPU times: user 1min 40s, sys: 472 ms, total: 1min 41s\n",
      "Wall time: 1min 41s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "def find_ngrams(text, ngrams):\n",
    "    a = range_ngrams(text, ngrams)\n",
    "    wi = []\n",
    "    for i in a:\n",
    "        w = ''.join(i)\n",
    "        try:\n",
    "            idx = words_list.index(w)\n",
    "        except:\n",
    "            idx = 0\n",
    "        wi.append(idx)\n",
    "    return wi\n",
    "\n",
    "# build X from index of n-gram sequence\n",
    "X = np.array(sdf.last_name_first_name.apply(lambda c: find_ngrams(c, NGRAMS)))\n",
    "\n",
    "# check max/avg feature\n",
    "X_len = []\n",
    "for x in X:\n",
    "    X_len.append(len(x))\n",
    "\n",
    "max_feature_len = max(X_len)\n",
    "avg_feature_len = int(np.mean(X_len))\n",
    "\n",
    "print(\"Max feature len = %d, Avg. feature len = %d\" % (max_feature_len, avg_feature_len))\n",
    "y = np.array(sdf.ethnic_code.astype('category').cat.codes)\n",
    "\n",
    "# Split train and test dataset\n",
    "X_train,  X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=21, stratify=y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train a LSTM model\n",
    "\n",
    "ref: http://machinelearningmastery.com/sequence-classification-lstm-recurrent-neural-networks-python-keras/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "800000 train sequences\n",
      "200000 test sequences\n",
      "Pad sequences (samples x time)\n",
      "X_train shape: (800000, 25)\n",
      "X_test shape: (200000, 25)\n",
      "2 classes\n",
      "Convert class vector to binary class matrix (for use with categorical_crossentropy)\n",
      "y_train shape: (800000, 2)\n",
      "y_test shape: (200000, 2)\n"
     ]
    }
   ],
   "source": [
    "import keras\n",
    "from keras.preprocessing import sequence\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Embedding, Dropout, Activation\n",
    "from keras.layers import LSTM\n",
    "from keras.layers.convolutional import Conv1D\n",
    "from keras.layers.convolutional import MaxPooling1D\n",
    "from keras.models import load_model\n",
    "\n",
    "max_features = num_words # 20000\n",
    "feature_len = 25 # avg_feature_len # cut texts after this number of words (among top max_features most common words)\n",
    "batch_size = 32\n",
    "\n",
    "print(len(X_train), 'train sequences')\n",
    "print(len(X_test), 'test sequences')\n",
    "\n",
    "print('Pad sequences (samples x time)')\n",
    "X_train = sequence.pad_sequences(X_train, maxlen=feature_len)\n",
    "X_test = sequence.pad_sequences(X_test, maxlen=feature_len)\n",
    "print('X_train shape:', X_train.shape)\n",
    "print('X_test shape:', X_test.shape)\n",
    "\n",
    "num_classes = np.max(y_train) + 1\n",
    "print(num_classes, 'classes')\n",
    "\n",
    "print('Convert class vector to binary class matrix '\n",
    "      '(for use with categorical_crossentropy)')\n",
    "y_train = keras.utils.to_categorical(y_train, num_classes)\n",
    "y_test = keras.utils.to_categorical(y_test, num_classes)\n",
    "print('y_train shape:', y_train.shape)\n",
    "print('y_test shape:', y_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Build model...\n",
      "WARNING:tensorflow:From /opt/venv37/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:74: The name tf.get_default_graph is deprecated. Please use tf.compat.v1.get_default_graph instead.\n",
      "\n",
      "WARNING:tensorflow:From /opt/venv37/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:517: The name tf.placeholder is deprecated. Please use tf.compat.v1.placeholder instead.\n",
      "\n",
      "WARNING:tensorflow:From /opt/venv37/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:4138: The name tf.random_uniform is deprecated. Please use tf.random.uniform instead.\n",
      "\n",
      "WARNING:tensorflow:From /opt/venv37/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:3976: The name tf.nn.max_pool is deprecated. Please use tf.nn.max_pool2d instead.\n",
      "\n",
      "WARNING:tensorflow:From /opt/venv37/lib/python3.7/site-packages/keras/optimizers.py:790: The name tf.train.Optimizer is deprecated. Please use tf.compat.v1.train.Optimizer instead.\n",
      "\n",
      "WARNING:tensorflow:From /opt/venv37/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:3295: The name tf.log is deprecated. Please use tf.math.log instead.\n",
      "\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding_1 (Embedding)      (None, 25, 50)            583300    \n",
      "_________________________________________________________________\n",
      "conv1d_1 (Conv1D)            (None, 25, 32)            4832      \n",
      "_________________________________________________________________\n",
      "max_pooling1d_1 (MaxPooling1 (None, 12, 32)            0         \n",
      "_________________________________________________________________\n",
      "lstm_1 (LSTM)                (None, 100)               53200     \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 2)                 202       \n",
      "=================================================================\n",
      "Total params: 641,534\n",
      "Trainable params: 641,534\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "print('Build model...')\n",
    "\n",
    "if False:\n",
    "    model = Sequential()\n",
    "    model.add(Embedding(num_words, 32, input_length=feature_len))\n",
    "    model.add(LSTM(128, dropout=0.2, recurrent_dropout=0.2))\n",
    "    model.add(Dense(num_classes, activation='softmax'))\n",
    "if True:\n",
    "    embedding_vecor_length = 50\n",
    "    model = Sequential()\n",
    "    model.add(Embedding(num_words, embedding_vecor_length, input_length=feature_len))\n",
    "    model.add(Conv1D(activation=\"relu\", padding=\"same\", filters=32, kernel_size=3))\n",
    "    model.add(MaxPooling1D(pool_size=2))\n",
    "    model.add(LSTM(100))\n",
    "    model.add(Dense(num_classes, activation='softmax'))\n",
    "\n",
    "# try using different optimizers and different optimizer configs\n",
    "model.compile(loss='categorical_crossentropy',\n",
    "              optimizer='adam',\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "print(model.summary())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train...\n",
      "WARNING:tensorflow:From /opt/venv37/lib/python3.7/site-packages/tensorflow_core/python/ops/math_grad.py:1424: where (from tensorflow.python.ops.array_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use tf.where in 2.0, which has the same broadcast rule as np.where\n",
      "WARNING:tensorflow:From /opt/venv37/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:986: The name tf.assign_add is deprecated. Please use tf.compat.v1.assign_add instead.\n",
      "\n",
      "WARNING:tensorflow:From /opt/venv37/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:973: The name tf.assign is deprecated. Please use tf.compat.v1.assign instead.\n",
      "\n",
      "WARNING:tensorflow:From /opt/venv37/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:2741: The name tf.Session is deprecated. Please use tf.compat.v1.Session instead.\n",
      "\n",
      "Train on 720000 samples, validate on 80000 samples\n",
      "Epoch 1/15\n",
      "WARNING:tensorflow:From /opt/venv37/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:174: The name tf.get_default_session is deprecated. Please use tf.compat.v1.get_default_session instead.\n",
      "\n",
      "WARNING:tensorflow:From /opt/venv37/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:181: The name tf.ConfigProto is deprecated. Please use tf.compat.v1.ConfigProto instead.\n",
      "\n",
      "WARNING:tensorflow:From /opt/venv37/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:190: The name tf.global_variables is deprecated. Please use tf.compat.v1.global_variables instead.\n",
      "\n",
      "WARNING:tensorflow:From /opt/venv37/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:199: The name tf.is_variable_initialized is deprecated. Please use tf.compat.v1.is_variable_initialized instead.\n",
      "\n",
      "WARNING:tensorflow:From /opt/venv37/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:206: The name tf.variables_initializer is deprecated. Please use tf.compat.v1.variables_initializer instead.\n",
      "\n",
      "720000/720000 [==============================] - 533s 740us/step - loss: 0.0765 - acc: 0.9771 - val_loss: 0.0703 - val_acc: 0.9788\n",
      "Epoch 2/15\n",
      "720000/720000 [==============================] - 506s 703us/step - loss: 0.0699 - acc: 0.9788 - val_loss: 0.0687 - val_acc: 0.9795\n",
      "Epoch 3/15\n",
      "720000/720000 [==============================] - 536s 745us/step - loss: 0.0679 - acc: 0.9794 - val_loss: 0.0684 - val_acc: 0.9800\n",
      "Epoch 4/15\n",
      "720000/720000 [==============================] - 517s 718us/step - loss: 0.0665 - acc: 0.9798 - val_loss: 0.0692 - val_acc: 0.9794\n",
      "Epoch 5/15\n",
      "720000/720000 [==============================] - 525s 730us/step - loss: 0.0654 - acc: 0.9800 - val_loss: 0.0681 - val_acc: 0.9796\n",
      "Epoch 6/15\n",
      "720000/720000 [==============================] - 542s 753us/step - loss: 0.0647 - acc: 0.9804 - val_loss: 0.0678 - val_acc: 0.9800\n",
      "Epoch 7/15\n",
      "720000/720000 [==============================] - 500s 694us/step - loss: 0.0638 - acc: 0.9807 - val_loss: 0.0685 - val_acc: 0.9796\n",
      "Epoch 8/15\n",
      "720000/720000 [==============================] - 538s 747us/step - loss: 0.0632 - acc: 0.9810 - val_loss: 0.0674 - val_acc: 0.9803\n",
      "Epoch 9/15\n",
      "720000/720000 [==============================] - 516s 717us/step - loss: 0.0627 - acc: 0.9810 - val_loss: 0.0680 - val_acc: 0.9799\n",
      "Epoch 10/15\n",
      "720000/720000 [==============================] - 505s 701us/step - loss: 0.0622 - acc: 0.9812 - val_loss: 0.0694 - val_acc: 0.9798\n",
      "Epoch 11/15\n",
      "720000/720000 [==============================] - 490s 681us/step - loss: 0.0622 - acc: 0.9811 - val_loss: 0.0687 - val_acc: 0.9790\n",
      "Epoch 12/15\n",
      "720000/720000 [==============================] - 475s 660us/step - loss: 0.0617 - acc: 0.9813 - val_loss: 0.0692 - val_acc: 0.9796\n",
      "Epoch 13/15\n",
      "720000/720000 [==============================] - 493s 684us/step - loss: 0.0617 - acc: 0.9813 - val_loss: 0.0687 - val_acc: 0.9799\n",
      "Epoch 14/15\n",
      "720000/720000 [==============================] - 501s 695us/step - loss: 0.0616 - acc: 0.9813 - val_loss: 0.0689 - val_acc: 0.9796\n",
      "Epoch 15/15\n",
      "720000/720000 [==============================] - 500s 694us/step - loss: 0.0615 - acc: 0.9814 - val_loss: 0.0693 - val_acc: 0.9795\n",
      "200000/200000 [==============================] - 25s 124us/step\n",
      "Test score: 0.0703102104614675\n",
      "Test accuracy: 0.97842\n",
      "CPU times: user 6h 33min 37s, sys: 54min 13s, total: 7h 27min 51s\n",
      "Wall time: 2h 8min 22s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "print('Train...')\n",
    "model.fit(X_train, y_train, batch_size=batch_size, epochs=EPOCHS,\n",
    "          validation_split=0.1, verbose=1)\n",
    "score, acc = model.evaluate(X_test, y_test,\n",
    "                            batch_size=batch_size, verbose=1)\n",
    "print('Test score:', score)\n",
    "print('Test accuracy:', acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test score: 0.0703102104614675\n",
      "Test accuracy: 0.97842\n"
     ]
    }
   ],
   "source": [
    "print('Test score:', score)\n",
    "print('Test accuracy:', acc)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Confusion Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "          HL       0.65      0.55      0.59      5747\n",
      "          NL       0.99      0.99      0.99    194253\n",
      "\n",
      "    accuracy                           0.98    200000\n",
      "   macro avg       0.82      0.77      0.79    200000\n",
      "weighted avg       0.98      0.98      0.98    200000\n",
      "\n",
      "[[  3152   2595]\n",
      " [  1721 192532]]\n"
     ]
    }
   ],
   "source": [
    "y_pred = model.predict_classes(X_test, verbose=2)\n",
    "p = model.predict_proba(X_test, verbose=2) # to predict probability\n",
    "target_names = list(sdf.ethnic_code.astype('category').cat.categories)\n",
    "print(classification_report(np.argmax(y_test, axis=1), y_pred, target_names=target_names))\n",
    "\n",
    "conf_mat = confusion_matrix(np.argmax(y_test, axis=1), y_pred)\n",
    "print(conf_mat)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "\n",
    "\n",
    "def plot_confusion_matrix(cm,\n",
    "                          target_names,\n",
    "                          title='Confusion matrix',\n",
    "                          cmap=None,\n",
    "                          normalize=True):\n",
    "    \"\"\"\n",
    "    given a sklearn confusion matrix (cm), make a nice plot\n",
    "\n",
    "    Arguments\n",
    "    ---------\n",
    "    cm:           confusion matrix from sklearn.metrics.confusion_matrix\n",
    "\n",
    "    target_names: given classification classes such as [0, 1, 2]\n",
    "                  the class names, for example: ['high', 'medium', 'low']\n",
    "\n",
    "    title:        the text to display at the top of the matrix\n",
    "\n",
    "    cmap:         the gradient of the values displayed from matplotlib.pyplot.cm\n",
    "                  see http://matplotlib.org/examples/color/colormaps_reference.html\n",
    "                  plt.get_cmap('jet') or plt.cm.Blues\n",
    "\n",
    "    normalize:    If False, plot the raw numbers\n",
    "                  If True, plot the proportions\n",
    "\n",
    "    Usage\n",
    "    -----\n",
    "    plot_confusion_matrix(cm           = cm,                  # confusion matrix created by\n",
    "                                                              # sklearn.metrics.confusion_matrix\n",
    "                          normalize    = True,                # show proportions\n",
    "                          target_names = y_labels_vals,       # list of names of the classes\n",
    "                          title        = best_estimator_name) # title of graph\n",
    "\n",
    "    Citiation\n",
    "    ---------\n",
    "    http://scikit-learn.org/stable/auto_examples/model_selection/plot_confusion_matrix.html\n",
    "\n",
    "    \"\"\"\n",
    "    import matplotlib.pyplot as plt\n",
    "    import numpy as np\n",
    "    import itertools\n",
    "\n",
    "    accuracy = np.trace(cm) / float(np.sum(cm))\n",
    "    misclass = 1 - accuracy\n",
    "\n",
    "    if cmap is None:\n",
    "        cmap = plt.get_cmap('Blues')\n",
    "\n",
    "    plt.figure(figsize=(8, 6))\n",
    "    plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
    "    plt.title(title)\n",
    "    plt.colorbar()\n",
    "\n",
    "    if target_names is not None:\n",
    "        tick_marks = np.arange(len(target_names))\n",
    "        plt.xticks(tick_marks, target_names, rotation=45)\n",
    "        plt.yticks(tick_marks, target_names)\n",
    "\n",
    "    if normalize:\n",
    "        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
    "\n",
    "\n",
    "    thresh = cm.max() / 1.5 if normalize else cm.max() / 2\n",
    "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
    "        if normalize:\n",
    "            plt.text(j, i, \"{:0.4f}\".format(cm[i, j]),\n",
    "                     horizontalalignment=\"center\",\n",
    "                     color=\"white\" if cm[i, j] > thresh else \"black\")\n",
    "        else:\n",
    "            plt.text(j, i, \"{:,}\".format(cm[i, j]),\n",
    "                     horizontalalignment=\"center\",\n",
    "                     color=\"white\" if cm[i, j] > thresh else \"black\")\n",
    "\n",
    "\n",
    "    plt.tight_layout()\n",
    "    plt.ylabel('True label')\n",
    "    plt.xlabel('Predicted label\\naccuracy={:0.4f}; misclass={:0.4f}'.format(accuracy, misclass))\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_confusion_matrix(conf_mat, target_names, normalize=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Save model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "#model.save('../models/nc_voter_name_lstm.h5')\n",
    "model.save('/opt/github/ethnicolr/ethnicolr/models/nc_voter_reg/nc_voter_name_hispanic_lstm.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "words_df = pd.DataFrame(words_list, columns=['vocab'])\n",
    "#words_df.to_csv('../models/nc_voter_name_vocab.csv', index=False, encoding='utf-8')\n",
    "words_df.to_csv('/opt/github/ethnicolr/ethnicolr/models/nc_voter_reg/nc_voter_name_vocab.csv', index=False, encoding='utf-8')"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}