wiki_bigrams.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from ipywidgets import FloatProgress, IntText\n",
    "import time\n",
    "import collections\n",
    "from math import log \n",
    "import pickle\n",
    "import os\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "def save_object(obj, filename):\n",
    "    with open(filename, 'wb') as outfile:  \n",
    "        pickle.dump(obj, outfile, pickle.HIGHEST_PROTOCOL)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('data/is_male.pkl', 'rb') as infile:\n",
    "    is_male = pickle.load(infile)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Python version + compiler & build info: 3.6.5 |Anaconda custom (64-bit)| (default, Mar 29 2018, 18:21:58) \n",
      "[GCC 7.2.0]\n",
      "Jupyter notebook version: 5.4.1\n",
      "\n"
     ]
    }
   ],
   "source": [
    "import sys\n",
    "print(\"Python version + compiler & build info:\", sys.version)\n",
    "print(\"Jupyter notebook version:\", os.popen(\"jupyter notebook --version\").readlines()[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Create Counters of Bigrams in Biographies of Women and Men, in Batches"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# This is purportedly the fastest way to get bigrams\n",
    "def get_bigrams(text):\n",
    "    return zip(*[text[i:] for i in range(2)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Building batches of counters for bigrams found in all Wiki bios of women and of men...\n",
      "Estimated time for completion: 46 minutes\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f1d1584e466048f1b6a595c2e7d13e59",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "A Jupyter Widget"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e0105367b8ee44cfb3d72185593afafd",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "A Jupyter Widget"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Articles: 850399 Time: 2681.5078961849213 seconds\n"
     ]
    }
   ],
   "source": [
    "# This cell takes about 46 minutes to run.\n",
    "\n",
    "male_bigrams = collections.Counter()\n",
    "female_bigrams = collections.Counter()\n",
    "\n",
    "print(\"Building batches of counters for bigrams found in all Wiki bios of women and of men...\")\n",
    "print(\"Estimated time for completion: 46 minutes\")\n",
    "\n",
    "fp = FloatProgress(min=0, max=850399)\n",
    "all_articles = IntText(value=0, description=\"Article Count:\")\n",
    "display(fp, all_articles)\n",
    "article_count = 0\n",
    "\n",
    "if not os.path.exists('data/bigrams'):\n",
    "    os.makedirs('data/bigrams')\n",
    "\n",
    "start = time.time()\n",
    "try:\n",
    "    with open(\"data/gendered-labeled-articles.stripped\", 'r') as f:\n",
    "        for line in f:\n",
    "            article_count += 1\n",
    "            all_articles.value += 1\n",
    "            if article_count % 1000 == 0:\n",
    "                fp.value = article_count\n",
    "            line = line.split()\n",
    "            male = is_male[line.pop(0)]\n",
    "            # Read text in batches of 100 words\n",
    "            line = [line[i-1 : i+99] for i in range(1, len(line), 99)]\n",
    "            while line:\n",
    "                bigrams = list(get_bigrams(line.pop(0)))\n",
    "                if male:\n",
    "                    male_bigrams.update(bigrams)\n",
    "                else:\n",
    "                    female_bigrams.update(bigrams)\n",
    "            # Pickle results after every 1000 articles for men; 3000 for women (since we are using\n",
    "            # a total article count, and we know there are fewer bios of women). These numbers seem\n",
    "            # to be about right to prevent memory errors.\n",
    "            if article_count % 1000 == 0:\n",
    "                save_object(male_bigrams, 'data/bigrams/male_bigrams_{}.pkl'.format(article_count // 1000))\n",
    "                male_bigrams = collections.Counter()\n",
    "                if article_count % 3000 == 0:\n",
    "                    save_object(female_bigrams, 'data/bigrams/female_bigrams_{}.pkl'.format(article_count // 3000))\n",
    "                    female_bigrams = collections.Counter()\n",
    "        save_object(male_bigrams, 'data/bigrams/male_bigrams_{}.pkl'.format(article_count // 1000 + 1))\n",
    "        save_object(female_bigrams, 'data/bigrams/female_bigrams_{}.pkl'.format(article_count // 3000 + 1))\n",
    "finally: \n",
    "    print(\"\\nArticles:\", article_count, \"Time:\", (time.time() - start), \"seconds\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of batches of bigrams for women collected: 284\n",
      "Number of batches of bigrams for men collected: 851\n"
     ]
    }
   ],
   "source": [
    "total_batches_female = article_count // 3000 + 1  \n",
    "total_batches_male = article_count // 1000 + 1 \n",
    "\n",
    "print(\"Number of batches of bigrams for women collected:\", total_batches_female)\n",
    "print(\"Number of batches of bigrams for men collected:\", total_batches_male)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Merging batches of the female_bigrams counters. Removing bigrams with counts <= 2, six times.\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "cce57ad8b73f4691a6ebb6e7e056c7aa",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "A Jupyter Widget"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "9c73e3a3a21447699c15af73bd382638",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "A Jupyter Widget"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Done.\n",
      "Master counter for female bigrams saved in 'data/bigrams/female_bigrams_master.pkl'\n",
      "Time: 552.6426672935486 seconds\n"
     ]
    }
   ],
   "source": [
    "# This cell takes about 10 minutes to run.\n",
    "\n",
    "print(\"Merging batches of the female_bigrams counters. Removing bigrams with counts <= 2, six times.\")\n",
    "\n",
    "fp = FloatProgress(min=0, max=total_batches_female)\n",
    "count = IntText(value=0, description=\"Count:\")\n",
    "display(fp, count)\n",
    "\n",
    "f_count = 0\n",
    "female_bigrams = collections.Counter()\n",
    "\n",
    "start = time.time()\n",
    "for i in range(1, total_batches_female + 1):\n",
    "    fp.value += 1\n",
    "    count.value += 1\n",
    "    f_count += 1\n",
    "    with open('data/bigrams/female_bigrams_{}.pkl'.format(i), 'rb') as infile:\n",
    "        female_bigrams += pickle.load(infile)\n",
    "        # Remove bigrams with counts <=2 six times.\n",
    "        if f_count % 48 == 0:\n",
    "            for k in set(k for k,v in female_bigrams.items() if v <= 2):\n",
    "                del female_bigrams[k]\n",
    "for k in set(k for k,v in female_bigrams.items() if v <= 2):\n",
    "    del female_bigrams[k]  \n",
    "save_object(female_bigrams, 'data/bigrams/female_bigrams_master.pkl')\n",
    "print(\"Done.\\nMaster counter for female bigrams saved in 'data/bigrams/female_bigrams_master.pkl'\")\n",
    "print(\"Time:\", (time.time() - start), \"seconds\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total number of distinct bigrams collected from bios of women: 2725813\n"
     ]
    }
   ],
   "source": [
    "print(\"Total number of distinct bigrams collected from bios of women:\", len(female_bigrams))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Deleting temporary batches...\n",
      "Done.\n"
     ]
    }
   ],
   "source": [
    "print(\"Deleting temporary batches...\")\n",
    "for i in range(1, total_batches_female + 1):\n",
    "    os.remove('data/bigrams/female_bigrams_{}.pkl'.format(i))\n",
    "print(\"Done.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "779b467daaad492a89d8d09d3b218636",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "A Jupyter Widget"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "81438b73cf784bba980e571cfe630ef3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "A Jupyter Widget"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Time: 1240.206268787384 seconds\n",
      "Number of batches of male bigrams reduced to 48\n"
     ]
    }
   ],
   "source": [
    "# This cell takes about 20 minutes to run.\n",
    "\n",
    "# NOTE: If you get a memory error here, restart kernel and re-run first cell of notebook only.\n",
    "# Then try again (a hacky way to solve memory issue that seems to work.)\n",
    "\n",
    "# Delete female_bigrams to save memory.  We can load it later from pickle file.\n",
    "# del female_bigrams\n",
    "total_batches_male = 851\n",
    "\n",
    "fp = FloatProgress(min=0, max=total_batches_male)\n",
    "count = IntText(value=0, description=\"Count:\")\n",
    "m_count = 0\n",
    "display(fp, count)\n",
    "\n",
    "male_bigrams = collections.Counter()\n",
    "n = 7\n",
    "\n",
    "start = time.time()\n",
    "for i in range(1, total_batches_male + 1):\n",
    "    fp.value += 1\n",
    "    count.value += 1\n",
    "    m_count += 1\n",
    "    with open('data/bigrams/male_bigrams_{}.pkl'.format(i), 'rb') as infile:\n",
    "        male_bigrams += pickle.load(infile)\n",
    "        # Remove bigrams with count <= n forty-eight times.\n",
    "        if m_count % 18 == 0:\n",
    "            for k in set(k for k,v in male_bigrams.items() if v <= n ):\n",
    "                del male_bigrams[k]\n",
    "            save_object(male_bigrams, 'data/bigrams/male_bigrams_b_{}.pkl'.format(m_count // 18))\n",
    "            male_bigrams = collections.Counter()\n",
    "# for k in set(k for k,v in male_bigrams.items() if v <= n ):\n",
    "#     del male_bigrams[k]\n",
    "save_object(male_bigrams, 'data/bigrams/male_bigrams_b_{}.pkl'.format(m_count // 18 + 1))\n",
    "male_bigrams = collections.Counter()\n",
    "print(\"Time:\", (time.time() - start), \"seconds\")\n",
    "print(\"Number of batches of male bigrams reduced to\", m_count // 18 + 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "82fc2731d2614302923bf363c5123f36",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "A Jupyter Widget"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "fa0183d35050417db0ad4a024fd87f10",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "A Jupyter Widget"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Creating final master counter for male bigrams...\n",
      "This should take under 2 minutes...\n",
      "Done.\n",
      "Master counter for male bigrams saved in 'data/bigrams/male_bigrams_master.pkl'\n",
      "Time: 35.66937756538391 seconds\n",
      "Total number of distinct bigrams collected from bios of men: 1612684\n"
     ]
    }
   ],
   "source": [
    "# This cell takes less than 2 minutes to run.\n",
    "\n",
    "fp = FloatProgress(min=0, max=48)\n",
    "count = IntText(value=0, description=\"Count:\")\n",
    "display(fp, count)\n",
    "\n",
    "print(\"Creating final master counter for male bigrams...\")\n",
    "print(\"This should take under 2 minutes...\")\n",
    "\n",
    "start = time.time()\n",
    "for i in range(1, 49):\n",
    "    fp.value += 1\n",
    "    count.value += 1\n",
    "    with open('data/bigrams/male_bigrams_b_{}.pkl'.format(i), 'rb') as infile:\n",
    "        male_bigrams += pickle.load(infile)\n",
    "# Limiting the number of bigrams to those with counts of 8 or more helps save memory\n",
    "for k in set(k for k,v in male_bigrams.items() if v < 8):\n",
    "    del male_bigrams[k]\n",
    "save_object(male_bigrams, 'data/bigrams/male_bigrams_master.pkl')\n",
    "print(\"Done.\\nMaster counter for male bigrams saved in 'data/bigrams/male_bigrams_master.pkl'\")\n",
    "print(\"Time:\", (time.time() - start), \"seconds\")\n",
    "print(\"Total number of distinct bigrams collected from bios of men:\", len(male_bigrams))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Deleting temporary batches...\n",
      "Done.\n"
     ]
    }
   ],
   "source": [
    "print(\"Deleting temporary batches...\")\n",
    "for i in range(1, total_batches_male + 1):\n",
    "    os.remove('data/bigrams/male_bigrams_{}.pkl'.format(i))\n",
    "for i in range(1, m_count // 18 + 2):\n",
    "    os.remove('data/bigrams/male_bigrams_b_{}.pkl'.format(i))\n",
    "print(\"Done.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load female_bigrams_master:\n",
    "with open('data/bigrams/female_bigrams_master.pkl', 'rb') as infile:\n",
    "    female_bigrams = pickle.load(infile)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1331985"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "for k in set(k for k,v in female_bigrams.items() if v < 8):\n",
    "    del female_bigrams[k]\n",
    "len(female_bigrams)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1612684"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Load male_bigrams_master:\n",
    "with open('data/bigrams/male_bigrams_master.pkl', 'rb') as infile:\n",
    "    male_bigrams = pickle.load(infile)\n",
    "len(male_bigrams)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# Exploratory Data Analysis of Bigrams"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(('http', 'www'), 813722),\n",
       " (('of', 'the'), 708824),\n",
       " (('ref', 'name'), 683958),\n",
       " (('url', 'http'), 578411),\n",
       " (('in', 'the'), 566040),\n",
       " (('ref', 'cite'), 448580),\n",
       " (('cite', 'web'), 440215),\n",
       " (('at', 'the'), 391527),\n",
       " (('web', 'url'), 299257),\n",
       " (('she', 'was'), 251644),\n",
       " (('women', 's'), 226877),\n",
       " (('for', 'the'), 190627),\n",
       " (('cite', 'news'), 179157),\n",
       " (('new', 'york'), 175253),\n",
       " (('to', 'the'), 173509),\n",
       " (('on', 'the'), 173402),\n",
       " (('and', 'the'), 169740),\n",
       " (('ref', 'http'), 160787),\n",
       " (('as', 'a'), 150900),\n",
       " (('birth', 'date'), 149470),\n",
       " (('ref', 'ref'), 144628),\n",
       " (('align', 'center'), 141886),\n",
       " (('united', 'states'), 121101),\n",
       " (('url', 'https'), 117811),\n",
       " (('in', 'a'), 110508),\n",
       " (('births', 'category'), 109660),\n",
       " (('ref', 'she'), 108566),\n",
       " (('living', 'people'), 107575),\n",
       " (('people', 'category'), 107094),\n",
       " (('archive', 'org'), 104657),\n",
       " (('category', 'american'), 104348),\n",
       " (('of', 'her'), 103947),\n",
       " (('university', 'of'), 102493),\n",
       " (('web', 'archive'), 100422),\n",
       " (('org', 'web'), 100273),\n",
       " (('from', 'the'), 98954),\n",
       " (('https', 'web'), 98542),\n",
       " (('is', 'a'), 97840),\n",
       " (('was', 'a'), 96771),\n",
       " (('tv', 'series'), 95971),\n",
       " (('references', 'reflist'), 95483),\n",
       " (('with', 'the'), 94900),\n",
       " (('ref', 'in'), 94165),\n",
       " (('category', 'living'), 92544),\n",
       " (('external', 'links'), 87502),\n",
       " (('the', 'first'), 86477),\n",
       " (('co', 'uk'), 85168),\n",
       " (('html', 'title'), 82992),\n",
       " (('web', 'title'), 81758),\n",
       " (('style', 'background'), 80678),\n",
       " (('https', 'www'), 79878),\n",
       " (('birth', 'place'), 79359),\n",
       " (('news', 'url'), 79138),\n",
       " (('death', 'date'), 77860),\n",
       " (('by', 'the'), 75581),\n",
       " (('and', 'age'), 73603),\n",
       " (('summer', 'olympics'), 72681),\n",
       " (('the', 'united'), 72487),\n",
       " (('she', 'is'), 72233),\n",
       " (('category', 'people'), 72171),\n",
       " (('as', 'the'), 71198),\n",
       " (('date', 'and'), 70526),\n",
       " (('date', 'birth'), 70489),\n",
       " (('cite', 'book'), 70159),\n",
       " (('was', 'the'), 67701),\n",
       " (('she', 'has'), 67227),\n",
       " (('text', 'align'), 67077),\n",
       " (('people', 'from'), 66838),\n",
       " (('the', 'new'), 65989),\n",
       " (('u', 's'), 65686)]"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "female_bigrams.most_common(70)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(('of', 'the'), 10743389),\n",
       " (('http', 'www'), 7115634),\n",
       " (('in', 'the'), 6708376),\n",
       " (('ref', 'name'), 6162263),\n",
       " (('url', 'http'), 4826060),\n",
       " (('ref', 'cite'), 3982254),\n",
       " (('he', 'was'), 3527017),\n",
       " (('cite', 'web'), 3441055),\n",
       " (('at', 'the'), 3024121),\n",
       " (('to', 'the'), 2744050),\n",
       " (('for', 'the'), 2462184),\n",
       " (('web', 'url'), 2362973),\n",
       " (('and', 'the'), 1935295),\n",
       " (('as', 'a'), 1894177),\n",
       " (('on', 'the'), 1864609),\n",
       " (('united', 'states'), 1833942),\n",
       " (('new', 'york'), 1697296),\n",
       " (('cite', 'news'), 1685356),\n",
       " (('birth', 'date'), 1576543),\n",
       " (('align', 'center'), 1555454),\n",
       " (('ref', 'http'), 1532831),\n",
       " (('players', 'category'), 1523481),\n",
       " (('with', 'the'), 1445508),\n",
       " (('f', 'c'), 1414783),\n",
       " (('was', 'a'), 1383429),\n",
       " (('births', 'category'), 1249080),\n",
       " (('university', 'of'), 1191819),\n",
       " (('of', 'his'), 1189126),\n",
       " (('from', 'the'), 1167930),\n",
       " (('in', 'a'), 1155019),\n",
       " (('by', 'the'), 1128407),\n",
       " (('ref', 'ref'), 1123085),\n",
       " (('style', 'background'), 1054393),\n",
       " (('url', 'https'), 1048855),\n",
       " (('ref', 'he'), 1043445),\n",
       " (('the', 'first'), 991601),\n",
       " (('the', 'united'), 971037),\n",
       " (('co', 'uk'), 955335),\n",
       " (('references', 'reflist'), 938507),\n",
       " (('death', 'date'), 923442),\n",
       " (('as', 'the'), 913630),\n",
       " (('cite', 'book'), 909289),\n",
       " (('archive', 'org'), 902547),\n",
       " (('category', 'people'), 893387),\n",
       " (('external', 'links'), 865936),\n",
       " (('birth', 'place'), 843330),\n",
       " (('ref', 'in'), 841856),\n",
       " (('0', '0'), 826471),\n",
       " (('br', 'small'), 808533),\n",
       " (('people', 'category'), 807615),\n",
       " (('was', 'the'), 805645),\n",
       " (('web', 'archive'), 797525),\n",
       " (('org', 'web'), 797152),\n",
       " (('align', 'left'), 794831),\n",
       " (('is', 'a'), 794489),\n",
       " (('https', 'web'), 780078),\n",
       " (('in', 'his'), 774684),\n",
       " (('member', 'of'), 770476),\n",
       " (('of', 'a'), 769138),\n",
       " (('category', 'american'), 768676),\n",
       " (('one', 'of'), 763881),\n",
       " (('and', 'age'), 761586),\n",
       " (('living', 'people'), 760814),\n",
       " (('people', 'from'), 755709),\n",
       " (('and', 'was'), 752151),\n",
       " (('news', 'url'), 751990),\n",
       " (('date', 'and'), 741749),\n",
       " (('html', 'title'), 732643),\n",
       " (('date', 'birth'), 730903),\n",
       " (('the', 'new'), 712533)]"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "male_bigrams.most_common(70)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load the unigram counters\n",
    "\n",
    "with open('data/female_counter.pkl', 'rb') as infile:\n",
    "    female_counter = pickle.load(infile)\n",
    "with open('data/male_counter.pkl', 'rb') as infile:\n",
    "    male_counter = pickle.load(infile)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "class PMICalculator:\n",
    "    \n",
    "    def __init__(self, counter1, counter2, n1, n2):\n",
    "        self.counter1 = counter1\n",
    "        self.counter2 = counter2\n",
    "        self.n1 = n1\n",
    "        self.n2 = n2\n",
    "        self.lpc1 = log(n1 / (n1 + n2))\n",
    "\n",
    "    def pmi(self, word):\n",
    "        try:\n",
    "            return (log(self.counter1[word]) - \n",
    "            self.lpc1 -\n",
    "            log(self.counter1[word] + self.counter2[word]))\n",
    "        except ValueError:\n",
    "            return 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "pc1 = PMICalculator(female_counter, male_counter, 130576, 718413)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-1.9380632450542024"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc1.pmi(\"his\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.7940272570088585"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc1.pmi(\"she\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.509566866723068"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc1.pmi(\"husband\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-0.057060278138187925"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc1.pmi(\"wife\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-0.12356478572158913"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc1.pmi(\"career\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.28344799212810123"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc1.pmi(\"family\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.20763811546007105"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc1.pmi(\"received\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.25115538983372687"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc1.pmi(\"felt\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-0.04150079746606927"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc1.pmi(\"thought\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-0.17359434944100727"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc1.pmi(\"feared\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.4251156630049806"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc1.pmi(\"married\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.49427551126943214"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc1.pmi(\"looks\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "all_unigrams = female_counter + male_counter\n",
    "top_unigrams = all_unigrams.most_common(20000)\n",
    "top_unigrams_with_pmi = [(p[0], p[1], pc1.pmi(p[0])) for p in top_unigrams]\n",
    "top_unigrams_with_pmi.sort(key = lambda x: x[-1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Men\n",
    "men_top_pmi = top_unigrams_with_pmi[:100]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Women\n",
    "women_top_pmi = top_unigrams_with_pmi[-100:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "# men_top_pmi"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('olga', 9979, 1.4244624807668984),\n",
       " ('karin', 3462, 1.4268502292553649),\n",
       " ('oricon', 6592, 1.4289086990300355),\n",
       " ('cindy', 5564, 1.4294589397572484),\n",
       " ('cheryl', 5233, 1.429642104677562),\n",
       " ('vogue', 14719, 1.4299182901017922),\n",
       " ('carla', 5561, 1.430278025021753),\n",
       " ('nurses', 4627, 1.4321053876736585),\n",
       " ('gymnast', 5323, 1.4326140040684354),\n",
       " ('angie', 2808, 1.4357037786344566),\n",
       " ('nicole', 12228, 1.4358578637140553),\n",
       " ('gymnastics', 27011, 1.4361599930835922),\n",
       " ('ballerina', 2369, 1.4394553364610196),\n",
       " ('melissa', 9778, 1.4416696954133652),\n",
       " ('katie', 9254, 1.4426203806579831),\n",
       " ('adriana', 2478, 1.4427348130173288),\n",
       " ('vanessa', 7303, 1.4432125532717706),\n",
       " ('girls', 57627, 1.443666437110494),\n",
       " ('tanya', 3190, 1.4439568891603312),\n",
       " ('carly', 2751, 1.4440165097617674),\n",
       " ('sopranos', 4737, 1.4471206741306606),\n",
       " ('suffrage', 6558, 1.447225542427251),\n",
       " ('abby', 3365, 1.448028872285004),\n",
       " ('bianca', 3231, 1.45024338952501),\n",
       " ('erika', 3397, 1.4516480233531563),\n",
       " ('jessica', 17538, 1.4518054969917902),\n",
       " ('monique', 2639, 1.4538502268189877),\n",
       " ('justine', 2306, 1.4539700542077787),\n",
       " ('michelle', 17688, 1.4548538222469443),\n",
       " ('tatiana', 3955, 1.4553103308129565),\n",
       " ('chloe', 2841, 1.4560098780253874),\n",
       " ('hye', 2421, 1.4572885529771042),\n",
       " ('megan', 5878, 1.4577397795135614),\n",
       " ('avn', 10121, 1.4594365854005744),\n",
       " ('chanel', 2576, 1.46000443790859),\n",
       " ('nadine', 2535, 1.4600955381334657),\n",
       " ('natalia', 5096, 1.4606226202147763),\n",
       " ('nikki', 5251, 1.4635739986398217),\n",
       " ('natasha', 4623, 1.4675986094324713),\n",
       " ('becky', 3007, 1.4692826151458007),\n",
       " ('melanie', 6026, 1.4695255625823087),\n",
       " ('martina', 4776, 1.4719506819786332),\n",
       " ('lesbian', 11601, 1.4728128703503014),\n",
       " ('nathalie', 2342, 1.4759755233413694),\n",
       " ('danielle', 5043, 1.4778650856087356),\n",
       " ('elle', 6924, 1.478470341287215),\n",
       " ('feminine', 2583, 1.480466740384914),\n",
       " ('kristina', 2815, 1.483708080233165),\n",
       " ('miss', 83631, 1.4843674303589847),\n",
       " ('serena', 4587, 1.485414709474675),\n",
       " ('gabriela', 2511, 1.4884877118158162),\n",
       " ('irina', 4768, 1.4900488761150061),\n",
       " ('sexiest', 3491, 1.4915208926340338),\n",
       " ('ellie', 2464, 1.4944413892355977),\n",
       " ('glamour', 5259, 1.4953035046743786),\n",
       " ('jenna', 2629, 1.505422050171231),\n",
       " ('husband', 75006, 1.509566866723068),\n",
       " ('dames', 2913, 1.5134031379984698),\n",
       " ('selena', 2387, 1.5172694217917977),\n",
       " ('samantha', 7174, 1.5255597765556708),\n",
       " ('daniela', 2709, 1.5402589400164963),\n",
       " ('ussoccer', 2946, 1.5463550152588699),\n",
       " ('ladies', 25061, 1.5567496105259568),\n",
       " ('diva', 5421, 1.565351069726276),\n",
       " ('hostess', 2330, 1.565635610121502),\n",
       " ('heroine', 4026, 1.5805928759512007),\n",
       " ('synchronized', 2695, 1.589220875318496),\n",
       " ('rhythmic', 6578, 1.5907733468114795),\n",
       " ('svetlana', 2508, 1.592878865263411),\n",
       " ('iafd', 2944, 1.5934259182796167),\n",
       " ('divas', 3064, 1.6468235824678423),\n",
       " ('softball', 6747, 1.663732906276806),\n",
       " ('consorts', 4180, 1.6646722397229912),\n",
       " ('heptathlon', 2473, 1.6810781243774144),\n",
       " ('uneven', 2870, 1.684879254993061),\n",
       " ('actress', 147860, 1.7006927238502563),\n",
       " ('princesses', 4711, 1.700704771164233),\n",
       " ('feminists', 4716, 1.7009029585357087),\n",
       " ('feminism', 6297, 1.7079919038089475),\n",
       " ('her', 1308263, 1.7244667675788659),\n",
       " ('pageant', 12683, 1.7338061198679977),\n",
       " ('women', 532471, 1.742348800649344),\n",
       " ('female', 123970, 1.7434816438590204),\n",
       " ('feminist', 15422, 1.7511335550504707),\n",
       " ('herself', 41532, 1.75370262315033),\n",
       " ('f0f8ff', 10135, 1.759633956570056),\n",
       " ('lightblue', 7937, 1.7710188060825605),\n",
       " ('womens', 7542, 1.7808032765943356),\n",
       " ('titleholder', 2282, 1.7851590918440001),\n",
       " ('fhm', 2824, 1.791752956152866),\n",
       " ('wnba', 11362, 1.7925247111220468),\n",
       " ('she', 1412916, 1.7940272570088585),\n",
       " ('camogie', 3807, 1.8108914106971437),\n",
       " ('netball', 4354, 1.8132015771732455),\n",
       " ('aagpbl', 3599, 1.8350128269026094),\n",
       " ('actresses', 93472, 1.8524058451615915),\n",
       " ('lpga', 11572, 1.8595688943358457),\n",
       " ('wta', 15923, 1.864272839097632),\n",
       " ('fbw', 3308, 1.8663305871266829),\n",
       " ('addfad', 3096, 1.8707979720789023)]"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "women_top_pmi"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "pc2 = PMICalculator(female_bigrams, male_bigrams, 130576, 718413)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "all_bigrams = female_bigrams + male_bigrams"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(('of', 'aberdeen'), 8253),\n",
       " (('46', 'ref'), 8252),\n",
       " (('the', 'eagles'), 8252),\n",
       " (('pcupdate', 'ntupdate'), 8252),\n",
       " (('houston', 'texans'), 8252)]"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "top_bigrams = all_bigrams.most_common(20000)\n",
    "top_bigrams[-5:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "top_bigrams_with_pmi = [(p[0], p[1], pc2.pmi(p[0])) for p in top_bigrams]\n",
    "top_bigrams_with_pmi.sort(key = lambda x: x[-1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Men\n",
    "top_bigrams_with_pmi[:50]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(('american', 'women'), 25328, 1.814727984961209),\n",
       " (('her', 'work'), 22365, 1.818924876053325),\n",
       " (('her', 'debut'), 10883, 1.8210054345701217),\n",
       " (('her', 'career'), 21648, 1.8261513329865693),\n",
       " (('she', 'has'), 70357, 1.8265834685324158),\n",
       " (('supporting', 'actress'), 12378, 1.8268087717543349),\n",
       " (('she', 'then'), 10979, 1.8288141422258697),\n",
       " (('she', 'received'), 17720, 1.8361243766480992),\n",
       " (('she', 'served'), 10180, 1.8412669513945836),\n",
       " (('category', 'women'), 29341, 1.8449019731513019),\n",
       " (('she', 'began'), 15514, 1.8513143813812842),\n",
       " (('she', 'won'), 28102, 1.8550074071748277),\n",
       " (('she', 'joined'), 9000, 1.8569771592711604),\n",
       " (('year', 'she'), 12683, 1.8577169145055379),\n",
       " (('she', 'attended'), 11307, 1.8598108776867885),\n",
       " (('she', 'appeared'), 17848, 1.860425078265246),\n",
       " (('she', 'played'), 19948, 1.8630263780836405),\n",
       " (('century', 'women'), 20836, 1.8657837541957925),\n",
       " (('began', 'her'), 8704, 1.865983028165859),\n",
       " (('she', 'studied'), 9917, 1.8669348436887674),\n",
       " (('female', 'singers'), 8999, 1.8684169827626445),\n",
       " (('she', 'graduated'), 9172, 1.8691427139283778),\n",
       " (('american', 'female'), 11263, 1.8692455919965472),\n",
       " (('women', 'writers'), 19960, 1.8695826442739953),\n",
       " (('olympics', 'women'), 16566, 1.86967329318413),\n",
       " (('actresses', 'category'), 61005, 1.870745740779741),\n",
       " (('championships', 'women'), 17658, 1.8711842814383939),\n",
       " (('category', 'actresses'), 18433, 1.8720907970812082),\n",
       " (('film', 'actresses'), 17840, 1.8720907970812082),\n",
       " (('open', 'women'), 15707, 1.8720907970812082),\n",
       " (('actresses', 'from'), 15092, 1.8720907970812082),\n",
       " (('television', 'actresses'), 13199, 1.8720907970812082),\n",
       " (('occupation', 'actress'), 12172, 1.8720907970812082),\n",
       " (('american', 'actresses'), 10033, 1.8720907970812082),\n",
       " (('category', 'female'), 9876, 1.8720907970812082)]"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Women\n",
    "top_bigrams_with_pmi[-35:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.6782626536456213"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc2.pmi((\"first\", \"woman\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-2.248461007969188"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc2.pmi((\"first\", \"man\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-0.5448207087013106"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc2.pmi((\"first\", \"person\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.677569700476507"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc2.pmi((\"her\", \"mother\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-2.3741492182690216"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc2.pmi((\"his\", \"mother\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-0.20183556391051916"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc2.pmi((\"husband\", \"of\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.46124905152146845"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc2.pmi((\"wife\", \"of\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-1.089677112699471"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc2.pmi((\"leader\", \"of\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-0.3949652180862451"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc2.pmi((\"known\", \"for\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-0.18826044315597024"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc2.pmi((\"recognized\", \"for\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-0.8254634190213821"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc2.pmi((\"first\", \"of\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-0.7645317867478543"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc2.pmi((\"the\", \"most\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-0.45633540156981844"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc2.pmi((\"the\", \"best\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-0.8233678688462245"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc2.pmi((\"the\", \"second\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-0.6509658253578845"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pc2.pmi((\"the\", \"first\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "912719349"
      ]
     },
     "execution_count": 97,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Number of words we are dealing with in our corpus: \n",
    "sum(female_counter.values()) + sum(male_counter.values())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# This makes the kernel die - don't run it\n",
    "\n",
    "# female = female_bigrams + female_counter\n",
    "# male = male_bigrams + male_counter\n",
    "\n",
    "# pc3 = PMICalculator(female, male, 130576, 718413)\n",
    "# all_counters = female + male\n",
    "# top_tokens = all_counters.most_common(10000)\n",
    "# top_tokens_with_pmi = [(p[0], p[1], pc3.pmi(p[0])) for p in top_tokens]\n",
    "# top_tokens_with_pmi.sort(key = lambda x: x[-1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Comparing Bigrams (LLR)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "147838368"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Number of words in corpus of women's bios: \n",
    "n_women = sum(female_counter.values())\n",
    "n_women"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "764880981"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Number of words in corpus of men's bios: \n",
    "n_men = sum(male_counter.values())\n",
    "n_men"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# bigrams beginning with he or she, filtered by words appearing at least 90 times.\n",
    "\n",
    "she = collections.Counter({ k[1]: v for k, v in female_bigrams.items() if v >= 90 and k[0] == \"she\" })\n",
    "he = collections.Counter({ k[1]: v for k, v in male_bigrams.items() if v >= 90 and k[0] == \"he\" })"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "615"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(she)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1252344"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sum(she.values())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1236"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(he)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "14487635"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sum(he.values())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of times 'he' appears in male bios: 7053558\n",
      "Number of times 'she' appears in female bios: 1306814\n"
     ]
    }
   ],
   "source": [
    "he_count = male_counter['he']\n",
    "she_count = female_counter['she']\n",
    "print(\"Number of times 'he' appears in male bios:\", he_count )\n",
    "print(\"Number of times 'she' appears in female bios:\", she_count)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Look up counts for all words in \"he\" counter not in \"she\" counter and vice versa\n",
    "for word in he:\n",
    "    if word not in she:\n",
    "        she[word] = female_bigrams[(\"she\", word)]\n",
    "for word in she:\n",
    "    if word not in he:\n",
    "        he[word] = male_bigrams[(\"he\", word)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "columns = ['Word following he/she', \"she+word count\"]\n",
    "df = pd.DataFrame([(k,v) for k,v in she.items()], columns=columns)\n",
    "df.set_index(['Word following he/she'], inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "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>she+word count</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Word following he/she</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>is</th>\n",
       "      <td>72233</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>called</th>\n",
       "      <td>1325</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>moved</th>\n",
       "      <td>9979</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>had</th>\n",
       "      <td>44820</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>achieved</th>\n",
       "      <td>1092</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                       she+word count\n",
       "Word following he/she                \n",
       "is                              72233\n",
       "called                           1325\n",
       "moved                            9979\n",
       "had                             44820\n",
       "achieved                         1092"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "columns2 = ['Word following he/she', \"he+word count\",]\n",
    "df2 = pd.DataFrame([(k,v) for k,v in he.items()], columns=columns2)\n",
    "df2.set_index(['Word following he/she'], inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.concat([df,df2], axis=1)\n",
    "\n",
    "df['total count'] = df.fillna(0)[\"she+word count\"] + df.fillna(0)[\"he+word count\"]\n",
    "\n",
    "df.sort_values(by=['total count'], ascending=False, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Pointwise Mutual Information, where $w_1$ is either \"he\" or \"she\", and `N` is the total number of words in the corpus\n",
    "\n",
    "$$\\text{PMI}(w_1 w_2) = \\log_2\\frac{p(w_1 w_2)}{p(w_1)\\ p(w_2)} = \\log_2\\frac{\\frac{C(w_1 w_2)}{N}}{\\frac{C(w_1)}{N} \\cdot \\frac{C(w_2)}{N}} = \\log_2(N * \\frac{C(w_1 w_2)}{C(w_1)\\ C(w_2)} )$$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "# count of x, count of y, count of xy\n",
    "def pmi(cx, cy, cxy, n):\n",
    "    if cy == 0:\n",
    "        return 0\n",
    "    if cxy == 0:\n",
    "        return -100\n",
    "    else:\n",
    "        return np.log2(n) + np.log2(cxy) - (np.log2(cx) + np.log2(cy))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "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>she+word count</th>\n",
       "      <th>he+word count</th>\n",
       "      <th>total count</th>\n",
       "      <th>PMI she+word</th>\n",
       "      <th>PMI he+word</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>was</th>\n",
       "      <td>251644</td>\n",
       "      <td>3527017</td>\n",
       "      <td>3778661</td>\n",
       "      <td>4.865221</td>\n",
       "      <td>5.984816</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>is</th>\n",
       "      <td>72233</td>\n",
       "      <td>581205</td>\n",
       "      <td>653438</td>\n",
       "      <td>4.234702</td>\n",
       "      <td>5.079859</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>had</th>\n",
       "      <td>44820</td>\n",
       "      <td>532181</td>\n",
       "      <td>577001</td>\n",
       "      <td>4.750093</td>\n",
       "      <td>5.740472</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>also</th>\n",
       "      <td>51109</td>\n",
       "      <td>516457</td>\n",
       "      <td>567566</td>\n",
       "      <td>4.753585</td>\n",
       "      <td>5.685964</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>has</th>\n",
       "      <td>67227</td>\n",
       "      <td>427506</td>\n",
       "      <td>494733</td>\n",
       "      <td>5.318918</td>\n",
       "      <td>6.043105</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>became</th>\n",
       "      <td>24076</td>\n",
       "      <td>321747</td>\n",
       "      <td>345823</td>\n",
       "      <td>5.027487</td>\n",
       "      <td>6.156175</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>played</th>\n",
       "      <td>19768</td>\n",
       "      <td>305679</td>\n",
       "      <td>325447</td>\n",
       "      <td>5.346162</td>\n",
       "      <td>6.008191</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>served</th>\n",
       "      <td>9871</td>\n",
       "      <td>248302</td>\n",
       "      <td>258173</td>\n",
       "      <td>5.054372</td>\n",
       "      <td>6.313847</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>made</th>\n",
       "      <td>14427</td>\n",
       "      <td>205207</td>\n",
       "      <td>219634</td>\n",
       "      <td>4.503840</td>\n",
       "      <td>5.617981</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>died</th>\n",
       "      <td>11680</td>\n",
       "      <td>200767</td>\n",
       "      <td>212447</td>\n",
       "      <td>4.594797</td>\n",
       "      <td>5.962816</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        she+word count  he+word count  total count  PMI she+word  PMI he+word\n",
       "was             251644        3527017      3778661      4.865221     5.984816\n",
       "is               72233         581205       653438      4.234702     5.079859\n",
       "had              44820         532181       577001      4.750093     5.740472\n",
       "also             51109         516457       567566      4.753585     5.685964\n",
       "has              67227         427506       494733      5.318918     6.043105\n",
       "became           24076         321747       345823      5.027487     6.156175\n",
       "played           19768         305679       325447      5.346162     6.008191\n",
       "served            9871         248302       258173      5.054372     6.313847\n",
       "made             14427         205207       219634      4.503840     5.617981\n",
       "died             11680         200767       212447      4.594797     5.962816"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['PMI she+word'] = df.apply(lambda row: pmi(she_count, female_counter[row.name], \n",
    "                                              row[\"she+word count\"], n_women), axis=1)\n",
    "df['PMI he+word'] = df.apply(lambda row: pmi(he_count, male_counter[row.name], \n",
    "                                             row[\"he+word count\"], n_men), axis=1)\n",
    "df.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Log probabilities of word given he/she. Add 1 for smoothing\n",
    "\n",
    "df['log prob of word | he'] = np.log2(\n",
    "    (df.fillna(0)[\"he+word count\"] + 1)) - np.log2((sum(he.values()) + 1))\n",
    "df['log prob of word | she'] = np.log2(\n",
    "    (df.fillna(0)[\"she+word count\"] + 1)) - np.log2((sum(she.values()) + 1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "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>she+word count</th>\n",
       "      <th>he+word count</th>\n",
       "      <th>total count</th>\n",
       "      <th>PMI she+word</th>\n",
       "      <th>PMI he+word</th>\n",
       "      <th>log prob of word | he</th>\n",
       "      <th>log prob of word | she</th>\n",
       "      <th>LLR</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>herself</th>\n",
       "      <td>744</td>\n",
       "      <td>0</td>\n",
       "      <td>744</td>\n",
       "      <td>1.189844</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>-23.788790</td>\n",
       "      <td>-10.746928</td>\n",
       "      <td>13.041862</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>peaked</th>\n",
       "      <td>242</td>\n",
       "      <td>0</td>\n",
       "      <td>242</td>\n",
       "      <td>2.912550</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>-23.788790</td>\n",
       "      <td>-12.363212</td>\n",
       "      <td>11.425578</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>her</th>\n",
       "      <td>203</td>\n",
       "      <td>0</td>\n",
       "      <td>203</td>\n",
       "      <td>-5.619086</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>-23.788790</td>\n",
       "      <td>-12.615599</td>\n",
       "      <td>11.173190</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>kicks</th>\n",
       "      <td>159</td>\n",
       "      <td>0</td>\n",
       "      <td>159</td>\n",
       "      <td>4.282957</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>-23.788790</td>\n",
       "      <td>-12.966096</td>\n",
       "      <td>10.822693</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>stoops</th>\n",
       "      <td>95</td>\n",
       "      <td>0</td>\n",
       "      <td>95</td>\n",
       "      <td>6.521314</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>-23.788790</td>\n",
       "      <td>-13.703062</td>\n",
       "      <td>10.085728</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>modeled</th>\n",
       "      <td>182</td>\n",
       "      <td>112</td>\n",
       "      <td>294</td>\n",
       "      <td>4.340933</td>\n",
       "      <td>2.627048</td>\n",
       "      <td>-16.968611</td>\n",
       "      <td>-12.772325</td>\n",
       "      <td>4.196286</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>contributes</th>\n",
       "      <td>95</td>\n",
       "      <td>60</td>\n",
       "      <td>155</td>\n",
       "      <td>4.352760</td>\n",
       "      <td>1.862693</td>\n",
       "      <td>-17.858052</td>\n",
       "      <td>-13.703062</td>\n",
       "      <td>4.154990</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>posed</th>\n",
       "      <td>354</td>\n",
       "      <td>249</td>\n",
       "      <td>603</td>\n",
       "      <td>4.876859</td>\n",
       "      <td>3.254666</td>\n",
       "      <td>-15.823005</td>\n",
       "      <td>-11.816349</td>\n",
       "      <td>4.006656</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>danced</th>\n",
       "      <td>513</td>\n",
       "      <td>458</td>\n",
       "      <td>971</td>\n",
       "      <td>5.207171</td>\n",
       "      <td>4.966041</td>\n",
       "      <td>-14.946439</td>\n",
       "      <td>-11.282400</td>\n",
       "      <td>3.664039</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>trains</th>\n",
       "      <td>88</td>\n",
       "      <td>91</td>\n",
       "      <td>179</td>\n",
       "      <td>3.745979</td>\n",
       "      <td>1.334844</td>\n",
       "      <td>-17.265228</td>\n",
       "      <td>-13.812291</td>\n",
       "      <td>3.452937</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>participates</th>\n",
       "      <td>105</td>\n",
       "      <td>109</td>\n",
       "      <td>214</td>\n",
       "      <td>4.846071</td>\n",
       "      <td>3.339100</td>\n",
       "      <td>-17.007430</td>\n",
       "      <td>-13.560104</td>\n",
       "      <td>3.447326</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>studies</th>\n",
       "      <td>109</td>\n",
       "      <td>117</td>\n",
       "      <td>226</td>\n",
       "      <td>-1.103480</td>\n",
       "      <td>-3.370457</td>\n",
       "      <td>-16.906147</td>\n",
       "      <td>-13.506665</td>\n",
       "      <td>3.399482</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>moves</th>\n",
       "      <td>80</td>\n",
       "      <td>101</td>\n",
       "      <td>181</td>\n",
       "      <td>1.763290</td>\n",
       "      <td>-0.434748</td>\n",
       "      <td>-17.116364</td>\n",
       "      <td>-13.948175</td>\n",
       "      <td>3.168190</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>stars</th>\n",
       "      <td>156</td>\n",
       "      <td>203</td>\n",
       "      <td>359</td>\n",
       "      <td>-0.393679</td>\n",
       "      <td>-1.824888</td>\n",
       "      <td>-16.116364</td>\n",
       "      <td>-12.993404</td>\n",
       "      <td>3.122961</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dated</th>\n",
       "      <td>227</td>\n",
       "      <td>303</td>\n",
       "      <td>530</td>\n",
       "      <td>3.088039</td>\n",
       "      <td>1.189732</td>\n",
       "      <td>-15.540862</td>\n",
       "      <td>-12.455135</td>\n",
       "      <td>3.085728</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>interned</th>\n",
       "      <td>90</td>\n",
       "      <td>125</td>\n",
       "      <td>215</td>\n",
       "      <td>4.620189</td>\n",
       "      <td>2.404596</td>\n",
       "      <td>-16.811510</td>\n",
       "      <td>-13.780230</td>\n",
       "      <td>3.031280</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>skated</th>\n",
       "      <td>94</td>\n",
       "      <td>132</td>\n",
       "      <td>226</td>\n",
       "      <td>4.821823</td>\n",
       "      <td>4.118634</td>\n",
       "      <td>-16.733507</td>\n",
       "      <td>-13.718169</td>\n",
       "      <td>3.015338</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>upset</th>\n",
       "      <td>129</td>\n",
       "      <td>182</td>\n",
       "      <td>311</td>\n",
       "      <td>2.981301</td>\n",
       "      <td>1.234424</td>\n",
       "      <td>-16.273090</td>\n",
       "      <td>-13.265657</td>\n",
       "      <td>3.007433</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>explores</th>\n",
       "      <td>123</td>\n",
       "      <td>175</td>\n",
       "      <td>298</td>\n",
       "      <td>3.112389</td>\n",
       "      <td>2.479784</td>\n",
       "      <td>-16.329358</td>\n",
       "      <td>-13.333828</td>\n",
       "      <td>2.995530</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>begged</th>\n",
       "      <td>61</td>\n",
       "      <td>90</td>\n",
       "      <td>151</td>\n",
       "      <td>4.686471</td>\n",
       "      <td>3.486064</td>\n",
       "      <td>-17.280995</td>\n",
       "      <td>-14.333828</td>\n",
       "      <td>2.947167</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>acts</th>\n",
       "      <td>75</td>\n",
       "      <td>113</td>\n",
       "      <td>188</td>\n",
       "      <td>-0.347033</td>\n",
       "      <td>-1.888532</td>\n",
       "      <td>-16.955900</td>\n",
       "      <td>-14.040097</td>\n",
       "      <td>2.915803</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>attends</th>\n",
       "      <td>82</td>\n",
       "      <td>128</td>\n",
       "      <td>210</td>\n",
       "      <td>3.808688</td>\n",
       "      <td>3.181424</td>\n",
       "      <td>-16.777562</td>\n",
       "      <td>-13.912985</td>\n",
       "      <td>2.864577</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>choreographed</th>\n",
       "      <td>127</td>\n",
       "      <td>198</td>\n",
       "      <td>325</td>\n",
       "      <td>3.728359</td>\n",
       "      <td>3.613664</td>\n",
       "      <td>-16.152165</td>\n",
       "      <td>-13.288025</td>\n",
       "      <td>2.864141</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>chairs</th>\n",
       "      <td>62</td>\n",
       "      <td>97</td>\n",
       "      <td>159</td>\n",
       "      <td>2.511577</td>\n",
       "      <td>0.818589</td>\n",
       "      <td>-17.174080</td>\n",
       "      <td>-14.310745</td>\n",
       "      <td>2.863335</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>turns</th>\n",
       "      <td>55</td>\n",
       "      <td>90</td>\n",
       "      <td>145</td>\n",
       "      <td>1.050514</td>\n",
       "      <td>0.151767</td>\n",
       "      <td>-17.280995</td>\n",
       "      <td>-14.480670</td>\n",
       "      <td>2.800325</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               she+word count  he+word count  total count  PMI she+word  \\\n",
       "herself                   744              0          744      1.189844   \n",
       "peaked                    242              0          242      2.912550   \n",
       "her                       203              0          203     -5.619086   \n",
       "kicks                     159              0          159      4.282957   \n",
       "stoops                     95              0           95      6.521314   \n",
       "modeled                   182            112          294      4.340933   \n",
       "contributes                95             60          155      4.352760   \n",
       "posed                     354            249          603      4.876859   \n",
       "danced                    513            458          971      5.207171   \n",
       "trains                     88             91          179      3.745979   \n",
       "participates              105            109          214      4.846071   \n",
       "studies                   109            117          226     -1.103480   \n",
       "moves                      80            101          181      1.763290   \n",
       "stars                     156            203          359     -0.393679   \n",
       "dated                     227            303          530      3.088039   \n",
       "interned                   90            125          215      4.620189   \n",
       "skated                     94            132          226      4.821823   \n",
       "upset                     129            182          311      2.981301   \n",
       "explores                  123            175          298      3.112389   \n",
       "begged                     61             90          151      4.686471   \n",
       "acts                       75            113          188     -0.347033   \n",
       "attends                    82            128          210      3.808688   \n",
       "choreographed             127            198          325      3.728359   \n",
       "chairs                     62             97          159      2.511577   \n",
       "turns                      55             90          145      1.050514   \n",
       "\n",
       "               PMI he+word  log prob of word | he  log prob of word | she  \\\n",
       "herself        -100.000000             -23.788790              -10.746928   \n",
       "peaked         -100.000000             -23.788790              -12.363212   \n",
       "her            -100.000000             -23.788790              -12.615599   \n",
       "kicks          -100.000000             -23.788790              -12.966096   \n",
       "stoops         -100.000000             -23.788790              -13.703062   \n",
       "modeled           2.627048             -16.968611              -12.772325   \n",
       "contributes       1.862693             -17.858052              -13.703062   \n",
       "posed             3.254666             -15.823005              -11.816349   \n",
       "danced            4.966041             -14.946439              -11.282400   \n",
       "trains            1.334844             -17.265228              -13.812291   \n",
       "participates      3.339100             -17.007430              -13.560104   \n",
       "studies          -3.370457             -16.906147              -13.506665   \n",
       "moves            -0.434748             -17.116364              -13.948175   \n",
       "stars            -1.824888             -16.116364              -12.993404   \n",
       "dated             1.189732             -15.540862              -12.455135   \n",
       "interned          2.404596             -16.811510              -13.780230   \n",
       "skated            4.118634             -16.733507              -13.718169   \n",
       "upset             1.234424             -16.273090              -13.265657   \n",
       "explores          2.479784             -16.329358              -13.333828   \n",
       "begged            3.486064             -17.280995              -14.333828   \n",
       "acts             -1.888532             -16.955900              -14.040097   \n",
       "attends           3.181424             -16.777562              -13.912985   \n",
       "choreographed     3.613664             -16.152165              -13.288025   \n",
       "chairs            0.818589             -17.174080              -14.310745   \n",
       "turns             0.151767             -17.280995              -14.480670   \n",
       "\n",
       "                     LLR  \n",
       "herself        13.041862  \n",
       "peaked         11.425578  \n",
       "her            11.173190  \n",
       "kicks          10.822693  \n",
       "stoops         10.085728  \n",
       "modeled         4.196286  \n",
       "contributes     4.154990  \n",
       "posed           4.006656  \n",
       "danced          3.664039  \n",
       "trains          3.452937  \n",
       "participates    3.447326  \n",
       "studies         3.399482  \n",
       "moves           3.168190  \n",
       "stars           3.122961  \n",
       "dated           3.085728  \n",
       "interned        3.031280  \n",
       "skated          3.015338  \n",
       "upset           3.007433  \n",
       "explores        2.995530  \n",
       "begged          2.947167  \n",
       "acts            2.915803  \n",
       "attends         2.864577  \n",
       "choreographed   2.864141  \n",
       "chairs          2.863335  \n",
       "turns           2.800325  "
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Log likelihood ratio she : he\n",
    "\n",
    "df['LLR'] = df['log prob of word | she'] - df['log prob of word | he']\n",
    "df.sort_values(by=['LLR'], ascending=False, inplace=True)\n",
    "df.head(25)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>she+word count</th>\n",
       "      <th>he+word count</th>\n",
       "      <th>total count</th>\n",
       "      <th>PMI she+word</th>\n",
       "      <th>PMI he+word</th>\n",
       "      <th>log prob of word | he</th>\n",
       "      <th>log prob of word | she</th>\n",
       "      <th>LLR</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>rushed</th>\n",
       "      <td>17</td>\n",
       "      <td>3238</td>\n",
       "      <td>3255</td>\n",
       "      <td>1.984473</td>\n",
       "      <td>5.710606</td>\n",
       "      <td>-12.127457</td>\n",
       "      <td>-16.118100</td>\n",
       "      <td>-3.990643</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>pronounced</th>\n",
       "      <td>0</td>\n",
       "      <td>180</td>\n",
       "      <td>180</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>1.929257</td>\n",
       "      <td>-16.288944</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-3.999081</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dispatched</th>\n",
       "      <td>0</td>\n",
       "      <td>181</td>\n",
       "      <td>181</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>2.766231</td>\n",
       "      <td>-16.280995</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-4.007029</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>captain</th>\n",
       "      <td>0</td>\n",
       "      <td>182</td>\n",
       "      <td>182</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>-2.714158</td>\n",
       "      <td>-16.273090</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-4.014935</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>calculated</th>\n",
       "      <td>0</td>\n",
       "      <td>191</td>\n",
       "      <td>191</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>3.397856</td>\n",
       "      <td>-16.203827</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-4.084197</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>downed</th>\n",
       "      <td>0</td>\n",
       "      <td>194</td>\n",
       "      <td>194</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>4.191981</td>\n",
       "      <td>-16.181459</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-4.106565</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>invaded</th>\n",
       "      <td>0</td>\n",
       "      <td>219</td>\n",
       "      <td>219</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>2.363412</td>\n",
       "      <td>-16.007430</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-4.280595</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>farmed</th>\n",
       "      <td>0</td>\n",
       "      <td>221</td>\n",
       "      <td>221</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>4.412934</td>\n",
       "      <td>-15.994374</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-4.293651</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>gallantly</th>\n",
       "      <td>0</td>\n",
       "      <td>229</td>\n",
       "      <td>229</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>5.751321</td>\n",
       "      <td>-15.943300</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-4.344725</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>trans</th>\n",
       "      <td>0</td>\n",
       "      <td>258</td>\n",
       "      <td>258</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>-0.914177</td>\n",
       "      <td>-15.771981</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-4.516043</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>bats</th>\n",
       "      <td>0</td>\n",
       "      <td>271</td>\n",
       "      <td>271</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>0.522018</td>\n",
       "      <td>-15.701327</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-4.586698</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>commanded</th>\n",
       "      <td>50</td>\n",
       "      <td>14651</td>\n",
       "      <td>14701</td>\n",
       "      <td>3.969824</td>\n",
       "      <td>6.257655</td>\n",
       "      <td>-9.949980</td>\n",
       "      <td>-14.615599</td>\n",
       "      <td>-4.665619</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>111</th>\n",
       "      <td>0</td>\n",
       "      <td>296</td>\n",
       "      <td>296</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>1.039456</td>\n",
       "      <td>-15.574471</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-4.713554</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>man</th>\n",
       "      <td>0</td>\n",
       "      <td>370</td>\n",
       "      <td>370</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>-2.796346</td>\n",
       "      <td>-15.253514</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-5.034510</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>engraved</th>\n",
       "      <td>0</td>\n",
       "      <td>382</td>\n",
       "      <td>382</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>3.237178</td>\n",
       "      <td>-15.207589</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-5.080435</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>erected</th>\n",
       "      <td>0</td>\n",
       "      <td>450</td>\n",
       "      <td>450</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>2.622690</td>\n",
       "      <td>-14.971806</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-5.316218</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>intercepted</th>\n",
       "      <td>0</td>\n",
       "      <td>489</td>\n",
       "      <td>489</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>4.380414</td>\n",
       "      <td>-14.852152</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-5.435873</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>umpired</th>\n",
       "      <td>0</td>\n",
       "      <td>679</td>\n",
       "      <td>679</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>5.935587</td>\n",
       "      <td>-14.379399</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-5.908626</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fathered</th>\n",
       "      <td>0</td>\n",
       "      <td>699</td>\n",
       "      <td>699</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>5.375418</td>\n",
       "      <td>-14.337579</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-5.950446</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>his</th>\n",
       "      <td>0</td>\n",
       "      <td>708</td>\n",
       "      <td>708</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>-6.023451</td>\n",
       "      <td>-14.319148</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-5.968877</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>n</th>\n",
       "      <td>0</td>\n",
       "      <td>712</td>\n",
       "      <td>712</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>-1.612143</td>\n",
       "      <td>-14.311031</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-5.976993</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>icon</th>\n",
       "      <td>0</td>\n",
       "      <td>797</td>\n",
       "      <td>797</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>-0.194769</td>\n",
       "      <td>-14.148545</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-6.139480</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>batting</th>\n",
       "      <td>0</td>\n",
       "      <td>861</td>\n",
       "      <td>861</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>0.701369</td>\n",
       "      <td>-14.037246</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-6.250779</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>totaled</th>\n",
       "      <td>0</td>\n",
       "      <td>883</td>\n",
       "      <td>883</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>5.875469</td>\n",
       "      <td>-14.000887</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-6.287137</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>himself</th>\n",
       "      <td>0</td>\n",
       "      <td>8331</td>\n",
       "      <td>8331</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>2.381361</td>\n",
       "      <td>-10.764343</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-9.523682</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             she+word count  he+word count  total count  PMI she+word  \\\n",
       "rushed                   17           3238         3255      1.984473   \n",
       "pronounced                0            180          180   -100.000000   \n",
       "dispatched                0            181          181   -100.000000   \n",
       "captain                   0            182          182   -100.000000   \n",
       "calculated                0            191          191   -100.000000   \n",
       "downed                    0            194          194   -100.000000   \n",
       "invaded                   0            219          219   -100.000000   \n",
       "farmed                    0            221          221   -100.000000   \n",
       "gallantly                 0            229          229   -100.000000   \n",
       "trans                     0            258          258   -100.000000   \n",
       "bats                      0            271          271   -100.000000   \n",
       "commanded                50          14651        14701      3.969824   \n",
       "111                       0            296          296   -100.000000   \n",
       "man                       0            370          370   -100.000000   \n",
       "engraved                  0            382          382   -100.000000   \n",
       "erected                   0            450          450   -100.000000   \n",
       "intercepted               0            489          489   -100.000000   \n",
       "umpired                   0            679          679   -100.000000   \n",
       "fathered                  0            699          699   -100.000000   \n",
       "his                       0            708          708   -100.000000   \n",
       "n                         0            712          712   -100.000000   \n",
       "icon                      0            797          797   -100.000000   \n",
       "batting                   0            861          861   -100.000000   \n",
       "totaled                   0            883          883   -100.000000   \n",
       "himself                   0           8331         8331   -100.000000   \n",
       "\n",
       "             PMI he+word  log prob of word | he  log prob of word | she  \\\n",
       "rushed          5.710606             -12.127457              -16.118100   \n",
       "pronounced      1.929257             -16.288944              -20.288025   \n",
       "dispatched      2.766231             -16.280995              -20.288025   \n",
       "captain        -2.714158             -16.273090              -20.288025   \n",
       "calculated      3.397856             -16.203827              -20.288025   \n",
       "downed          4.191981             -16.181459              -20.288025   \n",
       "invaded         2.363412             -16.007430              -20.288025   \n",
       "farmed          4.412934             -15.994374              -20.288025   \n",
       "gallantly       5.751321             -15.943300              -20.288025   \n",
       "trans          -0.914177             -15.771981              -20.288025   \n",
       "bats            0.522018             -15.701327              -20.288025   \n",
       "commanded       6.257655              -9.949980              -14.615599   \n",
       "111             1.039456             -15.574471              -20.288025   \n",
       "man            -2.796346             -15.253514              -20.288025   \n",
       "engraved        3.237178             -15.207589              -20.288025   \n",
       "erected         2.622690             -14.971806              -20.288025   \n",
       "intercepted     4.380414             -14.852152              -20.288025   \n",
       "umpired         5.935587             -14.379399              -20.288025   \n",
       "fathered        5.375418             -14.337579              -20.288025   \n",
       "his            -6.023451             -14.319148              -20.288025   \n",
       "n              -1.612143             -14.311031              -20.288025   \n",
       "icon           -0.194769             -14.148545              -20.288025   \n",
       "batting         0.701369             -14.037246              -20.288025   \n",
       "totaled         5.875469             -14.000887              -20.288025   \n",
       "himself         2.381361             -10.764343              -20.288025   \n",
       "\n",
       "                  LLR  \n",
       "rushed      -3.990643  \n",
       "pronounced  -3.999081  \n",
       "dispatched  -4.007029  \n",
       "captain     -4.014935  \n",
       "calculated  -4.084197  \n",
       "downed      -4.106565  \n",
       "invaded     -4.280595  \n",
       "farmed      -4.293651  \n",
       "gallantly   -4.344725  \n",
       "trans       -4.516043  \n",
       "bats        -4.586698  \n",
       "commanded   -4.665619  \n",
       "111         -4.713554  \n",
       "man         -5.034510  \n",
       "engraved    -5.080435  \n",
       "erected     -5.316218  \n",
       "intercepted -5.435873  \n",
       "umpired     -5.908626  \n",
       "fathered    -5.950446  \n",
       "his         -5.968877  \n",
       "n           -5.976993  \n",
       "icon        -6.139480  \n",
       "batting     -6.250779  \n",
       "totaled     -6.287137  \n",
       "himself     -9.523682  "
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.tail(25)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.drop(['111','n'], axis=0, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "# We are most interested in words actually having word associativity with he/she (+ PMI)\n",
    "he_words = df.drop(df[df['PMI he+word'] <= 0 ].index).tail(100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "she_words = df.drop(df[df['PMI she+word'] <= 0 ].index).head(100)"
   ]
  },
  {
   "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>she+word count</th>\n",
       "      <th>he+word count</th>\n",
       "      <th>total count</th>\n",
       "      <th>PMI she+word</th>\n",
       "      <th>PMI he+word</th>\n",
       "      <th>log prob of word | he</th>\n",
       "      <th>log prob of word | she</th>\n",
       "      <th>LLR</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>duly</th>\n",
       "      <td>0</td>\n",
       "      <td>165</td>\n",
       "      <td>165</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>3.265200</td>\n",
       "      <td>-16.413750</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-3.874274</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>vetoed</th>\n",
       "      <td>0</td>\n",
       "      <td>175</td>\n",
       "      <td>175</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>4.091714</td>\n",
       "      <td>-16.329358</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-3.958666</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rushed</th>\n",
       "      <td>17</td>\n",
       "      <td>3238</td>\n",
       "      <td>3255</td>\n",
       "      <td>1.984473</td>\n",
       "      <td>5.710606</td>\n",
       "      <td>-12.127457</td>\n",
       "      <td>-16.118100</td>\n",
       "      <td>-3.990643</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>pronounced</th>\n",
       "      <td>0</td>\n",
       "      <td>180</td>\n",
       "      <td>180</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>1.929257</td>\n",
       "      <td>-16.288944</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-3.999081</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dispatched</th>\n",
       "      <td>0</td>\n",
       "      <td>181</td>\n",
       "      <td>181</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>2.766231</td>\n",
       "      <td>-16.280995</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-4.007029</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>calculated</th>\n",
       "      <td>0</td>\n",
       "      <td>191</td>\n",
       "      <td>191</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>3.397856</td>\n",
       "      <td>-16.203827</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-4.084197</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>downed</th>\n",
       "      <td>0</td>\n",
       "      <td>194</td>\n",
       "      <td>194</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>4.191981</td>\n",
       "      <td>-16.181459</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-4.106565</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>invaded</th>\n",
       "      <td>0</td>\n",
       "      <td>219</td>\n",
       "      <td>219</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>2.363412</td>\n",
       "      <td>-16.007430</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-4.280595</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>farmed</th>\n",
       "      <td>0</td>\n",
       "      <td>221</td>\n",
       "      <td>221</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>4.412934</td>\n",
       "      <td>-15.994374</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-4.293651</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>gallantly</th>\n",
       "      <td>0</td>\n",
       "      <td>229</td>\n",
       "      <td>229</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>5.751321</td>\n",
       "      <td>-15.943300</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-4.344725</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>bats</th>\n",
       "      <td>0</td>\n",
       "      <td>271</td>\n",
       "      <td>271</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>0.522018</td>\n",
       "      <td>-15.701327</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-4.586698</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>commanded</th>\n",
       "      <td>50</td>\n",
       "      <td>14651</td>\n",
       "      <td>14701</td>\n",
       "      <td>3.969824</td>\n",
       "      <td>6.257655</td>\n",
       "      <td>-9.949980</td>\n",
       "      <td>-14.615599</td>\n",
       "      <td>-4.665619</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>engraved</th>\n",
       "      <td>0</td>\n",
       "      <td>382</td>\n",
       "      <td>382</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>3.237178</td>\n",
       "      <td>-15.207589</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-5.080435</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>erected</th>\n",
       "      <td>0</td>\n",
       "      <td>450</td>\n",
       "      <td>450</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>2.622690</td>\n",
       "      <td>-14.971806</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-5.316218</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>intercepted</th>\n",
       "      <td>0</td>\n",
       "      <td>489</td>\n",
       "      <td>489</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>4.380414</td>\n",
       "      <td>-14.852152</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-5.435873</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>umpired</th>\n",
       "      <td>0</td>\n",
       "      <td>679</td>\n",
       "      <td>679</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>5.935587</td>\n",
       "      <td>-14.379399</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-5.908626</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fathered</th>\n",
       "      <td>0</td>\n",
       "      <td>699</td>\n",
       "      <td>699</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>5.375418</td>\n",
       "      <td>-14.337579</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-5.950446</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>batting</th>\n",
       "      <td>0</td>\n",
       "      <td>861</td>\n",
       "      <td>861</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>0.701369</td>\n",
       "      <td>-14.037246</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-6.250779</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>totaled</th>\n",
       "      <td>0</td>\n",
       "      <td>883</td>\n",
       "      <td>883</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>5.875469</td>\n",
       "      <td>-14.000887</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-6.287137</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>himself</th>\n",
       "      <td>0</td>\n",
       "      <td>8331</td>\n",
       "      <td>8331</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>2.381361</td>\n",
       "      <td>-10.764343</td>\n",
       "      <td>-20.288025</td>\n",
       "      <td>-9.523682</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             she+word count  he+word count  total count  PMI she+word  \\\n",
       "duly                      0            165          165   -100.000000   \n",
       "vetoed                    0            175          175   -100.000000   \n",
       "rushed                   17           3238         3255      1.984473   \n",
       "pronounced                0            180          180   -100.000000   \n",
       "dispatched                0            181          181   -100.000000   \n",
       "calculated                0            191          191   -100.000000   \n",
       "downed                    0            194          194   -100.000000   \n",
       "invaded                   0            219          219   -100.000000   \n",
       "farmed                    0            221          221   -100.000000   \n",
       "gallantly                 0            229          229   -100.000000   \n",
       "bats                      0            271          271   -100.000000   \n",
       "commanded                50          14651        14701      3.969824   \n",
       "engraved                  0            382          382   -100.000000   \n",
       "erected                   0            450          450   -100.000000   \n",
       "intercepted               0            489          489   -100.000000   \n",
       "umpired                   0            679          679   -100.000000   \n",
       "fathered                  0            699          699   -100.000000   \n",
       "batting                   0            861          861   -100.000000   \n",
       "totaled                   0            883          883   -100.000000   \n",
       "himself                   0           8331         8331   -100.000000   \n",
       "\n",
       "             PMI he+word  log prob of word | he  log prob of word | she  \\\n",
       "duly            3.265200             -16.413750              -20.288025   \n",
       "vetoed          4.091714             -16.329358              -20.288025   \n",
       "rushed          5.710606             -12.127457              -16.118100   \n",
       "pronounced      1.929257             -16.288944              -20.288025   \n",
       "dispatched      2.766231             -16.280995              -20.288025   \n",
       "calculated      3.397856             -16.203827              -20.288025   \n",
       "downed          4.191981             -16.181459              -20.288025   \n",
       "invaded         2.363412             -16.007430              -20.288025   \n",
       "farmed          4.412934             -15.994374              -20.288025   \n",
       "gallantly       5.751321             -15.943300              -20.288025   \n",
       "bats            0.522018             -15.701327              -20.288025   \n",
       "commanded       6.257655              -9.949980              -14.615599   \n",
       "engraved        3.237178             -15.207589              -20.288025   \n",
       "erected         2.622690             -14.971806              -20.288025   \n",
       "intercepted     4.380414             -14.852152              -20.288025   \n",
       "umpired         5.935587             -14.379399              -20.288025   \n",
       "fathered        5.375418             -14.337579              -20.288025   \n",
       "batting         0.701369             -14.037246              -20.288025   \n",
       "totaled         5.875469             -14.000887              -20.288025   \n",
       "himself         2.381361             -10.764343              -20.288025   \n",
       "\n",
       "                  LLR  \n",
       "duly        -3.874274  \n",
       "vetoed      -3.958666  \n",
       "rushed      -3.990643  \n",
       "pronounced  -3.999081  \n",
       "dispatched  -4.007029  \n",
       "calculated  -4.084197  \n",
       "downed      -4.106565  \n",
       "invaded     -4.280595  \n",
       "farmed      -4.293651  \n",
       "gallantly   -4.344725  \n",
       "bats        -4.586698  \n",
       "commanded   -4.665619  \n",
       "engraved    -5.080435  \n",
       "erected     -5.316218  \n",
       "intercepted -5.435873  \n",
       "umpired     -5.908626  \n",
       "fathered    -5.950446  \n",
       "batting     -6.250779  \n",
       "totaled     -6.287137  \n",
       "himself     -9.523682  "
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "he_words.tail(20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "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>she+word count</th>\n",
       "      <th>he+word count</th>\n",
       "      <th>total count</th>\n",
       "      <th>PMI she+word</th>\n",
       "      <th>PMI he+word</th>\n",
       "      <th>log prob of word | he</th>\n",
       "      <th>log prob of word | she</th>\n",
       "      <th>LLR</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>herself</th>\n",
       "      <td>744</td>\n",
       "      <td>0</td>\n",
       "      <td>744</td>\n",
       "      <td>1.189844</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>-23.788790</td>\n",
       "      <td>-10.746928</td>\n",
       "      <td>13.041862</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>peaked</th>\n",
       "      <td>242</td>\n",
       "      <td>0</td>\n",
       "      <td>242</td>\n",
       "      <td>2.912550</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>-23.788790</td>\n",
       "      <td>-12.363212</td>\n",
       "      <td>11.425578</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>kicks</th>\n",
       "      <td>159</td>\n",
       "      <td>0</td>\n",
       "      <td>159</td>\n",
       "      <td>4.282957</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>-23.788790</td>\n",
       "      <td>-12.966096</td>\n",
       "      <td>10.822693</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>stoops</th>\n",
       "      <td>95</td>\n",
       "      <td>0</td>\n",
       "      <td>95</td>\n",
       "      <td>6.521314</td>\n",
       "      <td>-100.000000</td>\n",
       "      <td>-23.788790</td>\n",
       "      <td>-13.703062</td>\n",
       "      <td>10.085728</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>modeled</th>\n",
       "      <td>182</td>\n",
       "      <td>112</td>\n",
       "      <td>294</td>\n",
       "      <td>4.340933</td>\n",
       "      <td>2.627048</td>\n",
       "      <td>-16.968611</td>\n",
       "      <td>-12.772325</td>\n",
       "      <td>4.196286</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>contributes</th>\n",
       "      <td>95</td>\n",
       "      <td>60</td>\n",
       "      <td>155</td>\n",
       "      <td>4.352760</td>\n",
       "      <td>1.862693</td>\n",
       "      <td>-17.858052</td>\n",
       "      <td>-13.703062</td>\n",
       "      <td>4.154990</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>posed</th>\n",
       "      <td>354</td>\n",
       "      <td>249</td>\n",
       "      <td>603</td>\n",
       "      <td>4.876859</td>\n",
       "      <td>3.254666</td>\n",
       "      <td>-15.823005</td>\n",
       "      <td>-11.816349</td>\n",
       "      <td>4.006656</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>danced</th>\n",
       "      <td>513</td>\n",
       "      <td>458</td>\n",
       "      <td>971</td>\n",
       "      <td>5.207171</td>\n",
       "      <td>4.966041</td>\n",
       "      <td>-14.946439</td>\n",
       "      <td>-11.282400</td>\n",
       "      <td>3.664039</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>trains</th>\n",
       "      <td>88</td>\n",
       "      <td>91</td>\n",
       "      <td>179</td>\n",
       "      <td>3.745979</td>\n",
       "      <td>1.334844</td>\n",
       "      <td>-17.265228</td>\n",
       "      <td>-13.812291</td>\n",
       "      <td>3.452937</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>participates</th>\n",
       "      <td>105</td>\n",
       "      <td>109</td>\n",
       "      <td>214</td>\n",
       "      <td>4.846071</td>\n",
       "      <td>3.339100</td>\n",
       "      <td>-17.007430</td>\n",
       "      <td>-13.560104</td>\n",
       "      <td>3.447326</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>moves</th>\n",
       "      <td>80</td>\n",
       "      <td>101</td>\n",
       "      <td>181</td>\n",
       "      <td>1.763290</td>\n",
       "      <td>-0.434748</td>\n",
       "      <td>-17.116364</td>\n",
       "      <td>-13.948175</td>\n",
       "      <td>3.168190</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dated</th>\n",
       "      <td>227</td>\n",
       "      <td>303</td>\n",
       "      <td>530</td>\n",
       "      <td>3.088039</td>\n",
       "      <td>1.189732</td>\n",
       "      <td>-15.540862</td>\n",
       "      <td>-12.455135</td>\n",
       "      <td>3.085728</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>interned</th>\n",
       "      <td>90</td>\n",
       "      <td>125</td>\n",
       "      <td>215</td>\n",
       "      <td>4.620189</td>\n",
       "      <td>2.404596</td>\n",
       "      <td>-16.811510</td>\n",
       "      <td>-13.780230</td>\n",
       "      <td>3.031280</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>skated</th>\n",
       "      <td>94</td>\n",
       "      <td>132</td>\n",
       "      <td>226</td>\n",
       "      <td>4.821823</td>\n",
       "      <td>4.118634</td>\n",
       "      <td>-16.733507</td>\n",
       "      <td>-13.718169</td>\n",
       "      <td>3.015338</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>upset</th>\n",
       "      <td>129</td>\n",
       "      <td>182</td>\n",
       "      <td>311</td>\n",
       "      <td>2.981301</td>\n",
       "      <td>1.234424</td>\n",
       "      <td>-16.273090</td>\n",
       "      <td>-13.265657</td>\n",
       "      <td>3.007433</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>explores</th>\n",
       "      <td>123</td>\n",
       "      <td>175</td>\n",
       "      <td>298</td>\n",
       "      <td>3.112389</td>\n",
       "      <td>2.479784</td>\n",
       "      <td>-16.329358</td>\n",
       "      <td>-13.333828</td>\n",
       "      <td>2.995530</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>begged</th>\n",
       "      <td>61</td>\n",
       "      <td>90</td>\n",
       "      <td>151</td>\n",
       "      <td>4.686471</td>\n",
       "      <td>3.486064</td>\n",
       "      <td>-17.280995</td>\n",
       "      <td>-14.333828</td>\n",
       "      <td>2.947167</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>attends</th>\n",
       "      <td>82</td>\n",
       "      <td>128</td>\n",
       "      <td>210</td>\n",
       "      <td>3.808688</td>\n",
       "      <td>3.181424</td>\n",
       "      <td>-16.777562</td>\n",
       "      <td>-13.912985</td>\n",
       "      <td>2.864577</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>choreographed</th>\n",
       "      <td>127</td>\n",
       "      <td>198</td>\n",
       "      <td>325</td>\n",
       "      <td>3.728359</td>\n",
       "      <td>3.613664</td>\n",
       "      <td>-16.152165</td>\n",
       "      <td>-13.288025</td>\n",
       "      <td>2.864141</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>chairs</th>\n",
       "      <td>62</td>\n",
       "      <td>97</td>\n",
       "      <td>159</td>\n",
       "      <td>2.511577</td>\n",
       "      <td>0.818589</td>\n",
       "      <td>-17.174080</td>\n",
       "      <td>-14.310745</td>\n",
       "      <td>2.863335</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>turns</th>\n",
       "      <td>55</td>\n",
       "      <td>90</td>\n",
       "      <td>145</td>\n",
       "      <td>1.050514</td>\n",
       "      <td>0.151767</td>\n",
       "      <td>-17.280995</td>\n",
       "      <td>-14.480670</td>\n",
       "      <td>2.800325</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>admires</th>\n",
       "      <td>66</td>\n",
       "      <td>112</td>\n",
       "      <td>178</td>\n",
       "      <td>5.499895</td>\n",
       "      <td>5.108664</td>\n",
       "      <td>-16.968611</td>\n",
       "      <td>-14.221935</td>\n",
       "      <td>2.746675</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>competes</th>\n",
       "      <td>202</td>\n",
       "      <td>349</td>\n",
       "      <td>551</td>\n",
       "      <td>4.485681</td>\n",
       "      <td>3.976602</td>\n",
       "      <td>-15.337579</td>\n",
       "      <td>-12.622689</td>\n",
       "      <td>2.714890</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>lectures</th>\n",
       "      <td>73</td>\n",
       "      <td>130</td>\n",
       "      <td>203</td>\n",
       "      <td>1.574936</td>\n",
       "      <td>-0.553677</td>\n",
       "      <td>-16.755367</td>\n",
       "      <td>-14.078571</td>\n",
       "      <td>2.676796</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sings</th>\n",
       "      <td>552</td>\n",
       "      <td>1000</td>\n",
       "      <td>1552</td>\n",
       "      <td>4.071201</td>\n",
       "      <td>4.248261</td>\n",
       "      <td>-13.821563</td>\n",
       "      <td>-11.176889</td>\n",
       "      <td>2.644675</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>heads</th>\n",
       "      <td>78</td>\n",
       "      <td>145</td>\n",
       "      <td>223</td>\n",
       "      <td>1.721902</td>\n",
       "      <td>-0.110645</td>\n",
       "      <td>-16.598965</td>\n",
       "      <td>-13.984244</td>\n",
       "      <td>2.614721</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>directs</th>\n",
       "      <td>80</td>\n",
       "      <td>152</td>\n",
       "      <td>232</td>\n",
       "      <td>4.368964</td>\n",
       "      <td>3.421828</td>\n",
       "      <td>-16.531402</td>\n",
       "      <td>-13.948175</td>\n",
       "      <td>2.583227</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>headlined</th>\n",
       "      <td>118</td>\n",
       "      <td>224</td>\n",
       "      <td>342</td>\n",
       "      <td>4.527047</td>\n",
       "      <td>4.053381</td>\n",
       "      <td>-15.975009</td>\n",
       "      <td>-13.393207</td>\n",
       "      <td>2.581802</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>curated</th>\n",
       "      <td>117</td>\n",
       "      <td>224</td>\n",
       "      <td>341</td>\n",
       "      <td>3.078398</td>\n",
       "      <td>2.719081</td>\n",
       "      <td>-15.975009</td>\n",
       "      <td>-13.405382</td>\n",
       "      <td>2.569627</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rounded</th>\n",
       "      <td>49</td>\n",
       "      <td>95</td>\n",
       "      <td>144</td>\n",
       "      <td>4.302107</td>\n",
       "      <td>2.689447</td>\n",
       "      <td>-17.203827</td>\n",
       "      <td>-14.644168</td>\n",
       "      <td>2.559659</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>separated</th>\n",
       "      <td>113</td>\n",
       "      <td>324</td>\n",
       "      <td>437</td>\n",
       "      <td>1.930765</td>\n",
       "      <td>2.197273</td>\n",
       "      <td>-15.444494</td>\n",
       "      <td>-13.455135</td>\n",
       "      <td>1.989359</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>loves</th>\n",
       "      <td>399</td>\n",
       "      <td>1143</td>\n",
       "      <td>1542</td>\n",
       "      <td>3.988870</td>\n",
       "      <td>4.466079</td>\n",
       "      <td>-13.628918</td>\n",
       "      <td>-11.644168</td>\n",
       "      <td>1.984750</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>exited</th>\n",
       "      <td>36</td>\n",
       "      <td>106</td>\n",
       "      <td>142</td>\n",
       "      <td>4.753343</td>\n",
       "      <td>3.398548</td>\n",
       "      <td>-17.047323</td>\n",
       "      <td>-15.078571</td>\n",
       "      <td>1.968752</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>expresses</th>\n",
       "      <td>60</td>\n",
       "      <td>178</td>\n",
       "      <td>238</td>\n",
       "      <td>4.306649</td>\n",
       "      <td>3.788357</td>\n",
       "      <td>-16.304974</td>\n",
       "      <td>-14.357287</td>\n",
       "      <td>1.947687</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>carries</th>\n",
       "      <td>44</td>\n",
       "      <td>132</td>\n",
       "      <td>176</td>\n",
       "      <td>3.228687</td>\n",
       "      <td>1.166472</td>\n",
       "      <td>-16.733507</td>\n",
       "      <td>-14.796171</td>\n",
       "      <td>1.937336</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>favors</th>\n",
       "      <td>32</td>\n",
       "      <td>97</td>\n",
       "      <td>129</td>\n",
       "      <td>3.995275</td>\n",
       "      <td>3.167128</td>\n",
       "      <td>-17.174080</td>\n",
       "      <td>-15.243630</td>\n",
       "      <td>1.930449</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>manages</th>\n",
       "      <td>73</td>\n",
       "      <td>219</td>\n",
       "      <td>292</td>\n",
       "      <td>4.270181</td>\n",
       "      <td>3.617409</td>\n",
       "      <td>-16.007430</td>\n",
       "      <td>-14.078571</td>\n",
       "      <td>1.928859</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>filmed</th>\n",
       "      <td>367</td>\n",
       "      <td>1098</td>\n",
       "      <td>1465</td>\n",
       "      <td>3.437578</td>\n",
       "      <td>3.616890</td>\n",
       "      <td>-13.686814</td>\n",
       "      <td>-11.764463</td>\n",
       "      <td>1.922351</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>travels</th>\n",
       "      <td>112</td>\n",
       "      <td>344</td>\n",
       "      <td>456</td>\n",
       "      <td>2.559057</td>\n",
       "      <td>1.796299</td>\n",
       "      <td>-15.358337</td>\n",
       "      <td>-13.467846</td>\n",
       "      <td>1.890492</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>publishes</th>\n",
       "      <td>46</td>\n",
       "      <td>144</td>\n",
       "      <td>190</td>\n",
       "      <td>3.617465</td>\n",
       "      <td>3.021522</td>\n",
       "      <td>-16.608881</td>\n",
       "      <td>-14.733436</td>\n",
       "      <td>1.875445</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>offers</th>\n",
       "      <td>79</td>\n",
       "      <td>247</td>\n",
       "      <td>326</td>\n",
       "      <td>1.865272</td>\n",
       "      <td>1.268109</td>\n",
       "      <td>-15.834593</td>\n",
       "      <td>-13.966096</td>\n",
       "      <td>1.868497</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>confided</th>\n",
       "      <td>38</td>\n",
       "      <td>120</td>\n",
       "      <td>158</td>\n",
       "      <td>4.569905</td>\n",
       "      <td>4.507760</td>\n",
       "      <td>-16.869926</td>\n",
       "      <td>-15.002622</td>\n",
       "      <td>1.867304</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>forgot</th>\n",
       "      <td>36</td>\n",
       "      <td>114</td>\n",
       "      <td>150</td>\n",
       "      <td>3.362391</td>\n",
       "      <td>2.972494</td>\n",
       "      <td>-16.943300</td>\n",
       "      <td>-15.078571</td>\n",
       "      <td>1.864728</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>reprised</th>\n",
       "      <td>741</td>\n",
       "      <td>2315</td>\n",
       "      <td>3056</td>\n",
       "      <td>5.647794</td>\n",
       "      <td>6.578336</td>\n",
       "      <td>-12.611370</td>\n",
       "      <td>-10.752749</td>\n",
       "      <td>1.858621</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>prayed</th>\n",
       "      <td>47</td>\n",
       "      <td>150</td>\n",
       "      <td>197</td>\n",
       "      <td>4.433897</td>\n",
       "      <td>4.031006</td>\n",
       "      <td>-16.550385</td>\n",
       "      <td>-14.703062</td>\n",
       "      <td>1.847323</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>attempts</th>\n",
       "      <td>53</td>\n",
       "      <td>169</td>\n",
       "      <td>222</td>\n",
       "      <td>1.491752</td>\n",
       "      <td>-0.079321</td>\n",
       "      <td>-16.379399</td>\n",
       "      <td>-14.533137</td>\n",
       "      <td>1.846262</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>benefited</th>\n",
       "      <td>32</td>\n",
       "      <td>103</td>\n",
       "      <td>135</td>\n",
       "      <td>3.370612</td>\n",
       "      <td>2.428345</td>\n",
       "      <td>-17.088350</td>\n",
       "      <td>-15.243630</td>\n",
       "      <td>1.844720</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>interpreted</th>\n",
       "      <td>72</td>\n",
       "      <td>235</td>\n",
       "      <td>307</td>\n",
       "      <td>3.694832</td>\n",
       "      <td>2.908912</td>\n",
       "      <td>-15.906147</td>\n",
       "      <td>-14.098200</td>\n",
       "      <td>1.807947</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>acknowledges</th>\n",
       "      <td>70</td>\n",
       "      <td>229</td>\n",
       "      <td>299</td>\n",
       "      <td>4.928738</td>\n",
       "      <td>4.442597</td>\n",
       "      <td>-15.943300</td>\n",
       "      <td>-14.138277</td>\n",
       "      <td>1.805022</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>suffers</th>\n",
       "      <td>62</td>\n",
       "      <td>204</td>\n",
       "      <td>266</td>\n",
       "      <td>3.798739</td>\n",
       "      <td>3.077815</td>\n",
       "      <td>-16.109310</td>\n",
       "      <td>-14.310745</td>\n",
       "      <td>1.798565</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>descended</th>\n",
       "      <td>41</td>\n",
       "      <td>136</td>\n",
       "      <td>177</td>\n",
       "      <td>2.661706</td>\n",
       "      <td>1.508631</td>\n",
       "      <td>-16.690758</td>\n",
       "      <td>-14.895707</td>\n",
       "      <td>1.795050</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>joins</th>\n",
       "      <td>58</td>\n",
       "      <td>194</td>\n",
       "      <td>252</td>\n",
       "      <td>0.195988</td>\n",
       "      <td>-0.188229</td>\n",
       "      <td>-16.181459</td>\n",
       "      <td>-14.405382</td>\n",
       "      <td>1.776078</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>lists</th>\n",
       "      <td>65</td>\n",
       "      <td>218</td>\n",
       "      <td>283</td>\n",
       "      <td>1.057513</td>\n",
       "      <td>0.514032</td>\n",
       "      <td>-16.014003</td>\n",
       "      <td>-14.243630</td>\n",
       "      <td>1.770372</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>reports</th>\n",
       "      <td>72</td>\n",
       "      <td>243</td>\n",
       "      <td>315</td>\n",
       "      <td>0.393463</td>\n",
       "      <td>-0.416833</td>\n",
       "      <td>-15.858052</td>\n",
       "      <td>-14.098200</td>\n",
       "      <td>1.759852</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>legally</th>\n",
       "      <td>54</td>\n",
       "      <td>183</td>\n",
       "      <td>237</td>\n",
       "      <td>1.930152</td>\n",
       "      <td>2.224891</td>\n",
       "      <td>-16.265228</td>\n",
       "      <td>-14.506665</td>\n",
       "      <td>1.758563</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hails</th>\n",
       "      <td>84</td>\n",
       "      <td>285</td>\n",
       "      <td>369</td>\n",
       "      <td>4.503334</td>\n",
       "      <td>3.536180</td>\n",
       "      <td>-15.628918</td>\n",
       "      <td>-13.878634</td>\n",
       "      <td>1.750285</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>worried</th>\n",
       "      <td>26</td>\n",
       "      <td>90</td>\n",
       "      <td>116</td>\n",
       "      <td>2.349835</td>\n",
       "      <td>1.822664</td>\n",
       "      <td>-17.280995</td>\n",
       "      <td>-15.533137</td>\n",
       "      <td>1.747858</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>pleased</th>\n",
       "      <td>35</td>\n",
       "      <td>121</td>\n",
       "      <td>156</td>\n",
       "      <td>2.679643</td>\n",
       "      <td>1.820069</td>\n",
       "      <td>-16.858052</td>\n",
       "      <td>-15.118100</td>\n",
       "      <td>1.739953</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>abruptly</th>\n",
       "      <td>31</td>\n",
       "      <td>108</td>\n",
       "      <td>139</td>\n",
       "      <td>3.232988</td>\n",
       "      <td>2.657647</td>\n",
       "      <td>-17.020605</td>\n",
       "      <td>-15.288025</td>\n",
       "      <td>1.732581</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>produces</th>\n",
       "      <td>66</td>\n",
       "      <td>228</td>\n",
       "      <td>294</td>\n",
       "      <td>3.165778</td>\n",
       "      <td>2.730993</td>\n",
       "      <td>-15.949586</td>\n",
       "      <td>-14.221935</td>\n",
       "      <td>1.727651</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>100 rows × 8 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "               she+word count  he+word count  total count  PMI she+word  \\\n",
       "herself                   744              0          744      1.189844   \n",
       "peaked                    242              0          242      2.912550   \n",
       "kicks                     159              0          159      4.282957   \n",
       "stoops                     95              0           95      6.521314   \n",
       "modeled                   182            112          294      4.340933   \n",
       "contributes                95             60          155      4.352760   \n",
       "posed                     354            249          603      4.876859   \n",
       "danced                    513            458          971      5.207171   \n",
       "trains                     88             91          179      3.745979   \n",
       "participates              105            109          214      4.846071   \n",
       "moves                      80            101          181      1.763290   \n",
       "dated                     227            303          530      3.088039   \n",
       "interned                   90            125          215      4.620189   \n",
       "skated                     94            132          226      4.821823   \n",
       "upset                     129            182          311      2.981301   \n",
       "explores                  123            175          298      3.112389   \n",
       "begged                     61             90          151      4.686471   \n",
       "attends                    82            128          210      3.808688   \n",
       "choreographed             127            198          325      3.728359   \n",
       "chairs                     62             97          159      2.511577   \n",
       "turns                      55             90          145      1.050514   \n",
       "admires                    66            112          178      5.499895   \n",
       "competes                  202            349          551      4.485681   \n",
       "lectures                   73            130          203      1.574936   \n",
       "sings                     552           1000         1552      4.071201   \n",
       "heads                      78            145          223      1.721902   \n",
       "directs                    80            152          232      4.368964   \n",
       "headlined                 118            224          342      4.527047   \n",
       "curated                   117            224          341      3.078398   \n",
       "rounded                    49             95          144      4.302107   \n",
       "...                       ...            ...          ...           ...   \n",
       "separated                 113            324          437      1.930765   \n",
       "loves                     399           1143         1542      3.988870   \n",
       "exited                     36            106          142      4.753343   \n",
       "expresses                  60            178          238      4.306649   \n",
       "carries                    44            132          176      3.228687   \n",
       "favors                     32             97          129      3.995275   \n",
       "manages                    73            219          292      4.270181   \n",
       "filmed                    367           1098         1465      3.437578   \n",
       "travels                   112            344          456      2.559057   \n",
       "publishes                  46            144          190      3.617465   \n",
       "offers                     79            247          326      1.865272   \n",
       "confided                   38            120          158      4.569905   \n",
       "forgot                     36            114          150      3.362391   \n",
       "reprised                  741           2315         3056      5.647794   \n",
       "prayed                     47            150          197      4.433897   \n",
       "attempts                   53            169          222      1.491752   \n",
       "benefited                  32            103          135      3.370612   \n",
       "interpreted                72            235          307      3.694832   \n",
       "acknowledges               70            229          299      4.928738   \n",
       "suffers                    62            204          266      3.798739   \n",
       "descended                  41            136          177      2.661706   \n",
       "joins                      58            194          252      0.195988   \n",
       "lists                      65            218          283      1.057513   \n",
       "reports                    72            243          315      0.393463   \n",
       "legally                    54            183          237      1.930152   \n",
       "hails                      84            285          369      4.503334   \n",
       "worried                    26             90          116      2.349835   \n",
       "pleased                    35            121          156      2.679643   \n",
       "abruptly                   31            108          139      3.232988   \n",
       "produces                   66            228          294      3.165778   \n",
       "\n",
       "               PMI he+word  log prob of word | he  log prob of word | she  \\\n",
       "herself        -100.000000             -23.788790              -10.746928   \n",
       "peaked         -100.000000             -23.788790              -12.363212   \n",
       "kicks          -100.000000             -23.788790              -12.966096   \n",
       "stoops         -100.000000             -23.788790              -13.703062   \n",
       "modeled           2.627048             -16.968611              -12.772325   \n",
       "contributes       1.862693             -17.858052              -13.703062   \n",
       "posed             3.254666             -15.823005              -11.816349   \n",
       "danced            4.966041             -14.946439              -11.282400   \n",
       "trains            1.334844             -17.265228              -13.812291   \n",
       "participates      3.339100             -17.007430              -13.560104   \n",
       "moves            -0.434748             -17.116364              -13.948175   \n",
       "dated             1.189732             -15.540862              -12.455135   \n",
       "interned          2.404596             -16.811510              -13.780230   \n",
       "skated            4.118634             -16.733507              -13.718169   \n",
       "upset             1.234424             -16.273090              -13.265657   \n",
       "explores          2.479784             -16.329358              -13.333828   \n",
       "begged            3.486064             -17.280995              -14.333828   \n",
       "attends           3.181424             -16.777562              -13.912985   \n",
       "choreographed     3.613664             -16.152165              -13.288025   \n",
       "chairs            0.818589             -17.174080              -14.310745   \n",
       "turns             0.151767             -17.280995              -14.480670   \n",
       "admires           5.108664             -16.968611              -14.221935   \n",
       "competes          3.976602             -15.337579              -12.622689   \n",
       "lectures         -0.553677             -16.755367              -14.078571   \n",
       "sings             4.248261             -13.821563              -11.176889   \n",
       "heads            -0.110645             -16.598965              -13.984244   \n",
       "directs           3.421828             -16.531402              -13.948175   \n",
       "headlined         4.053381             -15.975009              -13.393207   \n",
       "curated           2.719081             -15.975009              -13.405382   \n",
       "rounded           2.689447             -17.203827              -14.644168   \n",
       "...                    ...                    ...                     ...   \n",
       "separated         2.197273             -15.444494              -13.455135   \n",
       "loves             4.466079             -13.628918              -11.644168   \n",
       "exited            3.398548             -17.047323              -15.078571   \n",
       "expresses         3.788357             -16.304974              -14.357287   \n",
       "carries           1.166472             -16.733507              -14.796171   \n",
       "favors            3.167128             -17.174080              -15.243630   \n",
       "manages           3.617409             -16.007430              -14.078571   \n",
       "filmed            3.616890             -13.686814              -11.764463   \n",
       "travels           1.796299             -15.358337              -13.467846   \n",
       "publishes         3.021522             -16.608881              -14.733436   \n",
       "offers            1.268109             -15.834593              -13.966096   \n",
       "confided          4.507760             -16.869926              -15.002622   \n",
       "forgot            2.972494             -16.943300              -15.078571   \n",
       "reprised          6.578336             -12.611370              -10.752749   \n",
       "prayed            4.031006             -16.550385              -14.703062   \n",
       "attempts         -0.079321             -16.379399              -14.533137   \n",
       "benefited         2.428345             -17.088350              -15.243630   \n",
       "interpreted       2.908912             -15.906147              -14.098200   \n",
       "acknowledges      4.442597             -15.943300              -14.138277   \n",
       "suffers           3.077815             -16.109310              -14.310745   \n",
       "descended         1.508631             -16.690758              -14.895707   \n",
       "joins            -0.188229             -16.181459              -14.405382   \n",
       "lists             0.514032             -16.014003              -14.243630   \n",
       "reports          -0.416833             -15.858052              -14.098200   \n",
       "legally           2.224891             -16.265228              -14.506665   \n",
       "hails             3.536180             -15.628918              -13.878634   \n",
       "worried           1.822664             -17.280995              -15.533137   \n",
       "pleased           1.820069             -16.858052              -15.118100   \n",
       "abruptly          2.657647             -17.020605              -15.288025   \n",
       "produces          2.730993             -15.949586              -14.221935   \n",
       "\n",
       "                     LLR  \n",
       "herself        13.041862  \n",
       "peaked         11.425578  \n",
       "kicks          10.822693  \n",
       "stoops         10.085728  \n",
       "modeled         4.196286  \n",
       "contributes     4.154990  \n",
       "posed           4.006656  \n",
       "danced          3.664039  \n",
       "trains          3.452937  \n",
       "participates    3.447326  \n",
       "moves           3.168190  \n",
       "dated           3.085728  \n",
       "interned        3.031280  \n",
       "skated          3.015338  \n",
       "upset           3.007433  \n",
       "explores        2.995530  \n",
       "begged          2.947167  \n",
       "attends         2.864577  \n",
       "choreographed   2.864141  \n",
       "chairs          2.863335  \n",
       "turns           2.800325  \n",
       "admires         2.746675  \n",
       "competes        2.714890  \n",
       "lectures        2.676796  \n",
       "sings           2.644675  \n",
       "heads           2.614721  \n",
       "directs         2.583227  \n",
       "headlined       2.581802  \n",
       "curated         2.569627  \n",
       "rounded         2.559659  \n",
       "...                  ...  \n",
       "separated       1.989359  \n",
       "loves           1.984750  \n",
       "exited          1.968752  \n",
       "expresses       1.947687  \n",
       "carries         1.937336  \n",
       "favors          1.930449  \n",
       "manages         1.928859  \n",
       "filmed          1.922351  \n",
       "travels         1.890492  \n",
       "publishes       1.875445  \n",
       "offers          1.868497  \n",
       "confided        1.867304  \n",
       "forgot          1.864728  \n",
       "reprised        1.858621  \n",
       "prayed          1.847323  \n",
       "attempts        1.846262  \n",
       "benefited       1.844720  \n",
       "interpreted     1.807947  \n",
       "acknowledges    1.805022  \n",
       "suffers         1.798565  \n",
       "descended       1.795050  \n",
       "joins           1.776078  \n",
       "lists           1.770372  \n",
       "reports         1.759852  \n",
       "legally         1.758563  \n",
       "hails           1.750285  \n",
       "worried         1.747858  \n",
       "pleased         1.739953  \n",
       "abruptly        1.732581  \n",
       "produces        1.727651  \n",
       "\n",
       "[100 rows x 8 columns]"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "she_words\n",
    "# \"She Kicks\" is a women's football magazine\n",
    "# \"She Stoops to Conquer\" is the name of a film"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "he_words.drop(['batting'], axis=0, inplace=True) # Doesn't make sense gramatically?\n",
    "Y2, X2 = he_words.tail(20).index, he_words.tail(20)['LLR']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "she_top = she_words.drop(she_words[she_words['she+word count'] < 90 ].index).head(20)['LLR']\n",
    "Y1, X1 = she_top.index, she_top "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 972x1440 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "title=\"Log likelihood ratio that word appears after she/he in articles about women and men, respectively\"\n",
    "subtitle=\"Limited to words appearing at least 90 times after pronoun and with positive word association (PMI) for that gender\"\n",
    "subtitle2=\"\\n\\nHe...                                       She...\"\n",
    "\n",
    "Y = np.arange(40)\n",
    "\n",
    "X_women = np.concatenate((np.zeros(20), np.flip(X1.values, 0)), axis=0)\n",
    "X_men = np.concatenate((np.flip(X2.values, 0), np.zeros(20)), axis=0)\n",
    "\n",
    "plt.figure(figsize=(13.5,20))\n",
    "plt.axes([0.025, 0.025, 0.95, 0.95])\n",
    "plt.barh(Y, X_women, facecolor=\"None\", edgecolor='#2ca79b')\n",
    "plt.barh(Y, X_men, facecolor=\"None\", edgecolor='#a52e3b')\n",
    "\n",
    "for x, y, label in zip(X_women, Y, np.concatenate((np.empty(20, dtype=\"U\") , np.flip(Y1,0)))):\n",
    "    plt.text(x-0.1, y, label, ha='right', va= 'center', color=\"black\", fontsize=13)\n",
    "for x, y, label in zip(X_men, Y, np.concatenate((np.flip(Y2,0), np.empty(20, dtype=\"U\")))):\n",
    "    plt.text(x+0.2, y, label, ha='left', va= 'center', color=\"black\", fontsize=13)\n",
    "\n",
    "\n",
    "plt.title(title, fontsize=17, y=1.09)\n",
    "plt.suptitle(subtitle + subtitle2, fontsize=14, y=1.05)\n",
    "plt.axvline(x=0, color=\"black\")\n",
    "plt.ylim(-.5, 39.6)\n",
    "plt.yticks(())\n",
    "plt.xlim(-12, 13.5)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "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": "3.6.5"
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}