Big-ARTM testing example.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Tagging and Visualizing Community Interests, Measuring Influence inside community, and impact of community in the whole network"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Plan\n",
    "1. Данные для анализа (выбрать сообщества для анализа (И))\n",
    "2. Introduction for paper (Обзор того, как делается теггирование (1,5 стр) + по 0,5 на каждый алгоритм кластеризации), См. Расстояние Уорда, косинус (Ward distance), сокращение размерности.\n",
    "3. Подсчитать наиболее значимые слова в полученном кластере Disp(TF/IDF), top 10\n",
    "4. Проверить аддитивность, транзитивность, гомогенность сообществ\n",
    "Profit"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.8.3\n"
     ]
    }
   ],
   "source": [
    "# -*- coding: utf-8 -*-\n",
    "%matplotlib inline\n",
    "import glob\n",
    "import os\n",
    "import matplotlib.pyplot as plt\n",
    "import time\n",
    "import artm\n",
    "\n",
    "print artm.version()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "a1= []\n",
    "a2 =[]\n",
    "a3 =[]\n",
    "a4 =[]\n",
    "a5 =[]\n",
    "\n",
    "for i in range(1):\n",
    "    batch_vectorizer = None\n",
    "    if len(glob.glob(os.path.join('vk', '*.batch'))) < 1:\n",
    "        batch_vectorizer = artm.BatchVectorizer(data_path='vw.vk.txt', data_format='vowpal_wabbit', target_folder='vk')\n",
    "    else:\n",
    "        batch_vectorizer = artm.BatchVectorizer(data_path='vk', data_format='batches')\n",
    "    dictionary = artm.Dictionary()\n",
    "    model_artm = artm.ARTM(topic_names=['topic_{}'.format(i) for i in xrange(50)],\n",
    "                           scores=[artm.PerplexityScore(name='PerplexityScore', dictionary=dictionary)],\n",
    "                           regularizers=[artm.SmoothSparseThetaRegularizer(name='SparseTheta', tau=-1.6)],\n",
    "                           cache_theta=True)\n",
    "    if not os.path.isfile('vk/dictionary.dict'):\n",
    "        dictionary.gather(data_path=batch_vectorizer.data_path)\n",
    "        dictionary.save(dictionary_path='vk/dictionary.dict')\n",
    "    dictionary.load(dictionary_path='vk/dictionary.dict')\n",
    "    model_artm.initialize(dictionary=dictionary)\n",
    "    model_artm.scores.add(artm.SparsityPhiScore(name='SparsityPhiScore'))\n",
    "    model_artm.scores.add(artm.SparsityThetaScore(name='SparsityThetaScore'))\n",
    "    model_artm.scores.add(artm.TopicKernelScore(name='TopicKernelScore', probability_mass_threshold=0.1))\n",
    "    \n",
    "    model_artm.regularizers.add(artm.SmoothSparsePhiRegularizer(name='SparsePhi', tau=-0.2))\n",
    "    model_artm.regularizers.add(artm.DecorrelatorPhiRegularizer(name='DecorrelatorPhi', tau=2.5e+5))\n",
    "    \n",
    "    model_artm.scores.add(artm.TopTokensScore(name='TopTokensScore', num_tokens=10))\n",
    "    model_artm.num_document_passes = 1\n",
    "    model_artm.fit_offline(batch_vectorizer=batch_vectorizer, num_collection_passes=1)\n",
    "    a1.append(model_artm.score_tracker['SparsityPhiScore'].last_value)\n",
    "    a2.append(model_artm.score_tracker['SparsityThetaScore'].last_value)\n",
    "    a3.append(model_artm.score_tracker['TopicKernelScore'].last_average_contrast)\n",
    "    a4.append(model_artm.score_tracker['TopicKernelScore'].last_average_purity)\n",
    "    a5.append(model_artm.score_tracker['PerplexityScore'].last_value)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.661718439084\n",
      "0.36649645995\n",
      "0.00497119698324\n",
      "117195.205017\n"
     ]
    }
   ],
   "source": [
    "print(model_artm.score_tracker['SparsityThetaScore'].last_value)\n",
    "print(model_artm.score_tracker['TopicKernelScore'].last_average_contrast)\n",
    "print(model_artm.score_tracker['TopicKernelScore'].last_average_purity)\n",
    "print(model_artm.score_tracker['PerplexityScore'].last_value)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.661715608322\n",
      "0.366545269102\n",
      "0.00497362809348\n",
      "117099.28948\n"
     ]
    }
   ],
   "source": [
    "print(model_artm.score_tracker['SparsityThetaScore'].last_value)\n",
    "print(model_artm.score_tracker['TopicKernelScore'].last_average_contrast)\n",
    "print(model_artm.score_tracker['TopicKernelScore'].last_average_purity)\n",
    "print(model_artm.score_tracker['PerplexityScore'].last_value)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "batch_vectorizer = None\n",
    "if len(glob.glob(os.path.join('vk', '*.batch'))) < 1:\n",
    "    batch_vectorizer = artm.BatchVectorizer(data_path='vw.vk.txt', data_format='vowpal_wabbit', target_folder='vk')\n",
    "else:\n",
    "    batch_vectorizer = artm.BatchVectorizer(data_path='vk', data_format='batches')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "dictionary = artm.Dictionary()\n",
    "\n",
    "model_plsa = artm.ARTM(topic_names=['topic_{}'.format(i) for i in xrange(50)],\n",
    "                       scores=[artm.PerplexityScore(name='PerplexityScore',\n",
    "                                                    dictionary=dictionary)],\n",
    "                       cache_theta=True)\n",
    "\n",
    "model_artm = artm.ARTM(topic_names=['topic_{}'.format(i) for i in xrange(50)],\n",
    "                       scores=[artm.PerplexityScore(name='PerplexityScore',\n",
    "                                                    dictionary=dictionary)],\n",
    "                       regularizers=[artm.SmoothSparseThetaRegularizer(name='SparseTheta', tau=-1.85)],\n",
    "                       cache_theta=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "if not os.path.isfile('vk/dictionary.dict'):\n",
    "    dictionary.gather(data_path=batch_vectorizer.data_path)\n",
    "    dictionary.save(dictionary_path='vk/dictionary.dict')\n",
    "\n",
    "dictionary.load(dictionary_path='vk/dictionary.dict')\n",
    "dictionary.load(dictionary_path='vk/dictionary.dict')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "model_plsa.initialize(dictionary=dictionary)\n",
    "model_artm.initialize(dictionary=dictionary)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "model_plsa.scores.add(artm.SparsityPhiScore(name='SparsityPhiScore'))\n",
    "model_plsa.scores.add(artm.SparsityThetaScore(name='SparsityThetaScore'))\n",
    "model_plsa.scores.add(artm.TopicKernelScore(name='TopicKernelScore', probability_mass_threshold=0.1))\n",
    "\n",
    "model_artm.scores.add(artm.SparsityPhiScore(name='SparsityPhiScore'))\n",
    "model_artm.scores.add(artm.SparsityThetaScore(name='SparsityThetaScore'))\n",
    "model_artm.scores.add(artm.TopicKernelScore(name='TopicKernelScore', probability_mass_threshold=0.1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "model_artm.regularizers.add(artm.SmoothSparsePhiRegularizer(name='SparsePhi', tau=-0.2))\n",
    "model_artm.regularizers.add(artm.DecorrelatorPhiRegularizer(name='DecorrelatorPhi', tau=2.5e+5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "model_plsa.scores.add(artm.TopTokensScore(name='TopTokensScore', num_tokens=10))\n",
    "model_artm.scores.add(artm.TopTokensScore(name='TopTokensScore', num_tokens=10))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "model_plsa.num_document_passes = 1\n",
    "model_artm.num_document_passes = 1\n",
    "\n",
    "model_plsa.fit_offline(batch_vectorizer=batch_vectorizer, num_collection_passes=1)\n",
    "model_artm.fit_offline(batch_vectorizer=batch_vectorizer, num_collection_passes=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model_plsa.fit_offline(batch_vectorizer=batch_vectorizer, num_collection_passes=4)\n",
    "model_artm.fit_offline(batch_vectorizer=batch_vectorizer, num_collection_passes=4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model_plsa.fit_offline(batch_vectorizer=batch_vectorizer, num_collection_passes=5)\n",
    "model_artm.fit_offline(batch_vectorizer=batch_vectorizer, num_collection_passes=5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def print_measures(model_plsa, model_artm):\n",
    "    print 'Sparsity Phi: {0:.3f} (PLSA) vs. {1:.3f} (ARTM)'.format(\n",
    "        model_plsa.score_tracker['SparsityPhiScore'].last_value,\n",
    "        model_artm.score_tracker['SparsityPhiScore'].last_value)\n",
    "\n",
    "    print 'Sparsity Theta: {0:.3f} (PLSA) vs. {1:.3f} (ARTM)'.format(\n",
    "        model_plsa.score_tracker['SparsityThetaScore'].last_value,\n",
    "        model_artm.score_tracker['SparsityThetaScore'].last_value)\n",
    "\n",
    "    print 'Kernel contrast: {0:.3f} (PLSA) vs. {1:.3f} (ARTM)'.format(\n",
    "        model_plsa.score_tracker['TopicKernelScore'].last_average_contrast,\n",
    "        model_artm.score_tracker['TopicKernelScore'].last_average_contrast)\n",
    "\n",
    "    print 'Kernel purity: {0:.3f} (PLSA) vs. {1:.3f} (ARTM)'.format(\n",
    "        model_plsa.score_tracker['TopicKernelScore'].last_average_purity,\n",
    "        model_artm.score_tracker['TopicKernelScore'].last_average_purity)\n",
    "\n",
    "    print 'Perplexity: {0:.3f} (PLSA) vs. {1:.3f} (ARTM)'.format(\n",
    "        model_plsa.score_tracker['PerplexityScore'].last_value,\n",
    "        model_artm.score_tracker['PerplexityScore'].last_value)\n",
    "\n",
    "    plt.plot(xrange(model_plsa.num_phi_updates), model_plsa.score_tracker['PerplexityScore'].value, 'b--',\n",
    "             xrange(model_artm.num_phi_updates), model_artm.score_tracker['PerplexityScore'].value, 'r--', linewidth=2)\n",
    "    plt.xlabel('Iterations count')\n",
    "    plt.ylabel('PLSA perp. (blue), ARTM perp. (red)')\n",
    "    plt.grid(True)\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def print_measures(model_artm):\n",
    "    print 'Sparsity Phi: {1:.3f} (ARTM)'.format(\n",
    "        model_artm.score_tracker['SparsityPhiScore'].last_value)\n",
    "\n",
    "    print 'Sparsity Theta: {1:.3f} (ARTM)'.format(\n",
    "        model_artm.score_tracker['SparsityThetaScore'].last_value)\n",
    "\n",
    "    print 'Kernel contrast: {1:.3f} (ARTM)'.format(\n",
    "        model_artm.score_tracker['TopicKernelScore'].last_average_contrast)\n",
    "\n",
    "    print 'Kernel purity: {1:.3f} (ARTM)'.format(\n",
    "        model_artm.score_tracker['TopicKernelScore'].last_average_purity)\n",
    "\n",
    "    print 'Perplexity: {1:.3f} (ARTM)'.format(\n",
    "        model_artm.score_tracker['PerplexityScore'].last_value)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "topic_words= []\n",
    "\n",
    "for topic_name in model_artm.topic_names:\n",
    "    topic_words.append(model_artm.score_tracker['TopTokensScore'].last_tokens[topic_name])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "topic 0:\n",
      "млрд_NOUN\n",
      "год_NOUN\n",
      "рубль_NOUN\n",
      "миллиард_NOUN\n",
      "составлять_VERB\n",
      "доллар_NOUN\n",
      "компания_NOUN\n",
      "руб_NOUN\n",
      "состояние_NOUN\n",
      "акционер_NOUN\n",
      "\n",
      "\n",
      "topic 1:\n",
      "этот_ADJ\n",
      "они_NOUN\n",
      "который_ADJ\n",
      "это_NOUN\n",
      "свой_ADJ\n",
      "весь_ADJ\n",
      "она_NOUN\n",
      "такой_ADJ\n",
      "либеральный_ADJ\n",
      "его_ADJ\n",
      "\n",
      "\n",
      "topic 2:\n",
      "музыка_NOUN\n",
      "язык_NOUN\n",
      "сергей_NOUN\n",
      "концерт_NOUN\n",
      "курехин_NOUN\n",
      "смотреть_INFN\n",
      "алгебра_NOUN\n",
      "музыкант_NOUN\n",
      "композитор_NOUN\n",
      "кино_NOUN\n",
      "\n",
      "\n",
      "topic 3:\n",
      "турция_NOUN\n",
      "отец_NOUN\n",
      "год_NOUN\n",
      "сын_NOUN\n",
      "карта_NOUN\n",
      "большой_ADJ\n",
      "дом_NOUN\n",
      "она_NOUN\n",
      "помогать_INFN\n",
      "фото_NOUN\n",
      "\n",
      "\n",
      "topic 4:\n",
      "форум_NOUN\n",
      "новый_ADJ\n",
      "литература_NOUN\n",
      "мочь_VERB\n",
      "язык_NOUN\n",
      "информация_NOUN\n",
      "технология_NOUN\n",
      "развитие_NOUN\n",
      "британский_ADJ\n",
      "российский_ADJ\n",
      "\n",
      "\n",
      "topic 5:\n",
      "гость_NOUN\n",
      "вечеринка_NOUN\n",
      "сечин_NOUN\n",
      "прием_NOUN\n",
      "отель_NOUN\n",
      "вечер_NOUN\n",
      "стол_NOUN\n",
      "праздник_NOUN\n",
      "вино_NOUN\n",
      "дерипаска_NOUN\n",
      "\n",
      "\n",
      "topic 6:\n",
      "борис_NOUN\n",
      "мочь_VERB\n",
      "дурицкая_NOUN\n",
      "человек_NOUN\n",
      "немцов_NOUN\n",
      "работать_VERB\n",
      "который_ADJ\n",
      "жизнь_NOUN\n",
      "киев_NOUN\n",
      "она_NOUN\n",
      "\n",
      "\n",
      "topic 7:\n",
      "год_NOUN\n",
      "дод_NOUN\n",
      "директор_NOUN\n",
      "русгидро_NOUN\n",
      "богатый_ADJ\n",
      "евгений_NOUN\n",
      "председатель_NOUN\n",
      "совет_NOUN\n",
      "бывший_ADJ\n",
      "компания_NOUN\n",
      "\n",
      "\n",
      "topic 8:\n",
      "республика_NOUN\n",
      "общественный_ADJ\n",
      "политический_ADJ\n",
      "участие_NOUN\n",
      "организация_NOUN\n",
      "книжный_ADJ\n",
      "работа_NOUN\n",
      "город_NOUN\n",
      "региональный_ADJ\n",
      "смена_NOUN\n",
      "\n",
      "\n",
      "topic 9:\n",
      "эстония_NOUN\n",
      "алла_NOUN\n",
      "фото_NOUN\n",
      "пугачева_NOUN\n",
      "приватизация_NOUN\n",
      "россия_NOUN\n",
      "эксперт_NOUN\n",
      "санкция_NOUN\n",
      "последний_ADJ\n",
      "песня_NOUN\n",
      "\n",
      "\n",
      "topic 10:\n",
      "это_NOUN\n",
      "который_ADJ\n",
      "свой_ADJ\n",
      "статус_NOUN\n",
      "работа_NOUN\n",
      "первый_ADJ\n",
      "этот_ADJ\n",
      "мероприятие_NOUN\n",
      "управление_NOUN\n",
      "другой_ADJ\n",
      "\n",
      "\n",
      "topic 11:\n",
      "театр_NOUN\n",
      "спектакль_NOUN\n",
      "михаил_NOUN\n",
      "прямой_ADJ\n",
      "фонд_NOUN\n",
      "организатор_NOUN\n",
      "рассказывать_VERB\n",
      "прохоров_NOUN\n",
      "театральный_ADJ\n",
      "собеседник_NOUN\n",
      "\n",
      "\n",
      "topic 12:\n",
      "военный_ADJ\n",
      "дед_NOUN\n",
      "ссср_NOUN\n",
      "автор_NOUN\n",
      "гайдар_NOUN\n",
      "член_NOUN\n",
      "отец_NOUN\n",
      "отдел_NOUN\n",
      "деятель_NOUN\n",
      "начальник_NOUN\n",
      "\n",
      "\n",
      "topic 13:\n",
      "время_NOUN\n",
      "страна_NOUN\n",
      "год_NOUN\n",
      "церковь_NOUN\n",
      "очередной_ADJ\n",
      "мочь_VERB\n",
      "государство_NOUN\n",
      "гражданин_NOUN\n",
      "православный_ADJ\n",
      "русский_ADJ\n",
      "\n",
      "\n",
      "topic 14:\n",
      "деньги_NOUN\n",
      "становиться_VERB\n",
      "человек_NOUN\n",
      "мочь_VERB\n",
      "россия_NOUN\n",
      "команда_NOUN\n",
      "раз_NOUN\n",
      "доллар_NOUN\n",
      "время_NOUN\n",
      "фридман_NOUN\n",
      "\n",
      "\n",
      "topic 15:\n",
      "холдинг_NOUN\n",
      "редактор_NOUN\n",
      "газета_NOUN\n",
      "главный_ADJ\n",
      "осетинский_ADJ\n",
      "редакция_NOUN\n",
      "путин_NOUN\n",
      "издание_NOUN\n",
      "елизавета_NOUN\n",
      "увольнение_NOUN\n",
      "\n",
      "\n",
      "topic 16:\n",
      "эстония_NOUN\n",
      "таллинн_NOUN\n",
      "нефть_NOUN\n",
      "цена_NOUN\n",
      "азия_NOUN\n",
      "год_NOUN\n",
      "россия_NOUN\n",
      "дом_NOUN\n",
      "становиться_VERB\n",
      "латвия_NOUN\n",
      "\n",
      "\n",
      "topic 17:\n",
      "новый_ADJ\n",
      "украина_NOUN\n",
      "который_ADJ\n",
      "страна_NOUN\n",
      "саакашвили_NOUN\n",
      "лещенко_NOUN\n",
      "заместитель_NOUN\n",
      "виктор_NOUN\n",
      "коррупция_NOUN\n",
      "бывший_ADJ\n",
      "\n",
      "\n",
      "topic 18:\n",
      "вопрос_NOUN\n",
      "человек_NOUN\n",
      "ссср_NOUN\n",
      "линия_NOUN\n",
      "время_NOUN\n",
      "будущий_ADJ\n",
      "прямой_ADJ\n",
      "кгб_NOUN\n",
      "система_NOUN\n",
      "проблема_NOUN\n",
      "\n",
      "\n",
      "topic 19:\n",
      "российский_ADJ\n",
      "финансирование_NOUN\n",
      "работа_NOUN\n",
      "сша_NOUN\n",
      "нужный_ADJ\n",
      "место_NOUN\n",
      "сми_NOUN\n",
      "антироссийский_ADJ\n",
      "вопрос_NOUN\n",
      "оказываться_VERB\n",
      "\n",
      "\n",
      "topic 20:\n",
      "ельцин_NOUN\n",
      "центр_NOUN\n",
      "убийство_NOUN\n",
      "остров_NOUN\n",
      "ройзман_NOUN\n",
      "адвокат_NOUN\n",
      "екатеринбург_NOUN\n",
      "борис_NOUN\n",
      "период_NOUN\n",
      "масштабный_ADJ\n",
      "\n",
      "\n",
      "topic 21:\n",
      "день_NOUN\n",
      "год_NOUN\n",
      "выступать_VERB\n",
      "рождение_NOUN\n",
      "яхта_NOUN\n",
      "дело_NOUN\n",
      "леонид_NOUN\n",
      "российский_ADJ\n",
      "кадыров_NOUN\n",
      "валерий_NOUN\n",
      "\n",
      "\n",
      "topic 22:\n",
      "стиль_NOUN\n",
      "человек_NOUN\n",
      "убеждение_NOUN\n",
      "пример_NOUN\n",
      "мочь_VERB\n",
      "эксперт_NOUN\n",
      "эмоциональный_ADJ\n",
      "логика_NOUN\n",
      "рациональный_ADJ\n",
      "харизматик_NOUN\n",
      "\n",
      "\n",
      "topic 23:\n",
      "крым_NOUN\n",
      "понимать_VERB\n",
      "год_NOUN\n",
      "правый_ADJ\n",
      "бизнес_NOUN\n",
      "говорить_VERB\n",
      "борис_NOUN\n",
      "новость_NOUN\n",
      "национальный_ADJ\n",
      "малый_ADJ\n",
      "\n",
      "\n",
      "topic 24:\n",
      "россия_NOUN\n",
      "депутат_NOUN\n",
      "округ_NOUN\n",
      "единый_ADJ\n",
      "дума_NOUN\n",
      "лдпр_NOUN\n",
      "госдума_NOUN\n",
      "область_NOUN\n",
      "председатель_NOUN\n",
      "патриот_NOUN\n",
      "\n",
      "\n",
      "topic 25:\n",
      "задание_NOUN\n",
      "русский_ADJ\n",
      "рабочий_ADJ\n",
      "язык_NOUN\n",
      "мир_NOUN\n",
      "служба_NOUN\n",
      "день_NOUN\n",
      "безопасность_NOUN\n",
      "россия_NOUN\n",
      "связь_NOUN\n",
      "\n",
      "\n",
      "topic 26:\n",
      "мирить_VERB\n",
      "письмо_NOUN\n",
      "москва_NOUN\n",
      "век_NOUN\n",
      "ооо_NOUN\n",
      "авто_NOUN\n",
      "угурлиев_NOUN\n",
      "захват_NOUN\n",
      "шанс_NOUN\n",
      "рейдерский_ADJ\n",
      "\n",
      "\n",
      "topic 27:\n",
      "документ_NOUN\n",
      "список_NOUN\n",
      "должный_ADJ\n",
      "день_NOUN\n",
      "съезд_NOUN\n",
      "оставаться_VERB\n",
      "предвыборный_ADJ\n",
      "счет_NOUN\n",
      "кандидат_NOUN\n",
      "россия_NOUN\n",
      "\n",
      "\n",
      "topic 28:\n",
      "собчак_NOUN\n",
      "этот_ADJ\n",
      "она_NOUN\n",
      "фильм_NOUN\n",
      "они_NOUN\n",
      "ребенок_NOUN\n",
      "русский_ADJ\n",
      "это_NOUN\n",
      "весь_ADJ\n",
      "гора_NOUN\n",
      "\n",
      "\n",
      "topic 29:\n",
      "сша_NOUN\n",
      "обама_NOUN\n",
      "заявлять_VERB\n",
      "америка_NOUN\n",
      "американский_ADJ\n",
      "первый_ADJ\n",
      "ракета_NOUN\n",
      "объявлять_VERB\n",
      "штат_NOUN\n",
      "американец_NOUN\n",
      "\n",
      "\n",
      "topic 30:\n",
      "наука_NOUN\n",
      "исследование_NOUN\n",
      "год_NOUN\n",
      "являться_VERB\n",
      "получать_VERB\n",
      "дело_NOUN\n",
      "степень_NOUN\n",
      "раз_NOUN\n",
      "сторона_NOUN\n",
      "россия_NOUN\n",
      "\n",
      "\n",
      "topic 31:\n",
      "компания_NOUN\n",
      "уралкалий_NOUN\n",
      "акция_NOUN\n",
      "капитал_NOUN\n",
      "кредит_NOUN\n",
      "ренессанс_NOUN\n",
      "рынок_NOUN\n",
      "дмитрий_NOUN\n",
      "год_NOUN\n",
      "крупный_ADJ\n",
      "\n",
      "\n",
      "topic 32:\n",
      "класс_NOUN\n",
      "банк_NOUN\n",
      "сообщать_VERB\n",
      "математика_NOUN\n",
      "михаил_NOUN\n",
      "международный_ADJ\n",
      "финансовый_ADJ\n",
      "фсб_NOUN\n",
      "прохоров_NOUN\n",
      "ответ_NOUN\n",
      "\n",
      "\n",
      "topic 33:\n",
      "ирина_NOUN\n",
      "прохорова_NOUN\n",
      "прохоров_NOUN\n",
      "михаил_NOUN\n",
      "пытаться_VERB\n",
      "главный_ADJ\n",
      "год_NOUN\n",
      "становиться_VERB\n",
      "интервью_NOUN\n",
      "проблема_NOUN\n",
      "\n",
      "\n",
      "topic 34:\n",
      "сталин_NOUN\n",
      "прохоров_NOUN\n",
      "мир_NOUN\n",
      "человек_NOUN\n",
      "бренд_NOUN\n",
      "менеджер_NOUN\n",
      "государство_NOUN\n",
      "собственный_ADJ\n",
      "страна_NOUN\n",
      "история_NOUN\n",
      "\n",
      "\n",
      "topic 35:\n",
      "владимир_NOUN\n",
      "протест_NOUN\n",
      "путин_NOUN\n",
      "акция_NOUN\n",
      "полиция_NOUN\n",
      "президент_NOUN\n",
      "митинг_NOUN\n",
      "молдова_NOUN\n",
      "главный_ADJ\n",
      "апрель_NOUN\n",
      "\n",
      "\n",
      "topic 36:\n",
      "партия_NOUN\n",
      "россия_NOUN\n",
      "выборы_NOUN\n",
      "региональный_ADJ\n",
      "кандидат_NOUN\n",
      "округ_NOUN\n",
      "единый_ADJ\n",
      "политический_ADJ\n",
      "госдума_NOUN\n",
      "список_NOUN\n",
      "\n",
      "\n",
      "topic 37:\n",
      "строительство_NOUN\n",
      "база_NOUN\n",
      "город_NOUN\n",
      "регион_NOUN\n",
      "американский_ADJ\n",
      "мэр_NOUN\n",
      "объект_NOUN\n",
      "житель_NOUN\n",
      "должный_ADJ\n",
      "власть_NOUN\n",
      "\n",
      "\n",
      "topic 38:\n",
      "фонд_NOUN\n",
      "конкурс_NOUN\n",
      "евро_NOUN\n",
      "тысяча_NOUN\n",
      "ребенок_NOUN\n",
      "школьник_NOUN\n",
      "прохоров_NOUN\n",
      "иностранный_ADJ\n",
      "работа_NOUN\n",
      "организация_NOUN\n",
      "\n",
      "\n",
      "topic 39:\n",
      "партия_NOUN\n",
      "гражданский_ADJ\n",
      "платформа_NOUN\n",
      "выборы_NOUN\n",
      "парнас_NOUN\n",
      "суд_NOUN\n",
      "иск_NOUN\n",
      "сентябрь_NOUN\n",
      "верховный_ADJ\n",
      "цик_NOUN\n",
      "\n",
      "\n",
      "topic 40:\n",
      "политический_ADJ\n",
      "партия_NOUN\n",
      "область_NOUN\n",
      "депутат_NOUN\n",
      "отделение_NOUN\n",
      "организация_NOUN\n",
      "мандат_NOUN\n",
      "всероссийский_ADJ\n",
      "региональный_ADJ\n",
      "который_ADJ\n",
      "\n",
      "\n",
      "topic 41:\n",
      "власть_NOUN\n",
      "глава_NOUN\n",
      "польша_NOUN\n",
      "оппозиционный_ADJ\n",
      "заявлять_VERB\n",
      "слово_NOUN\n",
      "правительство_NOUN\n",
      "страна_NOUN\n",
      "право_NOUN\n",
      "демократия_NOUN\n",
      "\n",
      "\n",
      "topic 42:\n",
      "кандидат_NOUN\n",
      "избирательный_ADJ\n",
      "депутат_NOUN\n",
      "список_NOUN\n",
      "округ_NOUN\n",
      "выборы_NOUN\n",
      "комиссия_NOUN\n",
      "одномандатный_ADJ\n",
      "избиратель_NOUN\n",
      "госдума_NOUN\n",
      "\n",
      "\n",
      "topic 43:\n",
      "немцов_NOUN\n",
      "февраль_NOUN\n",
      "москва_NOUN\n",
      "квартира_NOUN\n",
      "водитель_NOUN\n",
      "ребенок_NOUN\n",
      "показание_NOUN\n",
      "мост_NOUN\n",
      "ксения_NOUN\n",
      "упоминание_NOUN\n",
      "\n",
      "\n",
      "topic 44:\n",
      "прохоров_NOUN\n",
      "компания_NOUN\n",
      "обыск_NOUN\n",
      "онэксим_NOUN\n",
      "михаил_NOUN\n",
      "источник_NOUN\n",
      "группа_NOUN\n",
      "продавать_INFN\n",
      "медиахолдинг_NOUN\n",
      "сообщать_VERB\n",
      "\n",
      "\n",
      "topic 45:\n",
      "страна_NOUN\n",
      "мигрант_NOUN\n",
      "новый_ADJ\n",
      "россия_NOUN\n",
      "становиться_VERB\n",
      "время_NOUN\n",
      "день_NOUN\n",
      "средний_ADJ\n",
      "война_NOUN\n",
      "вопрос_NOUN\n",
      "\n",
      "\n",
      "topic 46:\n",
      "становиться_INFN\n",
      "быть_INFN\n",
      "медведев_NOUN\n",
      "мечтать_VERB\n",
      "детство_NOUN\n",
      "случай_NOUN\n",
      "хороший_ADJ\n",
      "это_NOUN\n",
      "русский_ADJ\n",
      "мечта_NOUN\n",
      "\n",
      "\n",
      "topic 47:\n",
      "маска_NOUN\n",
      "технология_NOUN\n",
      "мир_NOUN\n",
      "российский_ADJ\n",
      "высокий_ADJ\n",
      "русский_ADJ\n",
      "маск_NOUN\n",
      "джобс_NOUN\n",
      "стив_NOUN\n",
      "пример_NOUN\n",
      "\n",
      "\n",
      "topic 48:\n",
      "президент_NOUN\n",
      "путин_NOUN\n",
      "время_NOUN\n",
      "россия_NOUN\n",
      "политика_NOUN\n",
      "год_NOUN\n",
      "владимир_NOUN\n",
      "слово_NOUN\n",
      "кремль_NOUN\n",
      "российский_ADJ\n",
      "\n",
      "\n",
      "topic 49:\n",
      "мама_NOUN\n",
      "ирина_NOUN\n",
      "мочь_VERB\n",
      "операция_NOUN\n",
      "любить_VERB\n",
      "врач_NOUN\n",
      "жизнь_NOUN\n",
      "басанкин_NOUN\n",
      "нога_NOUN\n",
      "леня_NOUN\n",
      "\n",
      "\n",
      "topic 50:\n",
      "отделение_NOUN\n",
      "совет_NOUN\n",
      "управление_NOUN\n",
      "федеральный_ADJ\n",
      "рубль_NOUN\n",
      "тыс_NOUN\n",
      "один_ADJ\n",
      "россия_NOUN\n",
      "глава_NOUN\n",
      "федерация_NOUN\n",
      "\n",
      "\n",
      "topic 51:\n",
      "американский_ADJ\n",
      "местный_ADJ\n",
      "илона_NOUN\n",
      "потеря_NOUN\n",
      "мочь_VERB\n",
      "год_NOUN\n",
      "бюджет_NOUN\n",
      "изменять_VERB\n",
      "день_NOUN\n",
      "мир_NOUN\n",
      "\n",
      "\n",
      "topic 52:\n",
      "екатеринбург_NOUN\n",
      "центр_NOUN\n",
      "ельцин_NOUN\n",
      "господин_NOUN\n",
      "ройзман_NOUN\n",
      "дискуссия_NOUN\n",
      "известный_ADJ\n",
      "бюджетный_ADJ\n",
      "вопрос_NOUN\n",
      "организация_NOUN\n",
      "\n",
      "\n",
      "topic 53:\n",
      "фестиваль_NOUN\n",
      "культура_NOUN\n",
      "новый_ADJ\n",
      "программа_NOUN\n",
      "проект_NOUN\n",
      "гость_NOUN\n",
      "современный_ADJ\n",
      "литературный_ADJ\n",
      "человек_NOUN\n",
      "диалог_NOUN\n",
      "\n",
      "\n",
      "topic 54:\n",
      "путин_NOUN\n",
      "президент_NOUN\n",
      "андрей_NOUN\n",
      "вопрос_NOUN\n",
      "сторонник_NOUN\n",
      "отношение_NOUN\n",
      "власть_NOUN\n",
      "политика_NOUN\n",
      "являться_VERB\n",
      "гражданский_ADJ\n",
      "\n",
      "\n",
      "topic 55:\n",
      "россиянин_NOUN\n",
      "советский_ADJ\n",
      "уровень_NOUN\n",
      "режим_NOUN\n",
      "кремль_NOUN\n",
      "публицист_NOUN\n",
      "национализм_NOUN\n",
      "поддерживать_VERB\n",
      "лидер_NOUN\n",
      "отмечать_VERB\n",
      "\n",
      "\n",
      "topic 56:\n",
      "прохоров_NOUN\n",
      "актив_NOUN\n",
      "онэксим_NOUN\n",
      "продажа_NOUN\n",
      "михаил_NOUN\n",
      "группа_NOUN\n",
      "российский_ADJ\n",
      "продавать_VERB\n",
      "переговоры_NOUN\n",
      "ведомость_NOUN\n",
      "\n",
      "\n",
      "topic 57:\n",
      "страна_NOUN\n",
      "власть_NOUN\n",
      "россия_NOUN\n",
      "отношение_NOUN\n",
      "население_NOUN\n",
      "социальный_ADJ\n",
      "давать_VERB\n",
      "который_ADJ\n",
      "гражданин_NOUN\n",
      "система_NOUN\n",
      "\n",
      "\n",
      "topic 58:\n",
      "россия_NOUN\n",
      "рост_NOUN\n",
      "партия_NOUN\n",
      "правый_ADJ\n",
      "титов_NOUN\n",
      "предприниматель_NOUN\n",
      "сила_NOUN\n",
      "дмитрий_NOUN\n",
      "экономика_NOUN\n",
      "бизнес_NOUN\n",
      "\n",
      "\n",
      "topic 59:\n",
      "маск_NOUN\n",
      "россия_NOUN\n",
      "инженер_NOUN\n",
      "враг_NOUN\n",
      "улюкаев_NOUN\n",
      "экономический_ADJ\n",
      "энергетика_NOUN\n",
      "система_NOUN\n",
      "мир_NOUN\n",
      "солнечный_ADJ\n",
      "\n",
      "\n",
      "topic 60:\n",
      "место_NOUN\n",
      "бизнес_NOUN\n",
      "первый_ADJ\n",
      "рейтинг_NOUN\n",
      "олег_NOUN\n",
      "россия_NOUN\n",
      "секрет_NOUN\n",
      "президент_NOUN\n",
      "экономический_ADJ\n",
      "кавджарадзе_NOUN\n",
      "\n",
      "\n",
      "topic 61:\n",
      "руб_NOUN\n",
      "рассказывать_VERB\n",
      "бизнес_NOUN\n",
      "линия_NOUN\n",
      "компания_NOUN\n",
      "становиться_VERB\n",
      "крупный_ADJ\n",
      "человек_NOUN\n",
      "журнал_NOUN\n",
      "искать_VERB\n",
      "\n",
      "\n",
      "topic 62:\n",
      "учить_VERB\n",
      "попадать_INFN\n",
      "лекция_NOUN\n",
      "занятие_NOUN\n",
      "клуб_NOUN\n",
      "проходить_VERB\n",
      "адрес_NOUN\n",
      "обучение_NOUN\n",
      "телефон_NOUN\n",
      "искусство_NOUN\n",
      "\n",
      "\n",
      "topic 63:\n",
      "польский_ADJ\n",
      "польша_NOUN\n",
      "министр_NOUN\n",
      "правительство_NOUN\n",
      "оборона_NOUN\n",
      "памятник_NOUN\n",
      "год_NOUN\n",
      "армия_NOUN\n",
      "мэр_NOUN\n",
      "варшава_NOUN\n",
      "\n",
      "\n",
      "topic 64:\n",
      "украинский_ADJ\n",
      "украина_NOUN\n",
      "польский_ADJ\n",
      "польша_NOUN\n",
      "геноцид_NOUN\n",
      "память_NOUN\n",
      "июль_NOUN\n",
      "жертва_NOUN\n",
      "упа_NOUN\n",
      "поляков_NOUN\n",
      "\n",
      "\n",
      "topic 65:\n",
      "этот_ADJ\n",
      "человек_NOUN\n",
      "иметь_VERB\n",
      "они_NOUN\n",
      "она_NOUN\n",
      "все_NOUN\n",
      "говорить_VERB\n",
      "такой_ADJ\n",
      "дело_NOUN\n",
      "должный_ADJ\n",
      "\n",
      "\n",
      "topic 66:\n",
      "олигарх_NOUN\n",
      "завод_NOUN\n",
      "кризис_NOUN\n",
      "норильский_ADJ\n",
      "последний_ADJ\n",
      "жизнь_NOUN\n",
      "потанин_NOUN\n",
      "производство_NOUN\n",
      "финансовый_ADJ\n",
      "прохоров_NOUN\n",
      "\n",
      "\n",
      "topic 67:\n",
      "депутат_NOUN\n",
      "дело_NOUN\n",
      "политический_ADJ\n",
      "год_NOUN\n",
      "прокуратура_NOUN\n",
      "следствие_NOUN\n",
      "судья_NOUN\n",
      "заседание_NOUN\n",
      "союз_NOUN\n",
      "общественный_ADJ\n",
      "\n",
      "\n",
      "topic 68:\n",
      "мочь_VERB\n",
      "дело_NOUN\n",
      "сми_NOUN\n",
      "статья_NOUN\n",
      "написать_VERB\n",
      "знать_VERB\n",
      "путин_NOUN\n",
      "мошенничество_NOUN\n",
      "президент_NOUN\n",
      "быть_INFN\n",
      "\n",
      "\n",
      "topic 69:\n",
      "история_NOUN\n",
      "человек_NOUN\n",
      "участник_NOUN\n",
      "дом_NOUN\n",
      "мемориал_NOUN\n",
      "кино_NOUN\n",
      "библиотека_NOUN\n",
      "исторический_ADJ\n",
      "георгиевский_ADJ\n",
      "ленточка_NOUN\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "for j in range(70):\n",
    "    print('topic %d:' % j)\n",
    "    for i in topic_words[j]:\n",
    "        print i\n",
    "    print('\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "### Ну и как то так я пробовал подобрать наилучшее значение метрики, только сравнивал разные"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "a1 = []\n",
    "a2 = []\n",
    "a3 = []\n",
    "b = []\n",
    "k = -0.1\n",
    "for i in range(1):\n",
    "    k = k - 0.1\n",
    "    batch_vectorizer = None\n",
    "    if len(glob.glob(os.path.join('vk', '*.batch'))) < 1:\n",
    "        batch_vectorizer = artm.BatchVectorizer(data_path='', data_format='bow_uci', collection_name='vk', target_folder='vk')\n",
    "    else:\n",
    "        batch_vectorizer = artm.BatchVectorizer(data_path='vk', data_format='batches')\n",
    "\n",
    "    dictionary = artm.Dictionary()\n",
    "\n",
    "    model_artm = artm.ARTM(topic_names=['topic_{}'.format(i) for i in xrange(50)], #\n",
    "                           scores=[artm.PerplexityScore(name='PerplexityScore',\n",
    "                                                        dictionary=dictionary)],\n",
    "                           regularizers=[artm.SmoothSparseThetaRegularizer(name='SparseTheta', tau=-1.85)], #1\n",
    "                           cache_theta=True)\n",
    "\n",
    "    if not os.path.isfile('vk/dictionary.dict'):\n",
    "        dictionary.gather(data_path=batch_vectorizer.data_path)\n",
    "        dictionary.save(dictionary_path='vk/dictionary.dict')\n",
    "\n",
    "    dictionary.load(dictionary_path='vk/dictionary.dict')\n",
    "\n",
    "    model_artm.initialize(dictionary=dictionary)\n",
    "\n",
    "    model_artm.scores.add(artm.SparsityPhiScore(name='SparsityPhiScore'))\n",
    "    model_artm.scores.add(artm.SparsityThetaScore(name='SparsityThetaScore'))\n",
    "    model_artm.scores.add(artm.TopicKernelScore(name='TopicKernelScore', probability_mass_threshold=0.1)) #3\n",
    "\n",
    "    model_artm.regularizers.add(artm.SmoothSparsePhiRegularizer(name='SparsePhi', tau=-0.2))\n",
    "    model_artm.regularizers.add(artm.DecorrelatorPhiRegularizer(name='DecorrelatorPhi', tau=2.5e+5))\n",
    "  \n",
    "    model_artm.num_document_passes = 1 #4\n",
    "\n",
    "    model_artm.fit_offline(batch_vectorizer=batch_vectorizer, num_collection_passes=25) #5\n",
    "\n",
    "    a1.append(model_artm.score_tracker['TopicKernelScore'].last_average_contrast)\n",
    "    a2.append(model_artm.score_tracker['TopicKernelScore'].last_average_purity)\n",
    "    a3.append(model_artm.score_tracker['PerplexityScore'].last_value)\n",
    "    b.append(k)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 169,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x81677160>]"
      ]
     },
     "execution_count": 169,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAhQAAAFkCAYAAAB4sKK5AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x81677d68>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(b,a2)\n",
    "plt.plot(b,a1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              6001      6002      6003  6004      6005      6006      6007  \\\n",
      "topic_0   0.000000  0.003775  0.003775   0.0  0.000000  0.003775  0.003770   \n",
      "topic_1   0.000000  0.000094  0.000094   0.0  0.000000  0.000094  0.000093   \n",
      "topic_2   0.000000  0.000000  0.000000   0.0  0.000000  0.000000  0.000000   \n",
      "topic_3   0.000000  0.205269  0.205269   0.0  0.000000  0.205269  0.204997   \n",
      "topic_4   0.000000  0.000000  0.000000   0.0  0.029555  0.000000  0.000000   \n",
      "topic_5   0.000000  0.000000  0.000000   0.0  0.003082  0.000000  0.000000   \n",
      "topic_6   0.000000  0.009181  0.009181   0.0  0.005251  0.009181  0.009169   \n",
      "topic_7   0.000000  0.000000  0.000000   0.0  0.000000  0.000000  0.000000   \n",
      "topic_8   0.000000  0.008044  0.008044   0.0  0.002699  0.008044  0.008034   \n",
      "topic_9   0.000000  0.000000  0.000000   0.0  0.000000  0.000000  0.000000   \n",
      "topic_10  0.000000  0.000000  0.000000   0.0  0.001659  0.000000  0.000000   \n",
      "topic_11  0.000000  0.001246  0.001246   0.0  0.001221  0.001246  0.001245   \n",
      "topic_12  0.000000  0.000000  0.000000   0.0  0.000000  0.000000  0.000000   \n",
      "topic_13  0.000000  0.000000  0.000000   0.0  0.000000  0.000000  0.000000   \n",
      "topic_14  0.000000  0.003924  0.003924   0.0  0.000594  0.003924  0.003919   \n",
      "topic_15  0.000000  0.000417  0.000417   0.0  0.000000  0.000417  0.000416   \n",
      "topic_16  0.000000  0.004869  0.004869   0.0  0.022348  0.004869  0.004863   \n",
      "topic_17  0.000000  0.000000  0.000000   0.0  0.000407  0.000000  0.000000   \n",
      "topic_18  0.000000  0.001987  0.001987   0.0  0.001356  0.001987  0.001984   \n",
      "topic_19  0.000000  0.001814  0.001814   0.0  0.004370  0.001814  0.001812   \n",
      "topic_20  0.000000  0.000000  0.000000   0.0  0.000890  0.000000  0.000000   \n",
      "topic_21  0.000000  0.001439  0.001439   0.0  0.000807  0.001439  0.001437   \n",
      "topic_22  0.000000  0.000000  0.000000   0.0  0.744047  0.000000  0.000000   \n",
      "topic_23  0.000000  0.000213  0.000213   0.0  0.000299  0.000213  0.000213   \n",
      "topic_24  0.000000  0.001017  0.001017   0.0  0.000000  0.001017  0.001015   \n",
      "topic_25  0.000000  0.000000  0.000000   0.0  0.014927  0.000000  0.000000   \n",
      "topic_26  0.000000  0.000000  0.000000   0.0  0.002930  0.000000  0.000000   \n",
      "topic_27  0.000000  0.000000  0.000000   0.0  0.000000  0.000000  0.000000   \n",
      "topic_28  0.000000  0.000000  0.000000   0.0  0.000000  0.000000  0.000000   \n",
      "topic_29  0.000000  0.000000  0.000000   0.0  0.006838  0.000000  0.000000   \n",
      "...            ...       ...       ...   ...       ...       ...       ...   \n",
      "topic_40  0.000000  0.000000  0.000000   0.0  0.001585  0.000000  0.000000   \n",
      "topic_41  0.000000  0.000000  0.000000   0.0  0.004632  0.000000  0.000000   \n",
      "topic_42  0.000000  0.003318  0.003318   0.0  0.000000  0.003318  0.003313   \n",
      "topic_43  0.000000  0.003698  0.003698   1.0  0.002838  0.003698  0.003693   \n",
      "topic_44  0.133461  0.000000  0.000000   0.0  0.000734  0.000000  0.000000   \n",
      "topic_45  0.000000  0.000000  0.000000   0.0  0.000000  0.000000  0.000000   \n",
      "topic_46  0.000000  0.000000  0.000000   0.0  0.038124  0.000000  0.000000   \n",
      "topic_47  0.000000  0.000000  0.000000   0.0  0.001469  0.000000  0.000000   \n",
      "topic_48  0.000000  0.000000  0.000000   0.0  0.004742  0.000000  0.000000   \n",
      "topic_49  0.000000  0.223864  0.223864   0.0  0.000000  0.223864  0.223567   \n",
      "topic_50  0.000000  0.275642  0.275642   0.0  0.000000  0.275642  0.275277   \n",
      "topic_51  0.000000  0.000000  0.000000   0.0  0.004372  0.000000  0.000000   \n",
      "topic_52  0.000000  0.000000  0.000000   0.0  0.000000  0.000000  0.000000   \n",
      "topic_53  0.000000  0.002709  0.002709   0.0  0.005576  0.002709  0.002722   \n",
      "topic_54  0.000000  0.001072  0.001072   0.0  0.000000  0.001072  0.001070   \n",
      "topic_55  0.000000  0.001383  0.001383   0.0  0.002458  0.001383  0.001381   \n",
      "topic_56  0.000000  0.031322  0.031322   0.0  0.025428  0.031322  0.031352   \n",
      "topic_57  0.000000  0.000000  0.000000   0.0  0.000000  0.000000  0.000000   \n",
      "topic_58  0.000000  0.000825  0.000825   0.0  0.000000  0.000825  0.000824   \n",
      "topic_59  0.000000  0.001480  0.001480   0.0  0.000000  0.001480  0.001478   \n",
      "topic_60  0.000000  0.000000  0.000000   0.0  0.000000  0.000000  0.000000   \n",
      "topic_61  0.000000  0.003966  0.003966   0.0  0.000000  0.003966  0.003960   \n",
      "topic_62  0.000000  0.000000  0.000000   0.0  0.000000  0.000000  0.000000   \n",
      "topic_63  0.000000  0.000000  0.000000   0.0  0.000000  0.000000  0.000000   \n",
      "topic_64  0.000000  0.000098  0.000098   0.0  0.014554  0.000098  0.000098   \n",
      "topic_65  0.000000  0.000000  0.000000   0.0  0.000000  0.000000  0.000000   \n",
      "topic_66  0.000000  0.001568  0.001568   0.0  0.000807  0.001568  0.001566   \n",
      "topic_67  0.000000  0.000000  0.000000   0.0  0.000000  0.000000  0.000000   \n",
      "topic_68  0.000000  0.016062  0.016062   0.0  0.000000  0.016062  0.016041   \n",
      "topic_69  0.866538  0.000000  0.000000   0.0  0.000823  0.000000  0.000000   \n",
      "\n",
      "              6008      6009      6010  ...   2991  2992  2993  2994  2995  \\\n",
      "topic_0   0.003775  0.000000  0.003775  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_1   0.000094  0.000000  0.000094  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_2   0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_3   0.205269  0.000000  0.205269  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_4   0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_5   0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_6   0.009181  0.000000  0.009181  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_7   0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_8   0.008044  0.458185  0.008044  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_9   0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_10  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_11  0.001246  0.000000  0.001246  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_12  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_13  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_14  0.003924  0.000000  0.003924  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_15  0.000417  0.000000  0.000417  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_16  0.004869  0.000000  0.004869  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_17  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_18  0.001987  0.000000  0.001987  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_19  0.001814  0.000000  0.001814  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_20  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_21  0.001439  0.000000  0.001439  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_22  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_23  0.000213  0.000000  0.000213  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_24  0.001017  0.000000  0.001017  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_25  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_26  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_27  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_28  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_29  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "...            ...       ...       ...  ...    ...   ...   ...   ...   ...   \n",
      "topic_40  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_41  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_42  0.003318  0.000000  0.003318  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_43  0.003698  0.000000  0.003698  ...    0.0   0.0   1.0   0.0   0.0   \n",
      "topic_44  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_45  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_46  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_47  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_48  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_49  0.223864  0.000000  0.223864  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_50  0.275642  0.000000  0.275642  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_51  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_52  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_53  0.002709  0.000000  0.002709  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_54  0.001072  0.000000  0.001072  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_55  0.001383  0.000000  0.001383  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_56  0.031322  0.000000  0.031322  ...    0.0   1.0   0.0   1.0   1.0   \n",
      "topic_57  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_58  0.000825  0.000000  0.000825  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_59  0.001480  0.000000  0.001480  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_60  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_61  0.003966  0.000000  0.003966  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_62  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_63  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_64  0.000098  0.000000  0.000098  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_65  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_66  0.001568  0.000000  0.001568  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_67  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_68  0.016062  0.000000  0.016062  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "topic_69  0.000000  0.000000  0.000000  ...    0.0   0.0   0.0   0.0   0.0   \n",
      "\n",
      "          2996  2997  2998  2999  3000  \n",
      "topic_0    0.0   0.0   0.0   0.0   0.0  \n",
      "topic_1    0.0   0.0   0.0   0.0   0.0  \n",
      "topic_2    0.0   0.0   0.0   0.0   0.0  \n",
      "topic_3    0.0   0.0   0.0   0.0   0.0  \n",
      "topic_4    0.0   0.0   0.0   0.0   0.0  \n",
      "topic_5    0.0   0.0   0.0   0.0   0.0  \n",
      "topic_6    0.0   0.0   0.0   0.0   0.0  \n",
      "topic_7    0.0   0.0   0.0   0.0   0.0  \n",
      "topic_8    0.0   0.0   0.0   0.0   0.0  \n",
      "topic_9    0.0   0.0   0.0   0.0   0.0  \n",
      "topic_10   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_11   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_12   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_13   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_14   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_15   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_16   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_17   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_18   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_19   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_20   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_21   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_22   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_23   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_24   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_25   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_26   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_27   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_28   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_29   0.0   0.0   0.0   0.0   0.0  \n",
      "...        ...   ...   ...   ...   ...  \n",
      "topic_40   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_41   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_42   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_43   1.0   1.0   1.0   1.0   1.0  \n",
      "topic_44   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_45   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_46   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_47   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_48   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_49   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_50   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_51   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_52   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_53   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_54   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_55   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_56   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_57   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_58   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_59   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_60   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_61   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_62   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_63   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_64   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_65   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_66   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_67   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_68   0.0   0.0   0.0   0.0   0.0  \n",
      "topic_69   0.0   0.0   0.0   0.0   0.0  \n",
      "\n",
      "[70 rows x 10000 columns]\n"
     ]
    }
   ],
   "source": [
    "theta_matrix = model_artm.get_theta()\n",
    "print theta_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "name: \"nwt\"\n",
      "type: \"class artm::core::DensePhiMatrix\"\n",
      "num_topics: 70\n",
      "num_tokens: 44248\n",
      "byte_size: 16335816\n",
      "\n",
      "name: \"pwt\"\n",
      "type: \"class artm::core::AttachedPhiMatrix\"\n",
      "num_topics: 70\n",
      "num_tokens: 44248\n",
      "byte_size: 0\n",
      "\n"
     ]
    }
   ],
   "source": [
    "theta_matrix = model_artm.get_theta()\n",
    "print theta_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [conda root]",
   "language": "python",
   "name": "conda-root-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}