doc/source/topic_modeling.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Topic modeling\n",
    "\n",
    "The [topicmod module](api.rst#module-tmtoolkit.topicmod) offers a wide range of tools to facilitate [topic modeling](https://cacm.acm.org/magazines/2012/4/147361-probabilistic-topic-models/fulltext) with Python. This chapter will introduce the following techniques: \n",
    "\n",
    "- [parallel topic model computation for different copora and/or parameter sets](#Computing-topic-models-in-parallel)\n",
    "- [evaluation of topic models (including finding a good set of hyperparameters for the given dataset)](#Evaluation-of-topic-models)\n",
    "- [common statistics and tools for topic models](#Common-statistics-and-tools-for-topic-models)\n",
    "- [export of topic models and summaries to different file formats](#Displaying-and-exporting-topic-modeling-results)\n",
    "- [visualization of topic models](#Visualizing-topic-models)\n",
    "\n",
    "A quick note on terminology: So far, we spoke about *tokens* or sometimes *terms* when we meant the individual elements that our documents consist of after we applied text preprocessing such as *tokenization* to the raw input text strings. These tokens can be lexicographically correct words, but they don't have to, e.g. when you applied stemming you might have tokens like \"argu\" in your vocabulary. There may also be numbers or punctuation symbols in your vocabulary. For those topic modeling techniques that tmtoolkit supports, the results are always two probability distributions: a *document-topic distribution* and a *topic-word distribution*. Since the latter is called topic-*word* and not topic-*token* or *-term* distribution, we will also use the term *word* when we mean any token from the corpus' vocabulary.\n",
    "\n",
    "\n",
    "## An example document-term matrix\n",
    "\n",
    "tmtoolkit supports topic models that are computed from document-term matrices (DTMs). Just as in the previous chapter, we will at first generate a DTM. However, this time the sample will be bigger:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:53:27.240220Z",
     "iopub.status.busy": "2022-02-08T06:53:27.239366Z",
     "iopub.status.idle": "2022-02-08T06:53:37.131808Z",
     "shell.execute_reply": "2022-02-08T06:53:37.131392Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-02-08 07:53:29,472:INFO:tmtoolkit:creating Corpus instance with no documents\n",
      "2022-02-08 07:53:29,473:INFO:tmtoolkit:using serial processing\n",
      "2022-02-08 07:53:29,957:INFO:tmtoolkit:sampling 100 documents(s) out of 3824\n",
      "2022-02-08 07:53:29,958:INFO:tmtoolkit:adding text from 100 documents(s)\n",
      "2022-02-08 07:53:29,959:INFO:tmtoolkit:running NLP pipeline on 100 documents\n",
      "2022-02-08 07:53:37,126:INFO:tmtoolkit:generating document texts\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Corpus with 100 documents in English\n",
      "> NewsArticles-113 (1071 tokens): Use talk not tech to tame your children 's online ...\n",
      "> NewsArticles-1043 (270 tokens): Burhan Ozbilici wins 2017 World Press Photo compet...\n",
      "> NewsArticles-1032 (653 tokens): Germany 's right - wing AfD seeks to expel state l...\n",
      "> NewsArticles-104 (31 tokens): Your pictures : Broken resolutions    Each week , ...\n",
      "> NewsArticles-1137 (226 tokens): These Cool New ' Vertical Forest ' Skyscrapers Are...\n",
      "> NewsArticles-1036 (835 tokens): Amnesty accuses Tunisian authorities of torture ah...\n",
      "> NewsArticles-1048 (476 tokens): Espirito Santo police return to work after murder ...\n",
      "> NewsArticles-1126 (163 tokens): This Makeup Palette Has A Game - Changing Little S...\n",
      "> NewsArticles-1090 (1291 tokens): Martin challenges Fitzgerald over Tusla informatio...\n",
      "> NewsArticles-1141 (914 tokens): Aslef members reject Southern rail deal    Aslef m...\n",
      "(and 90 more documents)\n",
      "total number of tokens: 66637 / vocabulary size: 9469\n"
     ]
    }
   ],
   "source": [
    "import random\n",
    "random.seed(20191120)   # to make the sampling reproducible\n",
    "\n",
    "import numpy as np\n",
    "np.set_printoptions(precision=5)\n",
    "\n",
    "from tmtoolkit.utils import enable_logging\n",
    "enable_logging()\n",
    "\n",
    "from tmtoolkit.corpus import Corpus, print_summary\n",
    "\n",
    "\n",
    "corp = Corpus.from_builtin_corpus('en-NewsArticles', sample=100)\n",
    "print_summary(corp)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We will also now generate two DTMs, because we later want to show how you can compute topic models for two different DTMs in parallel. At first, we to some general preprocessing."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:53:37.136485Z",
     "iopub.status.busy": "2022-02-08T06:53:37.135873Z",
     "iopub.status.idle": "2022-02-08T06:53:37.223409Z",
     "shell.execute_reply": "2022-02-08T06:53:37.222931Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-02-08 07:53:37,165:INFO:tmtoolkit:replacing 2186 token hashes\n",
      "2022-02-08 07:53:37,202:INFO:tmtoolkit:replacing 502 token hashes\n",
      "2022-02-08 07:53:37,217:INFO:tmtoolkit:generating document texts\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Corpus with 100 documents in English\n",
      "> NewsArticles-113 (1071 tokens): use talk not tech to tame your child s online habi...\n",
      "> NewsArticles-1043 (270 tokens): burhan ozbilici win 2017 world press photo competi...\n",
      "> NewsArticles-1032 (653 tokens): germany s right  wing afd seek to expel state lead...\n",
      "> NewsArticles-104 (31 tokens): your picture  break resolution  each week  we publ...\n",
      "> NewsArticles-1137 (226 tokens): these cool new  vertical forest  skyscraper be des...\n",
      "> NewsArticles-1036 (835 tokens): amnesty accuse tunisian authority of torture ahead...\n",
      "> NewsArticles-1048 (476 tokens): espirito santo police return to work after murder ...\n",
      "> NewsArticles-1126 (163 tokens): this makeup palette have a game  change little sec...\n",
      "> NewsArticles-1090 (1291 tokens): martin challenge fitzgerald over tusla information...\n",
      "> NewsArticles-1141 (914 tokens): aslef member reject southern rail deal  aslef memb...\n",
      "(and 90 more documents)\n",
      "total number of tokens: 66637 / vocabulary size: 6758\n"
     ]
    }
   ],
   "source": [
    "from tmtoolkit.corpus import lemmatize, to_lowercase, remove_punctuation\n",
    "\n",
    "lemmatize(corp)\n",
    "to_lowercase(corp)\n",
    "remove_punctuation(corp)\n",
    "\n",
    "print_summary(corp)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Check if there are a few odd, unprintable characters in any tokens:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:53:37.236351Z",
     "iopub.status.busy": "2022-02-08T06:53:37.235491Z",
     "iopub.status.idle": "2022-02-08T06:53:37.238550Z",
     "shell.execute_reply": "2022-02-08T06:53:37.239189Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{('\\xa0', b'\\xc2\\xa0'),\n",
       " ('à', b'\\xc3\\xa0'),\n",
       " ('ó', b'\\xc3\\xb3'),\n",
       " ('™', b'\\xe2\\x84\\xa2'),\n",
       " ('�', b'\\xef\\xbf\\xbd')}"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import string\n",
    "from tmtoolkit.corpus import corpus_unique_chars\n",
    "\n",
    "{(c, c.encode('utf-8')) for c in corpus_unique_chars(corp) if c not in string.printable}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Remove all of them but \"à\" and \"ó\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:53:37.264242Z",
     "iopub.status.busy": "2022-02-08T06:53:37.263481Z",
     "iopub.status.idle": "2022-02-08T06:53:37.288213Z",
     "shell.execute_reply": "2022-02-08T06:53:37.287797Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-02-08 07:53:37,262:INFO:tmtoolkit:replacing 3 token hashes\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{('à', b'\\xc3\\xa0'), ('ó', b'\\xc3\\xb3')}"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.corpus import remove_chars\n",
    "\n",
    "unprintable_bytes = {b'\\xc2\\xa0', b'\\xe2\\x84\\xa2', b'\\xef\\xbf\\xbd'}\n",
    "unprintable_chars = set(map(lambda b: b.decode('utf-8'), unprintable_bytes))\n",
    "remove_chars(corp, unprintable_chars)\n",
    "\n",
    "# check again\n",
    "{(c, c.encode('utf-8')) for c in corpus_unique_chars(corp) if c not in string.printable}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we at first apply more \"relaxed\" cleaning:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:53:37.327032Z",
     "iopub.status.busy": "2022-02-08T06:53:37.326338Z",
     "iopub.status.idle": "2022-02-08T06:53:37.666910Z",
     "shell.execute_reply": "2022-02-08T06:53:37.666486Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-02-08 07:53:37,327:INFO:tmtoolkit:creating Corpus instance with no documents\n",
      "2022-02-08 07:53:37,327:INFO:tmtoolkit:using serial processing\n",
      "2022-02-08 07:53:37,447:INFO:tmtoolkit:filtered tokens by mask: num. tokens was 66637 and is now 30407\n",
      "2022-02-08 07:53:37,534:INFO:tmtoolkit:filtered tokens by mask: num. tokens was 30407 and is now 30407\n",
      "2022-02-08 07:53:37,626:INFO:tmtoolkit:filtered tokens by mask: num. tokens was 30407 and is now 16443\n",
      "2022-02-08 07:53:37,662:INFO:tmtoolkit:generating document texts\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Corpus with 100 documents in English\n",
      "> NewsArticles-113 (256 tokens): use talk child online like parent house like happy...\n",
      "> NewsArticles-1043 (64 tokens): win 2017 world press win 2017 world press image ru...\n",
      "> NewsArticles-1032 (158 tokens): germany right seek state leader leader germany ask...\n",
      "> NewsArticles-104 (7 tokens): break week publish set theme week break\n",
      "> NewsArticles-1137 (43 tokens): new design help fight world need dont china kind d...\n",
      "> NewsArticles-1036 (241 tokens): accuse authority ahead key talk germany right grou...\n",
      "> NewsArticles-1048 (141 tokens): police return work murder officer return work stat...\n",
      "> NewsArticles-1126 (19 tokens): game change little product reveal feature game new...\n",
      "> NewsArticles-1090 (266 tokens): challenge information leader claim minister justic...\n",
      "> NewsArticles-1141 (248 tokens): member reject southern deal member reject deal sou...\n",
      "(and 90 more documents)\n",
      "total number of tokens: 16443 / vocabulary size: 791\n"
     ]
    }
   ],
   "source": [
    "from copy import copy\n",
    "from tmtoolkit.corpus import filter_clean_tokens, remove_common_tokens, remove_uncommon_tokens\n",
    "\n",
    "corp_bigger = copy(corp)\n",
    "\n",
    "filter_clean_tokens(corp_bigger, remove_shorter_than=2)\n",
    "remove_common_tokens(corp_bigger, df_threshold=0.85)\n",
    "remove_uncommon_tokens(corp_bigger, df_threshold=0.05)\n",
    "\n",
    "print_summary(corp_bigger)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Another copy of `corp` will apply more aggressive cleaning and hence will result in a smaller vocabulary size:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:53:37.694801Z",
     "iopub.status.busy": "2022-02-08T06:53:37.684519Z",
     "iopub.status.idle": "2022-02-08T06:53:38.018637Z",
     "shell.execute_reply": "2022-02-08T06:53:38.019031Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-02-08 07:53:37,706:INFO:tmtoolkit:creating Corpus instance with no documents\n",
      "2022-02-08 07:53:37,706:INFO:tmtoolkit:using serial processing\n",
      "2022-02-08 07:53:37,814:INFO:tmtoolkit:filtered tokens by mask: num. tokens was 66637 and is now 19002\n",
      "2022-02-08 07:53:37,868:INFO:tmtoolkit:filtered tokens by mask: num. tokens was 19002 and is now 18551\n",
      "2022-02-08 07:53:37,939:INFO:tmtoolkit:filtered tokens by mask: num. tokens was 18551 and is now 18551\n",
      "2022-02-08 07:53:38,000:INFO:tmtoolkit:filtered tokens by mask: num. tokens was 18551 and is now 5065\n",
      "2022-02-08 07:53:38,014:INFO:tmtoolkit:generating document texts\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Corpus with 100 documents in English\n",
      "> NewsArticles-113 (69 tokens): child house service house service month time child...\n",
      "> NewsArticles-1043 (18 tokens): world press world press world press russia year ne...\n",
      "> NewsArticles-1032 (58 tokens): germany state leader leader germany state party le...\n",
      "> NewsArticles-104 (2 tokens): week week\n",
      "> NewsArticles-1137 (12 tokens): new world china press day area china house office ...\n",
      "> NewsArticles-1036 (91 tokens): authority germany right group security official mi...\n",
      "> NewsArticles-1048 (59 tokens): police work officer work state people day police o...\n",
      "> NewsArticles-1126 (3 tokens): product way twitter\n",
      "> NewsArticles-1090 (103 tokens): information leader minister child prime time week ...\n",
      "> NewsArticles-1141 (52 tokens): member member member member secretary decision mem...\n",
      "(and 90 more documents)\n",
      "total number of tokens: 5065 / vocabulary size: 141\n"
     ]
    }
   ],
   "source": [
    "from tmtoolkit.corpus import filter_for_pos\n",
    "corp_smaller = copy(corp)\n",
    "\n",
    "filter_for_pos(corp_smaller, 'N')\n",
    "filter_clean_tokens(corp_smaller, remove_shorter_than=2)\n",
    "remove_common_tokens(corp_smaller, df_threshold=0.8)\n",
    "remove_uncommon_tokens(corp_smaller, df_threshold=0.1)\n",
    "\n",
    "del corp   # remove original corpus\n",
    "\n",
    "print_summary(corp_smaller)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We will create the document labels, vocabulary arrays and DTMs for both versions now:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:53:38.031425Z",
     "iopub.status.busy": "2022-02-08T06:53:38.030927Z",
     "iopub.status.idle": "2022-02-08T06:53:38.080777Z",
     "shell.execute_reply": "2022-02-08T06:53:38.080302Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-02-08 07:53:38,029:INFO:tmtoolkit:generating sparse DTM with 100 documents and vocab size 791\n",
      "2022-02-08 07:53:38,063:INFO:tmtoolkit:generating sparse DTM with 100 documents and vocab size 141\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(<100x791 sparse matrix of type '<class 'numpy.int32'>'\n",
       " \twith 9247 stored elements in Compressed Sparse Row format>,\n",
       " <100x141 sparse matrix of type '<class 'numpy.int32'>'\n",
       " \twith 2378 stored elements in Compressed Sparse Row format>)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.corpus import dtm\n",
    "\n",
    "dtm_bg, doc_labels_bg, vocab_bg = dtm(corp_bigger, return_doc_labels=True, return_vocab=True)\n",
    "dtm_sm, doc_labels_sm, vocab_sm = dtm(corp_smaller, return_doc_labels=True, return_vocab=True)\n",
    "\n",
    "del corp_bigger, corp_smaller  # don't need these any more\n",
    "\n",
    "dtm_bg, dtm_sm"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We now have two sparse DTMs `dtm_bg` (from the bigger preprocessed data) and `dtm_sm` (from the smaller preprocessed data), a list of document labels `doc_labels` that represent the rows of both DTMs and vocabulary arrays `vocab_bg` and `vocab_sm` that represent the columns of the respective DTMs. We will use this data for the remainder of the chapter."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Computing topic models in parallel\n",
    "\n",
    "tmtoolkit allows to compute topic models in parallel, making use of all processor cores in your machine. Parallelization can be done per input DTM, per hyperparameter set and as combination of both. Hyperparameters control the number of topics and their \"granularity\". We will later have a look at the role of hyperparameters and how to find an optimal combination for a given dataset with the means of topic model evaluation.\n",
    "\n",
    "For now, we will concentrate on computing the topic models for both of our two DTMs in parallel. tmtoolkit supports three very popular packages for topic modeling, which provide the work of actually computing the model from the input matrix. They can all be accessed in separate sub-modules of the [topicmod module](api.rst#module-tmtoolkit.topicmod):\n",
    "\n",
    "- [topicmod.tm_lda](api.rst#module-tmtoolkit.topicmod.tm_lda) provides an interface for the [lda](https://lda.readthedocs.io/en/latest/) package\n",
    "- [topicmod.tm_sklearn](api.rst#module-tmtoolkit.topicmod.tm_sklearn) provides an interface for the [scikit-learn](http://scikit-learn.org/stable/modules/generated/sklearn.decomposition.LatentDirichletAllocation.html) package\n",
    "- [topicmod.tm_gensim](api.rst#module-tmtoolkit.topicmod.tm_gensim) provides an interface for the [Gensim](https://radimrehurek.com/gensim/) package\n",
    "\n",
    "Each of these sub-modules offer at least two functions that work with the respective package: `compute_models_parallel` for general parallel model computation and `evaluate_topic_models` for parallel model computation and evaluation (discussed later). For now, we want to compute two models in parallel with the [lda](https://lda.readthedocs.io/en/latest/) package and hence use [compute_models_parallel](api.rst#tmtoolkit.topicmod.tm_lda.compute_models_parallel) from [topicmod.tm_lda](api.rst#module-tmtoolkit.topicmod.tm_lda).\n",
    "\n",
    "We need to provide two things for this function: First, the input matrices as a dict that maps labels to the respective DTMs. Second, hyperparameters to use for the model computations. Note that each topic modeling package has different hyperparameters and you should refer to their documentation in order to find out which hyperparameters you need to provide. For lda, we set the number of topics `n_topics` to 10 and the number of iterations for the Gibbs sampling process `n_iter` to 1000. We always want to use the same hyperparameters, so we pass these as `constant_parameters`. If we wanted to create models for a whole range of parameters, e.g. for different numbers of topics, we could provide `varying_parameters`. We will check this out later when we evaluate topic models.\n",
    "\n",
    "<div class=\"alert alert-info\">\n",
    "    \n",
    "**Note**\n",
    "\n",
    "For proper topic modeling, we shouldn't just set the number of topics, but try to find it out via evaluation methods. We should also check if the algorithm converged using the provided likelihood estimations. We will do both later on, but now focus on `compute_models_parallel`.\n",
    "\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:53:38.086201Z",
     "iopub.status.busy": "2022-02-08T06:53:38.085717Z",
     "iopub.status.idle": "2022-02-08T06:53:42.728984Z",
     "shell.execute_reply": "2022-02-08T06:53:42.729561Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "defaultdict(list,\n",
       "            {'smaller': [({'n_topics': 10,\n",
       "                'n_iter': 1000,\n",
       "                'random_state': 20191122},\n",
       "               <lda.lda.LDA at 0x7fb69dd0d580>)],\n",
       "             'bigger': [({'n_topics': 10,\n",
       "                'n_iter': 1000,\n",
       "                'random_state': 20191122},\n",
       "               <lda.lda.LDA at 0x7fb69dd0d460>)]})"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import logging\n",
    "import warnings\n",
    "from tmtoolkit.utils import disable_logging\n",
    "from tmtoolkit.topicmod.tm_lda import compute_models_parallel\n",
    "\n",
    "# disable tmtoolkit logging for now (too much output)\n",
    "disable_logging()\n",
    "\n",
    "# suppress the \"INFO\" messages and warnings from lda\n",
    "logger = logging.getLogger('lda')\n",
    "logger.addHandler(logging.NullHandler())\n",
    "logger.propagate = False\n",
    "\n",
    "warnings.filterwarnings('ignore')\n",
    "\n",
    "# set data to use\n",
    "dtms = {\n",
    "    'bigger': dtm_bg,\n",
    "    'smaller': dtm_sm\n",
    "}\n",
    "\n",
    "# and fixed hyperparameters\n",
    "lda_params = {\n",
    "    'n_topics': 10,\n",
    "    'n_iter': 1000,\n",
    "    'random_state': 20191122  # to make results reproducible\n",
    "}\n",
    "\n",
    "models = compute_models_parallel(dtms, constant_parameters=lda_params)\n",
    "models"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As expected, two models were created. These can be accessed via the labels that we used in the `dtms` dict:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:53:42.735351Z",
     "iopub.status.busy": "2022-02-08T06:53:42.734579Z",
     "iopub.status.idle": "2022-02-08T06:53:42.738391Z",
     "shell.execute_reply": "2022-02-08T06:53:42.737935Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[({'n_topics': 10, 'n_iter': 1000, 'random_state': 20191122},\n",
       "  <lda.lda.LDA at 0x7fb69dd0d580>)]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "models['smaller']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see that for each input DTM, we get a list of 2-tuples. The first element in each tuple is a dict that represents the hyperparameters that were used to compute the model, the second element is the actual topic model (the `<lda.lda.LDA ...>` object). This structure looks a bit complex, but this is because it also supports varying parameters. Since we only have one fixed set of hyperparameters per DTM, we only have a list of length 1 for each DTM.\n",
    "\n",
    "We will now access the models and print the top words per topic by using [print_ldamodel_topic_words](api.rst#tmtoolkit.topicmod.model_io.print_ldamodel_topic_words):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:53:42.749164Z",
     "iopub.status.busy": "2022-02-08T06:53:42.748308Z",
     "iopub.status.idle": "2022-02-08T06:53:42.762215Z",
     "shell.execute_reply": "2022-02-08T06:53:42.762623Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "topic_1\n",
      "> #1. mr (0.159627)\n",
      "> #2. germany (0.134861)\n",
      "> #3. member (0.096337)\n",
      "topic_2\n",
      "> #1. police (0.229620)\n",
      "> #2. man (0.185890)\n",
      "> #3. officer (0.145804)\n",
      "topic_3\n",
      "> #1. party (0.132964)\n",
      "> #2. election (0.132964)\n",
      "> #3. leader (0.072724)\n",
      "topic_4\n",
      "> #1. people (0.149059)\n",
      "> #2. country (0.117747)\n",
      "> #3. attack (0.056375)\n",
      "topic_5\n",
      "> #1. al (0.118424)\n",
      "> #2. syria (0.109317)\n",
      "> #3. force (0.097174)\n",
      "topic_6\n",
      "> #1. company (0.173780)\n",
      "> #2. percent (0.086901)\n",
      "> #3. business (0.069525)\n",
      "topic_7\n",
      "> #1. trump (0.196396)\n",
      "> #2. house (0.109759)\n",
      "> #3. president (0.088581)\n",
      "topic_8\n",
      "> #1. year (0.217796)\n",
      "> #2. time (0.102423)\n",
      "> #3. day (0.073904)\n",
      "topic_9\n",
      "> #1. china (0.299307)\n",
      "> #2. development (0.074851)\n",
      "> #3. european (0.071598)\n",
      "topic_10\n",
      "> #1. report (0.087935)\n",
      "> #2. official (0.058197)\n",
      "> #3. president (0.058197)\n"
     ]
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.model_io import print_ldamodel_topic_words\n",
    "\n",
    "model_sm = models['smaller'][0][1]\n",
    "print_ldamodel_topic_words(model_sm.topic_word_, vocab_sm, top_n=3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:53:42.767699Z",
     "iopub.status.busy": "2022-02-08T06:53:42.766896Z",
     "iopub.status.idle": "2022-02-08T06:53:42.785129Z",
     "shell.execute_reply": "2022-02-08T06:53:42.784527Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "topic_1\n",
      "> #1. american (0.041231)\n",
      "> #2. new (0.036285)\n",
      "> #3. america (0.029689)\n",
      "topic_2\n",
      "> #1. year (0.032174)\n",
      "> #2. day (0.026613)\n",
      "> #3. work (0.025025)\n",
      "topic_3\n",
      "> #1. party (0.051625)\n",
      "> #2. election (0.051625)\n",
      "> #3. vote (0.040334)\n",
      "topic_4\n",
      "> #1. people (0.090103)\n",
      "> #2. country (0.067925)\n",
      "> #3. million (0.032580)\n",
      "topic_5\n",
      "> #1. say (0.109116)\n",
      "> #2. report (0.040041)\n",
      "> #3. mr (0.025522)\n",
      "topic_6\n",
      "> #1. trump (0.085939)\n",
      "> #2. president (0.063024)\n",
      "> #3. russian (0.045838)\n",
      "topic_7\n",
      "> #1. police (0.048848)\n",
      "> #2. man (0.039545)\n",
      "> #3. officer (0.034119)\n",
      "topic_8\n",
      "> #1. china (0.066056)\n",
      "> #2. company (0.057441)\n",
      "> #3. market (0.030878)\n",
      "topic_9\n",
      "> #1. help (0.029090)\n",
      "> #2. child (0.028429)\n",
      "> #3. good (0.024463)\n",
      "topic_10\n",
      "> #1. say (0.091405)\n",
      "> #2. year (0.040949)\n",
      "> #3. come (0.021510)\n"
     ]
    }
   ],
   "source": [
    "model_bg = models['bigger'][0][1]\n",
    "print_ldamodel_topic_words(model_bg.topic_word_, vocab_bg, top_n=3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We could also generate models from different parameters in parallel, either for a single DTM or several. In the following example we generate models for a series of four different values for the `alpha` parameter. The parameters `n_iter` and `n_topics` are held constant across all models."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:53:42.791904Z",
     "iopub.status.busy": "2022-02-08T06:53:42.791307Z",
     "iopub.status.idle": "2022-02-08T06:53:46.306355Z",
     "shell.execute_reply": "2022-02-08T06:53:46.307525Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[({'alpha': 0.0001, 'n_iter': 500, 'n_topics': 10, 'random_state': 20191122},\n",
       "  <lda.lda.LDA at 0x7fb69dcec430>),\n",
       " ({'alpha': 0.001, 'n_iter': 500, 'n_topics': 10, 'random_state': 20191122},\n",
       "  <lda.lda.LDA at 0x7fb69dd01ee0>),\n",
       " ({'alpha': 0.01, 'n_iter': 500, 'n_topics': 10, 'random_state': 20191122},\n",
       "  <lda.lda.LDA at 0x7fb69dd015e0>),\n",
       " ({'alpha': 0.1, 'n_iter': 500, 'n_topics': 10, 'random_state': 20191122},\n",
       "  <lda.lda.LDA at 0x7fb69dd017c0>)]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "var_params = [{'alpha': 1/(10**x)} for x in range(1, 5)]\n",
    "\n",
    "const_params = {\n",
    "    'n_iter': 500,\n",
    "    'n_topics': 10,\n",
    "    'random_state': 20191122  # to make results reproducible\n",
    "}\n",
    "\n",
    "models = compute_models_parallel(dtm_sm,  # smaller DTM\n",
    "                                 varying_parameters=var_params,\n",
    "                                 constant_parameters=const_params)\n",
    "models"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We could compare these models now, e.g. by investigating their topics.\n",
    "\n",
    "A more systematic approach on comparing and evaluating topic models, also in order to find a good set of hyperparameters for a given dataset, will be presented in the next section.\n",
    "\n",
    "## Evaluation of topic models\n",
    "\n",
    "The package tmtoolkit provides several metrics for comparing and evaluating topic models. This can be used for finding a good hyperparameter set for a given dataset, e.g. a good combination of the number of topics and concentration paramaters (often called alpha and beta in literature). For some background on hyperparameters in topic modeling, see [this blog post](https://datascience.blog.wzb.eu/2017/11/09/topic-modeling-evaluation-in-python-with-tmtoolkit/).\n",
    "\n",
    "For each candidate hyperparameter set, a model can be generated and evaluated in parallel. We will do this now for the \"big\" DTM `dtm_bg`. Our candidate values for the number of topics `k` range between 20 and 120, with steps of 10. We make the concentration parameter for a prior over the document-specific topic distributions, alpha, depending on `k` as `1/k`:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:53:46.315849Z",
     "iopub.status.busy": "2022-02-08T06:53:46.312566Z",
     "iopub.status.idle": "2022-02-08T06:53:46.333242Z",
     "shell.execute_reply": "2022-02-08T06:53:46.331311Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[{'n_topics': 20, 'alpha': 0.05},\n",
       " {'n_topics': 30, 'alpha': 0.03333333333333333},\n",
       " {'n_topics': 40, 'alpha': 0.025},\n",
       " {'n_topics': 50, 'alpha': 0.02},\n",
       " {'n_topics': 60, 'alpha': 0.016666666666666666},\n",
       " {'n_topics': 70, 'alpha': 0.014285714285714285},\n",
       " {'n_topics': 80, 'alpha': 0.0125},\n",
       " {'n_topics': 90, 'alpha': 0.011111111111111112},\n",
       " {'n_topics': 100, 'alpha': 0.01},\n",
       " {'n_topics': 110, 'alpha': 0.00909090909090909},\n",
       " {'n_topics': 120, 'alpha': 0.008333333333333333}]"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "var_params = [{'n_topics': k, 'alpha': 1/k}\n",
    "               for k in range(20, 121, 10)]\n",
    "var_params"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The heart of the model evaluation process is the function [evaluate_topic_models](api.rst#tmtoolkit.topicmod.tm_lda.evaluate_topic_models), which is available for all three topic modeling packages. We stick with lda and import that function from [topicmod.tm_lda](api.rst#module-tmtoolkit.topicmod.tm_lda). It is similar to [compute_models_parallel](api.rst#tmtoolkit.topicmod.tm_lda.compute_models_parallel) as it accepts varying and constant hyperparameters. However, it doesn't only compute the models in parallel, but also applies several metrics to these models in order to evaluate them. This can be controlled with the `metric` parameter that accepts a string or a list of strings that specify the used metric(s). These metrics refer to functions that are implemented in [topicmod.evaluate](api.rst#module-tmtoolkit.topicmod.evaluate).\n",
    "\n",
    "Each topic modeling sub-module defines two important sequences: `AVAILABLE_METRICS` and `DEFAULT_METRICS`. The former lists all available metrics for that sub-module, the latter lists the default metrics that are used when you don't specify anything with the `metric` parameter. Let's have a look at both sequences in [topicmod.tm_lda](api.rst#module-tmtoolkit.topicmod.tm_lda):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:53:46.343307Z",
     "iopub.status.busy": "2022-02-08T06:53:46.339496Z",
     "iopub.status.idle": "2022-02-08T06:53:46.351293Z",
     "shell.execute_reply": "2022-02-08T06:53:46.350536Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "('loglikelihood',\n",
       " 'cao_juan_2009',\n",
       " 'arun_2010',\n",
       " 'coherence_mimno_2011',\n",
       " 'griffiths_2004',\n",
       " 'held_out_documents_wallach09',\n",
       " 'coherence_gensim_u_mass',\n",
       " 'coherence_gensim_c_v',\n",
       " 'coherence_gensim_c_uci',\n",
       " 'coherence_gensim_c_npmi')"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod import tm_lda\n",
    "\n",
    "tm_lda.AVAILABLE_METRICS"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:53:46.356225Z",
     "iopub.status.busy": "2022-02-08T06:53:46.355210Z",
     "iopub.status.idle": "2022-02-08T06:53:46.365373Z",
     "shell.execute_reply": "2022-02-08T06:53:46.366866Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "('cao_juan_2009', 'arun_2010', 'coherence_mimno_2011')"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tm_lda.DEFAULT_METRICS"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For details about the metrics and the academic references, see the respective implementations in the [topicmod.evaluate](api.rst#module-tmtoolkit.topicmod.evaluate) module.\n",
    "\n",
    "We will now run the model evaluations with [evaluate_topic_models](api.rst#tmtoolkit.topicmod.tm_lda.evaluate_topic_models) using our previously generated list of varying hyperparameters `var_params`, some constant hyperparameters and the default set of metrics. We also set `return_models=True` which means to retain the generated models in the evaluation results:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:53:46.377285Z",
     "iopub.status.busy": "2022-02-08T06:53:46.370234Z",
     "iopub.status.idle": "2022-02-08T06:54:39.457698Z",
     "shell.execute_reply": "2022-02-08T06:54:39.457302Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[({'n_topics': 20,\n",
       "   'alpha': 0.05,\n",
       "   'n_iter': 1000,\n",
       "   'random_state': 20191122,\n",
       "   'eta': 0.1},\n",
       "  {'model': <lda.lda.LDA at 0x7fb69df052e0>,\n",
       "   'cao_juan_2009': 0.1583051825037188,\n",
       "   'arun_2010': 6.813827298707812,\n",
       "   'coherence_mimno_2011': -1.696975792080187}),\n",
       " ({'n_topics': 30,\n",
       "   'alpha': 0.03333333333333333,\n",
       "   'n_iter': 1000,\n",
       "   'random_state': 20191122,\n",
       "   'eta': 0.1},\n",
       "  {'model': <lda.lda.LDA at 0x7fb69dd01610>,\n",
       "   'cao_juan_2009': 0.11654514557276367,\n",
       "   'arun_2010': 4.260416549019389,\n",
       "   'coherence_mimno_2011': -1.563392437504735}),\n",
       " ({'n_topics': 40,\n",
       "   'alpha': 0.025,\n",
       "   'n_iter': 1000,\n",
       "   'random_state': 20191122,\n",
       "   'eta': 0.1},\n",
       "  {'model': <lda.lda.LDA at 0x7fb69dd01730>,\n",
       "   'cao_juan_2009': 0.11915454452526894,\n",
       "   'arun_2010': 3.0999249948241188,\n",
       "   'coherence_mimno_2011': -1.688220159791409})]"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.tm_lda import evaluate_topic_models\n",
    "from tmtoolkit.topicmod.evaluate import results_by_parameter\n",
    "\n",
    "const_params = {\n",
    "    'n_iter': 1000,\n",
    "    'random_state': 20191122,  # to make results reproducible\n",
    "    'eta': 0.1,                # sometimes also called \"beta\"\n",
    "}\n",
    "\n",
    "eval_results = evaluate_topic_models(dtm_bg,\n",
    "                                     varying_parameters=var_params,\n",
    "                                     constant_parameters=const_params,\n",
    "                                     return_models=True)\n",
    "eval_results[:3]  # only show first three models"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The evaluation results are a list with pairs of hyperparameters and their evaluation results for each metric. Additionally, there is the generated model for each hyperparameter set.\n",
    "\n",
    "We now use [results_by_parameter](api.rst#tmtoolkit.topicmod.evaluate.results_by_parameter), which takes the \"raw\" evaluation results and sorts them by a specific hyperparameter, in this case `n_topics`. This is important because this is the way that the function for visualizing evaluation results, [plot_eval_results](api.rst#tmtoolkit.topicmod.visualize.plot_eval_results), expects the input."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:39.462865Z",
     "iopub.status.busy": "2022-02-08T06:54:39.462301Z",
     "iopub.status.idle": "2022-02-08T06:54:39.465191Z",
     "shell.execute_reply": "2022-02-08T06:54:39.464721Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(20,\n",
       "  {'model': <lda.lda.LDA at 0x7fb69df052e0>,\n",
       "   'cao_juan_2009': 0.1583051825037188,\n",
       "   'arun_2010': 6.813827298707812,\n",
       "   'coherence_mimno_2011': -1.696975792080187}),\n",
       " (30,\n",
       "  {'model': <lda.lda.LDA at 0x7fb69dd01610>,\n",
       "   'cao_juan_2009': 0.11654514557276367,\n",
       "   'arun_2010': 4.260416549019389,\n",
       "   'coherence_mimno_2011': -1.563392437504735}),\n",
       " (40,\n",
       "  {'model': <lda.lda.LDA at 0x7fb69dd01730>,\n",
       "   'cao_juan_2009': 0.11915454452526894,\n",
       "   'arun_2010': 3.0999249948241188,\n",
       "   'coherence_mimno_2011': -1.688220159791409})]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "eval_results_by_topics = results_by_parameter(eval_results, 'n_topics')\n",
    "eval_results_by_topics[:3]  # again only the first three models"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can now see the results for each metric across the specified range of number of topics using [plot_eval_results](api.rst#tmtoolkit.topicmod.visualize.plot_eval_results):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:39.476912Z",
     "iopub.status.busy": "2022-02-08T06:54:39.476400Z",
     "iopub.status.idle": "2022-02-08T06:54:39.799464Z",
     "shell.execute_reply": "2022-02-08T06:54:39.799849Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x432 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.visualize import plot_eval_results\n",
    "\n",
    "plot_eval_results(eval_results_by_topics);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "These results suggest to set the number of topics, `n_topics`, to 50 and alpha to `0.02`. We don't have to generate a model with these hyperparameters again, because it's already in the evaluation results (thanks to `return_models=True`). We extract the model from there in order to use it in the rest of the chapter."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:39.804293Z",
     "iopub.status.busy": "2022-02-08T06:54:39.803831Z",
     "iopub.status.idle": "2022-02-08T06:54:39.808202Z",
     "shell.execute_reply": "2022-02-08T06:54:39.807709Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(50, 0.02, 0.1)"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "best_tm = [m for k, m in eval_results_by_topics if k == 50][0]['model']\n",
    "best_tm.n_topics, best_tm.alpha, best_tm.eta  # just to make sure"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Common statistics and tools for topic models\n",
    "\n",
    "The [topicmod.model_stats](api.rst#module-tmtoolkit.topicmod.model_stats) module mostly contains functions that compute statistics from the document-topic and topic-word distribution of a topic model and also some helper functions for working with such distributions. We'll start with an important helper function, [generate_topic_labels_from_top_words](api.rst#tmtoolkit.topicmod.model_stats.generate_topic_labels_from_top_words)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Generating labels for topics\n",
    "\n",
    "In topic modeling, topics are numbered because they're *abstract* – they're simply a probability distribution across all words in the vocabulary. Still, it's useful to give them labels for better identification. The function [generate_topic_labels_from_top_words](api.rst#tmtoolkit.topicmod.model_stats.generate_topic_labels_from_top_words) is very useful for that, as it finds labels according to the most \"relevant\" words in each topic. We'll later see how we can identify the most relevant words per topic using a special [relevance statistic](#Topic-word-relevance). Note that you can adjust the weight of the relevance measure for the ranking by using the parameter `lambda_` which is in range $[0, 1]$.\n",
    "\n",
    "The function requires at least the topic-word and document-topic distributions from the model, the document lengths and the vocabulary. It then finds the minimum number of relevant words that uniquely label each topic. You can also use a fixed number for that minimum number with the parameter `n_words`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:39.813218Z",
     "iopub.status.busy": "2022-02-08T06:54:39.812699Z",
     "iopub.status.idle": "2022-02-08T06:54:39.838438Z",
     "shell.execute_reply": "2022-02-08T06:54:39.838854Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['1_error_case', '2_germany_german', '3_trump_president',\n",
       "       '4_circumstance_describe', '5_flight_air', '6_report_agency',\n",
       "       '7_company_reform', '8_intelligence_russia', '9_party_vote',\n",
       "       '10_al_syria'], dtype='<U23')"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.bow.bow_stats import doc_lengths\n",
    "from tmtoolkit.topicmod.model_stats import generate_topic_labels_from_top_words\n",
    "\n",
    "vocab_bg = np.array(vocab_bg)   # we need this to be an array\n",
    "\n",
    "doc_lengths_bg = doc_lengths(dtm_bg)\n",
    "topic_labels = generate_topic_labels_from_top_words(\n",
    "    best_tm.topic_word_,\n",
    "    best_tm.doc_topic_,\n",
    "    doc_lengths_bg,\n",
    "    np.array(vocab_bg),\n",
    "    lambda_=0.6\n",
    ")\n",
    "\n",
    "topic_labels[:10]   # showing only the first 5 topics here"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As we can see, two words are necessary to label each topic uniquely. By default, each label is prefixed with a number. You can change that with the parameter `labels_format`.\n",
    "\n",
    "Let's have a look at the top words for a specific topic. We can use [ldamodel_top_topic_words](api.rst#tmtoolkit.topicmod.model_io.ldamodel_top_topic_words) for that from the module [topicmod.model_io](api.rst#module-tmtoolkit.topicmod.model_io), which we will have a closer look at [later](#Displaying-and-exporting-topic-modeling-results):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:39.843601Z",
     "iopub.status.busy": "2022-02-08T06:54:39.843070Z",
     "iopub.status.idle": "2022-02-08T06:54:40.735198Z",
     "shell.execute_reply": "2022-02-08T06:54:40.734677Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>rank_1</th>\n",
       "      <th>rank_2</th>\n",
       "      <th>rank_3</th>\n",
       "      <th>rank_4</th>\n",
       "      <th>rank_5</th>\n",
       "      <th>rank_6</th>\n",
       "      <th>rank_7</th>\n",
       "      <th>rank_8</th>\n",
       "      <th>rank_9</th>\n",
       "      <th>rank_10</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>9_party_vote</th>\n",
       "      <td>party (0.09073)</td>\n",
       "      <td>vote (0.04245)</td>\n",
       "      <td>leader (0.02635)</td>\n",
       "      <td>right (0.02434)</td>\n",
       "      <td>minister (0.02233)</td>\n",
       "      <td>percent (0.02233)</td>\n",
       "      <td>voter (0.02032)</td>\n",
       "      <td>political (0.01629)</td>\n",
       "      <td>call (0.01629)</td>\n",
       "      <td>come (0.01629)</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                       rank_1          rank_2            rank_3  \\\n",
       "topic                                                             \n",
       "9_party_vote  party (0.09073)  vote (0.04245)  leader (0.02635)   \n",
       "\n",
       "                       rank_4              rank_5             rank_6  \\\n",
       "topic                                                                  \n",
       "9_party_vote  right (0.02434)  minister (0.02233)  percent (0.02233)   \n",
       "\n",
       "                       rank_7               rank_8          rank_9  \\\n",
       "topic                                                                \n",
       "9_party_vote  voter (0.02032)  political (0.01629)  call (0.01629)   \n",
       "\n",
       "                     rank_10  \n",
       "topic                         \n",
       "9_party_vote  come (0.01629)  "
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.model_io import ldamodel_top_topic_words\n",
    "\n",
    "top_topic_word = ldamodel_top_topic_words(best_tm.topic_word_,\n",
    "                                          vocab_bg,\n",
    "                                          row_labels=topic_labels)\n",
    "top_topic_word[top_topic_word.index == '9_party_vote']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Marginal topic and word distributions\n",
    "\n",
    "We'll now focus on the marginal topic and word distributions. Let's get the marginal topic distribution first by using [marginal_topic_distrib](api.rst#tmtoolkit.topicmod.model_stats.marginal_topic_distrib):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:40.739797Z",
     "iopub.status.busy": "2022-02-08T06:54:40.738416Z",
     "iopub.status.idle": "2022-02-08T06:54:40.742401Z",
     "shell.execute_reply": "2022-02-08T06:54:40.742805Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.01576, 0.01888, 0.01706, 0.00615, 0.00833, 0.03309, 0.03259,\n",
       "       0.01403, 0.02541, 0.03231, 0.02303, 0.01714, 0.02143, 0.03394,\n",
       "       0.01467, 0.01534, 0.0367 , 0.02148, 0.01134, 0.01369, 0.02523,\n",
       "       0.01143, 0.0359 , 0.01353, 0.01892, 0.01301, 0.01765, 0.02398,\n",
       "       0.01649, 0.02521, 0.02488, 0.01322, 0.00826, 0.03753, 0.00964,\n",
       "       0.08694, 0.01579, 0.0151 , 0.01658, 0.02005, 0.01848, 0.00534,\n",
       "       0.01172, 0.00715, 0.01603, 0.02789, 0.01261, 0.00995, 0.01288,\n",
       "       0.01625])"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.model_stats import marginal_topic_distrib\n",
    "\n",
    "marg_topic = marginal_topic_distrib(best_tm.doc_topic_, doc_lengths_bg)\n",
    "marg_topic"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The marginal topic distribution can be interpreted as the \"importance\" of each topic for the whole corpus. Let's get the sorted indices into `topic_labels` with `np.argsort` and get the top five topics:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:40.747181Z",
     "iopub.status.busy": "2022-02-08T06:54:40.746629Z",
     "iopub.status.idle": "2022-02-08T06:54:40.749412Z",
     "shell.execute_reply": "2022-02-08T06:54:40.749813Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['36_say_year', '34_america_nation', '17_white_trump',\n",
       "       '23_country_love', '14_say_committee'], dtype='<U23')"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# np.argsort gives ascending order, hence reverse via [::-1]\n",
    "topic_labels[np.argsort(marg_topic)[::-1][:5]]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Likewise, we can get the marginal word distribution with [marginal_word_distrib](api.rst#tmtoolkit.topicmod.model_stats.marginal_word_distrib) from the model's topic-word distribution and the marginal topic distribution. We'll use this to list the most probable words for the corpus. As expected, these are mostly quite common words:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:40.758559Z",
     "iopub.status.busy": "2022-02-08T06:54:40.757908Z",
     "iopub.status.idle": "2022-02-08T06:54:40.764444Z",
     "shell.execute_reply": "2022-02-08T06:54:40.763758Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['say', 'year', 'people', 'country', 'new', 'time', 'trump',\n",
       "       'report', 'china', 'president'], dtype='<U14')"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.model_stats import marginal_word_distrib\n",
    "\n",
    "marg_word = marginal_word_distrib(best_tm.topic_word_, marg_topic)\n",
    "vocab_bg[np.argsort(marg_word)[::-1][:10]]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Two helper functions exist for this purpose: [most_probable_words](api.rst#tmtoolkit.topicmod.model_stats.most_probable_words) and [least_probable_words](api.rst#tmtoolkit.topicmod.model_stats.least_probable_words) sort the vocabulary according to the marginal probability:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:40.769728Z",
     "iopub.status.busy": "2022-02-08T06:54:40.768443Z",
     "iopub.status.idle": "2022-02-08T06:54:40.773001Z",
     "shell.execute_reply": "2022-02-08T06:54:40.773437Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['say', 'year', 'people', 'country', 'new', 'time', 'trump',\n",
       "       'report', 'china', 'president'], dtype='<U14')"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.model_stats import most_probable_words, least_probable_words\n",
    "\n",
    "most_probable_words(vocab_bg, best_tm.topic_word_,\n",
    "                    best_tm.doc_topic_, doc_lengths_bg,\n",
    "                    n=10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:40.778057Z",
     "iopub.status.busy": "2022-02-08T06:54:40.777214Z",
     "iopub.status.idle": "2022-02-08T06:54:40.783587Z",
     "shell.execute_reply": "2022-02-08T06:54:40.784219Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['urge', 'series', 'reveal', 'protection', 'associate', 'argue',\n",
       "       'elect', 'analysis', 'seven', 'guarantee'], dtype='<U14')"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "least_probable_words(vocab_bg, best_tm.topic_word_,\n",
    "                     best_tm.doc_topic_, doc_lengths_bg,\n",
    "                     n=10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Word distinctiveness and saliency\n",
    "\n",
    "Word *distinctiveness* and *saliency* (see below) help to identify the most \"informative\" words in a corpus given its topic model. Both measures are introduced in [Chuang et al. 2012](https://dl.acm.org/citation.cfm?id=2254572).\n",
    "\n",
    "Word distinctiveness is calculated for each word $w$ as\n",
    "\n",
    "$\\text{distinctiveness}(w) = \\sum_T(P(T|w) \\log \\frac{P(T|w)}{P(T)})$.\n",
    "\n",
    "where $P(T)$ is the marginal topic distribution and $P(T|w)$ is the probability of a topic given a word $w$.\n",
    "\n",
    "We can calculate this measure using [word_distinctiveness](api.rst#tmtoolkit.topicmod.model_stats.word_distinctiveness). To use this measure directly to rank words, we can use [most_distinct_words](api.rst#tmtoolkit.topicmod.model_stats.most_distinct_words) and [least_distinct_words](api.rst#tmtoolkit.topicmod.model_stats.least_distinct_words):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:40.789985Z",
     "iopub.status.busy": "2022-02-08T06:54:40.789248Z",
     "iopub.status.idle": "2022-02-08T06:54:40.794149Z",
     "shell.execute_reply": "2022-02-08T06:54:40.793727Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.78042, 1.13865, 1.21893, 1.01726, 1.20055, 1.5611 , 1.18047,\n",
       "       1.58108, 0.74311, 0.97515])"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.model_stats import word_distinctiveness, \\\n",
    "    most_distinct_words, least_distinct_words\n",
    "\n",
    "word_distinct = word_distinctiveness(best_tm.topic_word_, marg_topic)\n",
    "word_distinct[:10]   # first 10 words in vocab"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:40.798059Z",
     "iopub.status.busy": "2022-02-08T06:54:40.797523Z",
     "iopub.status.idle": "2022-02-08T06:54:40.801379Z",
     "shell.execute_reply": "2022-02-08T06:54:40.801796Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['note', 'space', 'china', '7', 'bank', 'judge', 'north', 'flight',\n",
       "       'police', 'mr'], dtype='<U14')"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "most_distinct_words(vocab_bg, best_tm.topic_word_,\n",
    "                    best_tm.doc_topic_, doc_lengths_bg,\n",
    "                    n=10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:40.805564Z",
     "iopub.status.busy": "2022-02-08T06:54:40.805024Z",
     "iopub.status.idle": "2022-02-08T06:54:40.809478Z",
     "shell.execute_reply": "2022-02-08T06:54:40.809897Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['away', 'adviser', 'agree', 'adopt', 'place', 'effect',\n",
       "       'conference', 'currently', 'mind', 'explain'], dtype='<U14')"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "least_distinct_words(vocab_bg, best_tm.topic_word_,\n",
    "                     best_tm.doc_topic_, doc_lengths_bg,\n",
    "                     n=10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Word *saliency* weights each words' distinctiveness by it's marginal probability $P(w)$:\n",
    "\n",
    "$\\text{saliency}(w) = P(w) \\cdot \\text{distinctiveness}(w)$.\n",
    "\n",
    "The respective functions in tmtoolkit are [word_saliency](api.rst#tmtoolkit.topicmod.model_stats.word_saliency), [most_salient_words](api.rst#tmtoolkit.topicmod.model_stats.most_salient_words) and [least_salient_words](api.rst#tmtoolkit.topicmod.model_stats.least_salient_words):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:40.814316Z",
     "iopub.status.busy": "2022-02-08T06:54:40.813604Z",
     "iopub.status.idle": "2022-02-08T06:54:40.817966Z",
     "shell.execute_reply": "2022-02-08T06:54:40.818373Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.00054, 0.00106, 0.00136, 0.00081, 0.00093, 0.00084, 0.00091,\n",
       "       0.00105, 0.00044, 0.00072])"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.model_stats import word_saliency, \\\n",
    "    most_salient_words, least_salient_words\n",
    "\n",
    "word_sal = word_saliency(best_tm.topic_word_, best_tm.doc_topic_, doc_lengths_bg)\n",
    "word_sal[:10]   # first 10 words in vocab"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:40.821990Z",
     "iopub.status.busy": "2022-02-08T06:54:40.821461Z",
     "iopub.status.idle": "2022-02-08T06:54:40.825853Z",
     "shell.execute_reply": "2022-02-08T06:54:40.826265Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['say', 'china', 'trump', 'people', 'report', 'new', 'year',\n",
       "       'country', 'police', 'mr'], dtype='<U14')"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "most_salient_words(vocab_bg, best_tm.topic_word_,\n",
    "                   best_tm.doc_topic_, doc_lengths_bg,\n",
    "                   n=10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:40.830459Z",
     "iopub.status.busy": "2022-02-08T06:54:40.829698Z",
     "iopub.status.idle": "2022-02-08T06:54:40.833660Z",
     "shell.execute_reply": "2022-02-08T06:54:40.834069Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['adopt', 'effect', 'personal', 'adviser', 'positive', 'shortly',\n",
       "       'strengthen', 'conduct', 'arab', 'conference'], dtype='<U14')"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "least_salient_words(vocab_bg, best_tm.topic_word_,\n",
    "                    best_tm.doc_topic_, doc_lengths_bg,\n",
    "                    n=10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Topic-word relevance\n",
    "\n",
    "The topic-word relevance measure as introduced by [Sievert and Shirley 2014](https://www.aclweb.org/anthology/W14-3110/) helps to identify the most relevant words within a topic by also accounting for the marginal probability of each word across the corpus. This is done by integrating a *lift* value, which is the \"ratio of a term's probability within a topic to its marginal probability across the corpus.\" (ibid.)\n",
    "\n",
    "Thus for each word $w$, given a topic-word distribution $\\phi$, a topic $t$ and a weight parameter $\\lambda$, it is calculated as:\n",
    "\n",
    "$\\text{relevance}_{\\phi, \\lambda}(w, t) = \\lambda \\log \\phi_{t,w} + (1-\\lambda) \\log \\frac{\\phi_{t,w}}{p(w)}$.\n",
    "\n",
    "The first term $\\log \\phi_{t,w}$ is the log of the topic-word distribution, the second term $\\log \\frac{\\phi_{t,w}}{p(w)}$ is the *log lift* and $\\lambda$ can be used to control the weight between both terms. The lower $\\lambda$, the more weight is put on the lift term, i.e. the more different are the results from the original topic-word distribution.\n",
    "\n",
    "This measure is implemented in [topic_word_relevance](api.rst#tmtoolkit.topicmod.model_stats.topic_word_relevance). It returns a matrix of the same shape as the topic-word distribution, i.e. each row represents a topic with a (log-transformed) distribution across all words in the vocabulary. Please note that the lambda parameter ends with an underscore: `lambda_`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:40.838043Z",
     "iopub.status.busy": "2022-02-08T06:54:40.837426Z",
     "iopub.status.idle": "2022-02-08T06:54:40.843604Z",
     "shell.execute_reply": "2022-02-08T06:54:40.843182Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-5.21276, -5.33602, -5.40768, ..., -6.27353, -5.35915, -5.51232],\n",
       "       [-5.35582, -5.47908, -5.55074, ..., -2.70302, -1.78864, -1.54451],\n",
       "       [-5.27302, -2.35175, -5.46794, ..., -6.33379, -5.41941, -5.57258],\n",
       "       ...,\n",
       "       [-4.87877, -5.00203, -5.07369, ..., -5.93954, -5.02516, -5.17833],\n",
       "       [-5.05964, -2.78501, -5.25456, ..., -6.12041, -5.20604, -5.3592 ],\n",
       "       [-5.23614, -5.3594 , -5.43107, ..., -2.18604, -5.38254, -5.5357 ]])"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.model_stats import topic_word_relevance\n",
    "\n",
    "topic_word_rel = topic_word_relevance(best_tm.topic_word_, best_tm.doc_topic_,\n",
    "                                      doc_lengths_bg, lambda_=0.6)\n",
    "topic_word_rel"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To confirm that it's 50 topics across all words in the vocabulary:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:40.847604Z",
     "iopub.status.busy": "2022-02-08T06:54:40.847050Z",
     "iopub.status.idle": "2022-02-08T06:54:40.849744Z",
     "shell.execute_reply": "2022-02-08T06:54:40.850138Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(50, 791)"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "topic_word_rel.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Two functions can be used to get the most or least relevant words for a topic: [most_relevant_words_for_topic](api.rst#tmtoolkit.topicmod.model_stats.most_relevant_words_for_topic) and [least_relevant_words_for_topic](api.rst#tmtoolkit.topicmod.model_stats.least_relevant_words_for_topic). You can select the topic with the `topic` parameter which is a **zero-based topic index**.\n",
    "\n",
    "We'll do it for topic with index 9, which is:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:40.854401Z",
     "iopub.status.busy": "2022-02-08T06:54:40.853652Z",
     "iopub.status.idle": "2022-02-08T06:54:40.857062Z",
     "shell.execute_reply": "2022-02-08T06:54:40.856519Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'10_al_syria'"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "topic_labels[9]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:40.864285Z",
     "iopub.status.busy": "2022-02-08T06:54:40.863507Z",
     "iopub.status.idle": "2022-02-08T06:54:40.868411Z",
     "shell.execute_reply": "2022-02-08T06:54:40.868932Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['al', 'syria', 'syrian', 'opposition', 'city', 'area', 'military',\n",
       "       'kill', 'war', 'strike'], dtype='<U14')"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.model_stats import most_relevant_words_for_topic, \\\n",
    "    least_relevant_words_for_topic\n",
    "\n",
    "most_relevant_words_for_topic(vocab_bg, topic_word_rel, topic=9, n=10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:40.875009Z",
     "iopub.status.busy": "2022-02-08T06:54:40.874424Z",
     "iopub.status.idle": "2022-02-08T06:54:40.877271Z",
     "shell.execute_reply": "2022-02-08T06:54:40.876842Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['year', 'people', 'time', 'trump', 'report', 'china', 'president',\n",
       "       'come', 'state', 'company'], dtype='<U14')"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "least_relevant_words_for_topic(vocab_bg, topic_word_rel, topic=9, n=10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Topic coherence\n",
    "\n",
    "We already used the *coherence* metric ([Mimno et al. 2011](https://dl.acm.org/citation.cfm?id=2145462)) for topic model evaluation. However, this metric cannot only be used to assess the overall quality of a topic model, but also to evaluate the individual topics' coherence."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:40.881508Z",
     "iopub.status.busy": "2022-02-08T06:54:40.881009Z",
     "iopub.status.idle": "2022-02-08T06:54:41.149508Z",
     "shell.execute_reply": "2022-02-08T06:54:41.149130Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([-2.06169, -0.69593, -1.13214, -1.48083, -2.53062, -1.14   ,\n",
       "       -1.30559, -1.66265, -1.18398, -1.24934, -1.50454, -1.24552,\n",
       "       -1.27484, -1.26755, -1.45137, -2.55208, -1.20722, -1.6191 ,\n",
       "       -2.06777, -1.2751 , -2.14256, -1.50499, -1.10072, -1.98659,\n",
       "       -2.34686, -1.56978, -1.97106, -1.54649, -1.58539, -1.20991,\n",
       "       -1.50709, -2.20169, -1.56954, -1.4693 , -3.6065 , -0.93041,\n",
       "       -1.18229, -1.30928, -3.1338 , -1.35259, -1.34244, -2.3644 ,\n",
       "       -2.73258, -3.99957, -1.45934, -1.34668, -2.20253, -2.63822,\n",
       "       -1.52807, -3.48115])"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.evaluate import metric_coherence_mimno_2011\n",
    "\n",
    "# use top 20 words per topic for metric\n",
    "coh = metric_coherence_mimno_2011(best_tm.topic_word_, dtm_bg, top_n=20)\n",
    "coh"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This generates a coherence value for each topic. Let's show the distribution of these values:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:41.170614Z",
     "iopub.status.busy": "2022-02-08T06:54:41.164369Z",
     "iopub.status.idle": "2022-02-08T06:54:41.320927Z",
     "shell.execute_reply": "2022-02-08T06:54:41.320452Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "plt.hist(coh, bins=20)\n",
    "plt.xlabel('coherence')\n",
    "plt.ylabel('n')\n",
    "plt.show();"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And print the best and worst topics according to this metric:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:41.325343Z",
     "iopub.status.busy": "2022-02-08T06:54:41.324771Z",
     "iopub.status.idle": "2022-02-08T06:54:41.327469Z",
     "shell.execute_reply": "2022-02-08T06:54:41.327846Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['2_germany_german', '36_say_year', '23_country_love',\n",
       "       '3_trump_president', '6_report_agency'], dtype='<U23')"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import numpy as np\n",
    "\n",
    "top10_t_indices = np.argsort(coh)[::-1][:5]\n",
    "bottom10_t_indices = np.argsort(coh)[:5]\n",
    "\n",
    "topic_labels[top10_t_indices]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:41.331790Z",
     "iopub.status.busy": "2022-02-08T06:54:41.331264Z",
     "iopub.status.idle": "2022-02-08T06:54:41.333784Z",
     "shell.execute_reply": "2022-02-08T06:54:41.334176Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['44_note_7', '35_force_group', '50_officer_police',\n",
       "       '39_russian_diplomat', '43_bank_investor'], dtype='<U23')"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "topic_labels[bottom10_t_indices]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note that this metric also doesn't spare oneself careful manual evaluation, because it can also be off for some topics. For example, topic `36_say_year` is certainly not a coherent topic as it mostly ranks very common but but less meaningful words high:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:41.344353Z",
     "iopub.status.busy": "2022-02-08T06:54:41.343678Z",
     "iopub.status.idle": "2022-02-08T06:54:41.346209Z",
     "shell.execute_reply": "2022-02-08T06:54:41.346648Z"
    }
   },
   "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>rank_1</th>\n",
       "      <th>rank_2</th>\n",
       "      <th>rank_3</th>\n",
       "      <th>rank_4</th>\n",
       "      <th>rank_5</th>\n",
       "      <th>rank_6</th>\n",
       "      <th>rank_7</th>\n",
       "      <th>rank_8</th>\n",
       "      <th>rank_9</th>\n",
       "      <th>rank_10</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>36_say_year</th>\n",
       "      <td>say (0.185)</td>\n",
       "      <td>year (0.04498)</td>\n",
       "      <td>time (0.02648)</td>\n",
       "      <td>tell (0.02582)</td>\n",
       "      <td>think (0.01922)</td>\n",
       "      <td>lot (0.01922)</td>\n",
       "      <td>take (0.01922)</td>\n",
       "      <td>people (0.01856)</td>\n",
       "      <td>know (0.01658)</td>\n",
       "      <td>want (0.01592)</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  rank_1          rank_2          rank_3          rank_4  \\\n",
       "topic                                                                      \n",
       "36_say_year  say (0.185)  year (0.04498)  time (0.02648)  tell (0.02582)   \n",
       "\n",
       "                      rank_5         rank_6          rank_7            rank_8  \\\n",
       "topic                                                                           \n",
       "36_say_year  think (0.01922)  lot (0.01922)  take (0.01922)  people (0.01856)   \n",
       "\n",
       "                     rank_9         rank_10  \n",
       "topic                                        \n",
       "36_say_year  know (0.01658)  want (0.01592)  "
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "top_topic_word[top_topic_word.index == '36_say_year']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "More coherence metrics can be used with the function [metric_coherence_gensim](api.rst#tmtoolkit.topicmod.evaluate.metric_coherence_gensim). This requires that [gensim](https://radimrehurek.com/gensim/) is installed. Furthemore, most metrics require that a parameter `texts` is passed which is the tokenized text that was used to create the document-term matrix."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Filtering topics\n",
    "\n",
    "With the function [filter_topics](api.rst#tmtoolkit.topicmod.model_stats.filter_topics), you can filter the topics according to their topic-word distribution and the following search criteria:\n",
    "\n",
    "- `search_pattern`: one or more search patterns according to the [common parameters for pattern matching](preprocessing.ipynb#Common-parameters-for-pattern-matching-functions)\n",
    "- `top_n`: pattern match(es) must occur in the first `top_n` most probable words in the topic\n",
    "- `thresh`: matched words' probability must be above this threshold\n",
    "\n",
    "You must specify at least one of `top_n` and `thresh`, but you can also specify both. The function returns an array of topic indices (which start with zero!).\n",
    "\n",
    "Let's find all topics that have the glob pattern (`match_type='glob'`) \"russ*\" (to match both \"russia\" and \"russian\") in the top 5 most probable words:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:41.351038Z",
     "iopub.status.busy": "2022-02-08T06:54:41.350284Z",
     "iopub.status.idle": "2022-02-08T06:54:41.359369Z",
     "shell.execute_reply": "2022-02-08T06:54:41.358948Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 7, 24, 38, 41])"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.model_stats import filter_topics\n",
    "\n",
    "found_topics = filter_topics('russ*', vocab_bg,\n",
    "                             best_tm.topic_word_, match_type='glob',\n",
    "                             top_n=5)\n",
    "found_topics"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can use these indices with our `topic_labels`:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:41.363272Z",
     "iopub.status.busy": "2022-02-08T06:54:41.362607Z",
     "iopub.status.idle": "2022-02-08T06:54:41.365288Z",
     "shell.execute_reply": "2022-02-08T06:54:41.365675Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['8_intelligence_russia', '25_sanction_visit',\n",
       "       '39_russian_diplomat', '42_win_anti'], dtype='<U23')"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "topic_labels[found_topics]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next, we want to select all topics where *any* of the words matched by the glob patterns `\"chin*\"` *or* `\"business\"` achieve at least a probability of 0.01 (`thresh=0.01`):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:41.369296Z",
     "iopub.status.busy": "2022-02-08T06:54:41.368810Z",
     "iopub.status.idle": "2022-02-08T06:54:41.411815Z",
     "shell.execute_reply": "2022-02-08T06:54:41.412163Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['7_company_reform', '22_space_european', '30_china_chinese',\n",
       "       '47_family_refugee'], dtype='<U23')"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "found_topics = filter_topics(['chin*', 'business'], vocab_bg,\n",
    "                             best_tm.topic_word_, thresh=0.01, match_type='glob')\n",
    "topic_labels[found_topics]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "When we specify `cond='all'`, *all* patterns must have at least one match (here in the top 10 list of words per topic):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:41.415817Z",
     "iopub.status.busy": "2022-02-08T06:54:41.415358Z",
     "iopub.status.idle": "2022-02-08T06:54:41.425808Z",
     "shell.execute_reply": "2022-02-08T06:54:41.425349Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([], dtype='<U23')"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "found_topics = filter_topics(['chin*', 'business'], vocab_bg,\n",
    "                             best_tm.topic_word_, top_n=10, match_type='glob',\n",
    "                             cond='all')\n",
    "topic_labels[found_topics]   # no result"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You could also pass a topic-word relevance matrix instead of a topic-word probability distribution."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:41.429917Z",
     "iopub.status.busy": "2022-02-08T06:54:41.429191Z",
     "iopub.status.idle": "2022-02-08T06:54:41.439078Z",
     "shell.execute_reply": "2022-02-08T06:54:41.438661Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['8_intelligence_russia', '25_sanction_visit',\n",
       "       '39_russian_diplomat', '42_win_anti'], dtype='<U23')"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "found_topics = filter_topics('russ*', vocab_bg,\n",
    "                             topic_word_rel, match_type='glob',\n",
    "                             top_n=5)\n",
    "topic_labels[found_topics]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Excluding topics\n",
    "\n",
    "It is often the case that some topics of a topic model rank a lot of uninformative (e.g. very common) words the highest. This results in some uninformative topics, which you may want to exclude from further analysis. Note that if a large fraction of topics seems uninformative, it points to a problem with your topic model and/or your preprocessing steps. You should [evaluate](#Evaluation-of-topic-models) your candidate models carefully with the mentioned metrics and/or adjust your text preprocessing pipeline.\n",
    "\n",
    "The function [exclude_topics](api.rst#tmtoolkit.topicmod.model_stats.exclude_topics) allows to remove a specified set of topics from the document-topic and topic-word distributions. You need to pass the **zero-based** indices of the topics that you want to remove, and both distributions.\n",
    "\n",
    "For example, suppose the following topics were identified as uninformative:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:41.443600Z",
     "iopub.status.busy": "2022-02-08T06:54:41.442985Z",
     "iopub.status.idle": "2022-02-08T06:54:41.445353Z",
     "shell.execute_reply": "2022-02-08T06:54:41.445747Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['1_error_case', '4_circumstance_describe', '36_say_year'],\n",
       "      dtype='<U23')"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "uninform_topics = [0, 3, 35]\n",
    "topic_labels[uninform_topics]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can now pass these indices to [exclude_topics](api.rst#tmtoolkit.topicmod.model_stats.exclude_topics) along with the topic model distributions. We'll get back new, filtered, distributions."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:41.450054Z",
     "iopub.status.busy": "2022-02-08T06:54:41.449097Z",
     "iopub.status.idle": "2022-02-08T06:54:41.452475Z",
     "shell.execute_reply": "2022-02-08T06:54:41.452859Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((100, 47), (47, 791))"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.model_stats import exclude_topics\n",
    "\n",
    "new_doc_topic, new_topic_word, new_topic_mapping = \\\n",
    "    exclude_topics(uninform_topics, best_tm.doc_topic_,\n",
    "                best_tm.topic_word_, return_new_topic_mapping=True)\n",
    "new_doc_topic.shape, new_topic_word.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see in the new distributions' shapes that we now have 47 instead of 50 topics, because we removed three of them. We shouldn't forget to also update the topic labels and remove the unwanted topics:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:41.457184Z",
     "iopub.status.busy": "2022-02-08T06:54:41.456629Z",
     "iopub.status.idle": "2022-02-08T06:54:41.459010Z",
     "shell.execute_reply": "2022-02-08T06:54:41.459393Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['2_germany_german', '3_trump_president', '5_flight_air',\n",
       "       '6_report_agency', '7_company_reform', '8_intelligence_russia',\n",
       "       '9_party_vote', '10_al_syria', '11_play_game', '12_parent_child',\n",
       "       '13_economic_meeting', '14_say_committee', '15_northern_election',\n",
       "       '16_mr_damage', '17_white_trump', '18_minister_election',\n",
       "       '19_police_shoot', '20_percent_limit', '21_attack_group',\n",
       "       '22_space_european', '23_country_love', '24_industry_service',\n",
       "       '25_sanction_visit', '26_judge_victim', '27_man_new', '28_feel_mp',\n",
       "       '29_club_sell', '30_china_chinese', '31_company_investment',\n",
       "       '32_britain_democratic', '33_general_authority',\n",
       "       '34_america_nation', '35_force_group', '37_north_discuss',\n",
       "       '38_southern_deal', '39_russian_diplomat', '40_turkish_country',\n",
       "       '41_bill_house', '42_win_anti', '43_bank_investor', '44_note_7',\n",
       "       '45_day_morning', '46_support_provide', '47_family_refugee',\n",
       "       '48_body_north', '49_fire_death', '50_officer_police'],\n",
       "      dtype='<U23')"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_topic_labels = np.delete(topic_labels, uninform_topics)\n",
    "new_topic_labels"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Displaying and exporting topic modeling results\n",
    "\n",
    "The [topicmod.model_io](api.rst#module-tmtoolkit.topicmod.model_io) module provides several functions for displaying and exporting topic modeling results, i.e. results derived from the document-topic and topic-word distribution of a given topic model.\n",
    "\n",
    "We already used [ldamodel_top_topic_words](api.rst#tmtoolkit.topicmod.model_io.ldamodel_top_topic_words) briefly, which generates a dataframe with the top words from a topic-word distribution. You can also use the topic-word relevance matrix instead. With `top_n` we can control the number of top words:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:41.463163Z",
     "iopub.status.busy": "2022-02-08T06:54:41.462607Z",
     "iopub.status.idle": "2022-02-08T06:54:41.932160Z",
     "shell.execute_reply": "2022-02-08T06:54:41.931757Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>rank_1</th>\n",
       "      <th>rank_2</th>\n",
       "      <th>rank_3</th>\n",
       "      <th>rank_4</th>\n",
       "      <th>rank_5</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>topic_1</th>\n",
       "      <td>error (-0.8448)</td>\n",
       "      <td>case (-0.8618)</td>\n",
       "      <td>stage (-0.897)</td>\n",
       "      <td>statement (-0.9278)</td>\n",
       "      <td>detail (-0.9412)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic_2</th>\n",
       "      <td>germany (-0.1098)</td>\n",
       "      <td>german (-0.9475)</td>\n",
       "      <td>linkedin (-1.021)</td>\n",
       "      <td>tumblr (-1.021)</td>\n",
       "      <td>stumble (-1.021)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic_3</th>\n",
       "      <td>trump (0.02926)</td>\n",
       "      <td>president (-0.4054)</td>\n",
       "      <td>course (-0.9024)</td>\n",
       "      <td>play (-1.059)</td>\n",
       "      <td>fact (-1.187)</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                    rank_1               rank_2             rank_3  \\\n",
       "topic                                                                \n",
       "topic_1    error (-0.8448)       case (-0.8618)     stage (-0.897)   \n",
       "topic_2  germany (-0.1098)     german (-0.9475)  linkedin (-1.021)   \n",
       "topic_3    trump (0.02926)  president (-0.4054)   course (-0.9024)   \n",
       "\n",
       "                      rank_4            rank_5  \n",
       "topic                                           \n",
       "topic_1  statement (-0.9278)  detail (-0.9412)  \n",
       "topic_2      tumblr (-1.021)  stumble (-1.021)  \n",
       "topic_3        play (-1.059)     fact (-1.187)  "
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# using relevance matrix here and showing only the first 3 topics\n",
    "ldamodel_top_topic_words(topic_word_rel, vocab_bg, top_n=5)[:3]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "If you're interested in the top *topics for each word/token*, you can use [ldamodel_top_word_topics](api.rst#tmtoolkit.topicmod.model_io.ldamodel_top_word_topics). Here, we generate the top five topics for each token in the vocabulary, but only display the output for four specific words. Instead of the generic `\"topic_...\"` topic names, we additionally pass our previously generated topic labels `topic_labels`:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:41.937699Z",
     "iopub.status.busy": "2022-02-08T06:54:41.936978Z",
     "iopub.status.idle": "2022-02-08T06:54:48.904146Z",
     "shell.execute_reply": "2022-02-08T06:54:48.903720Z"
    }
   },
   "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>rank_1</th>\n",
       "      <th>rank_2</th>\n",
       "      <th>rank_3</th>\n",
       "      <th>rank_4</th>\n",
       "      <th>rank_5</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>token</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>china</th>\n",
       "      <td>30_china_chinese (0.4125)</td>\n",
       "      <td>37_north_discuss (-2.959)</td>\n",
       "      <td>34_america_nation (-4.323)</td>\n",
       "      <td>42_win_anti (-5.29)</td>\n",
       "      <td>4_circumstance_describe (-5.365)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>chinese</th>\n",
       "      <td>30_china_chinese (-0.3812)</td>\n",
       "      <td>22_space_european (-1.433)</td>\n",
       "      <td>42_win_anti (-4.893)</td>\n",
       "      <td>4_circumstance_describe (-4.969)</td>\n",
       "      <td>44_note_7 (-5.059)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>russia</th>\n",
       "      <td>8_intelligence_russia (-0.539)</td>\n",
       "      <td>25_sanction_visit (-0.907)</td>\n",
       "      <td>39_russian_diplomat (-0.9549)</td>\n",
       "      <td>42_win_anti (-1.286)</td>\n",
       "      <td>31_company_investment (-3.681)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>russian</th>\n",
       "      <td>39_russian_diplomat (-0.1902)</td>\n",
       "      <td>8_intelligence_russia (-1.083)</td>\n",
       "      <td>25_sanction_visit (-1.224)</td>\n",
       "      <td>42_win_anti (-2.033)</td>\n",
       "      <td>24_industry_service (-2.628)</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 rank_1                          rank_2  \\\n",
       "token                                                                     \n",
       "china         30_china_chinese (0.4125)       37_north_discuss (-2.959)   \n",
       "chinese      30_china_chinese (-0.3812)      22_space_european (-1.433)   \n",
       "russia   8_intelligence_russia (-0.539)      25_sanction_visit (-0.907)   \n",
       "russian   39_russian_diplomat (-0.1902)  8_intelligence_russia (-1.083)   \n",
       "\n",
       "                                rank_3                            rank_4  \\\n",
       "token                                                                      \n",
       "china       34_america_nation (-4.323)               42_win_anti (-5.29)   \n",
       "chinese           42_win_anti (-4.893)  4_circumstance_describe (-4.969)   \n",
       "russia   39_russian_diplomat (-0.9549)              42_win_anti (-1.286)   \n",
       "russian     25_sanction_visit (-1.224)              42_win_anti (-2.033)   \n",
       "\n",
       "                                   rank_5  \n",
       "token                                      \n",
       "china    4_circumstance_describe (-5.365)  \n",
       "chinese                44_note_7 (-5.059)  \n",
       "russia     31_company_investment (-3.681)  \n",
       "russian      24_industry_service (-2.628)  "
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.model_io import ldamodel_top_word_topics\n",
    "\n",
    "top_word_topics = ldamodel_top_word_topics(topic_word_rel, vocab_bg,\n",
    "                                           top_n=5, topic_labels=topic_labels)\n",
    "top_word_topics[top_word_topics.index.isin(['china', 'chinese', 'russia', 'russian'])]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note that the values in parantheses are the corresponding values from the matrix for that word in that topic. They're negative because of the log transformation that is applied in the topic-word relevance measure.\n",
    "\n",
    "Similar functions can be used for the document-topic distribution: [ldamodel_top_doc_topics](api.rst#tmtoolkit.topicmod.model_io.ldamodel_top_doc_topics) gives the top topics per document and [ldamodel_top_topic_docs](api.rst#tmtoolkit.topicmod.model_io.ldamodel_top_topic_docs) gives the top documents per topic. Here, `top_n` controls the number of top-ranked topics or documents to return, respectively. This time, we use the filtered document-topic distribution `new_doc_topics`:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:48.922508Z",
     "iopub.status.busy": "2022-02-08T06:54:48.906707Z",
     "iopub.status.idle": "2022-02-08T06:54:50.179005Z",
     "shell.execute_reply": "2022-02-08T06:54:50.179429Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>rank_1</th>\n",
       "      <th>rank_2</th>\n",
       "      <th>rank_3</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>document</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>NewsArticles-1032</th>\n",
       "      <td>9_party_vote (0.6419)</td>\n",
       "      <td>2_germany_german (0.1385)</td>\n",
       "      <td>18_minister_election (0.1008)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NewsArticles-1036</th>\n",
       "      <td>21_attack_group (0.3321)</td>\n",
       "      <td>6_report_agency (0.2325)</td>\n",
       "      <td>2_germany_german (0.1495)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NewsArticles-104</th>\n",
       "      <td>38_southern_deal (0.6322)</td>\n",
       "      <td>32_britain_democratic (0.2544)</td>\n",
       "      <td>24_industry_service (0.002519)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NewsArticles-1043</th>\n",
       "      <td>42_win_anti (0.6779)</td>\n",
       "      <td>10_al_syria (0.1851)</td>\n",
       "      <td>6_report_agency (0.0619)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NewsArticles-1048</th>\n",
       "      <td>50_officer_police (0.5356)</td>\n",
       "      <td>45_day_morning (0.2608)</td>\n",
       "      <td>6_report_agency (0.1974)</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                       rank_1                          rank_2  \\\n",
       "document                                                                        \n",
       "NewsArticles-1032       9_party_vote (0.6419)       2_germany_german (0.1385)   \n",
       "NewsArticles-1036    21_attack_group (0.3321)        6_report_agency (0.2325)   \n",
       "NewsArticles-104    38_southern_deal (0.6322)  32_britain_democratic (0.2544)   \n",
       "NewsArticles-1043        42_win_anti (0.6779)            10_al_syria (0.1851)   \n",
       "NewsArticles-1048  50_officer_police (0.5356)         45_day_morning (0.2608)   \n",
       "\n",
       "                                           rank_3  \n",
       "document                                           \n",
       "NewsArticles-1032   18_minister_election (0.1008)  \n",
       "NewsArticles-1036       2_germany_german (0.1495)  \n",
       "NewsArticles-104   24_industry_service (0.002519)  \n",
       "NewsArticles-1043        6_report_agency (0.0619)  \n",
       "NewsArticles-1048        6_report_agency (0.1974)  "
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.model_io import ldamodel_top_doc_topics\n",
    "\n",
    "ldamodel_top_doc_topics(new_doc_topic, doc_labels_bg, top_n=3,\n",
    "                        topic_labels=new_topic_labels)[:5]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And now for the top documents per topic:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:50.183862Z",
     "iopub.status.busy": "2022-02-08T06:54:50.183109Z",
     "iopub.status.idle": "2022-02-08T06:54:50.570658Z",
     "shell.execute_reply": "2022-02-08T06:54:50.570159Z"
    }
   },
   "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>rank_1</th>\n",
       "      <th>rank_2</th>\n",
       "      <th>rank_3</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>topic</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2_germany_german</th>\n",
       "      <td>NewsArticles-293 (0.8348)</td>\n",
       "      <td>NewsArticles-879 (0.4695)</td>\n",
       "      <td>NewsArticles-3119 (0.2692)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3_trump_president</th>\n",
       "      <td>NewsArticles-3140 (0.8479)</td>\n",
       "      <td>NewsArticles-19 (0.2168)</td>\n",
       "      <td>NewsArticles-3730 (0.172)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5_flight_air</th>\n",
       "      <td>NewsArticles-1539 (0.6423)</td>\n",
       "      <td>NewsArticles-1982 (0.6268)</td>\n",
       "      <td>NewsArticles-526 (0.4505)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6_report_agency</th>\n",
       "      <td>NewsArticles-1545 (0.8991)</td>\n",
       "      <td>NewsArticles-3125 (0.6466)</td>\n",
       "      <td>NewsArticles-2269 (0.4598)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7_company_reform</th>\n",
       "      <td>NewsArticles-2799 (0.8006)</td>\n",
       "      <td>NewsArticles-461 (0.5785)</td>\n",
       "      <td>NewsArticles-2613 (0.4432)</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                       rank_1                      rank_2  \\\n",
       "topic                                                                       \n",
       "2_germany_german    NewsArticles-293 (0.8348)   NewsArticles-879 (0.4695)   \n",
       "3_trump_president  NewsArticles-3140 (0.8479)    NewsArticles-19 (0.2168)   \n",
       "5_flight_air       NewsArticles-1539 (0.6423)  NewsArticles-1982 (0.6268)   \n",
       "6_report_agency    NewsArticles-1545 (0.8991)  NewsArticles-3125 (0.6466)   \n",
       "7_company_reform   NewsArticles-2799 (0.8006)   NewsArticles-461 (0.5785)   \n",
       "\n",
       "                                       rank_3  \n",
       "topic                                          \n",
       "2_germany_german   NewsArticles-3119 (0.2692)  \n",
       "3_trump_president   NewsArticles-3730 (0.172)  \n",
       "5_flight_air        NewsArticles-526 (0.4505)  \n",
       "6_report_agency    NewsArticles-2269 (0.4598)  \n",
       "7_company_reform   NewsArticles-2613 (0.4432)  "
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.model_io import ldamodel_top_topic_docs\n",
    "\n",
    "ldamodel_top_topic_docs(new_doc_topic, doc_labels_bg, top_n=3,\n",
    "                        topic_labels=new_topic_labels)[:5]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There are also two functions that generate datatables for the full topic-word and document-topic distributions: [ldamodel_full_topic_words](api.rst#tmtoolkit.topicmod.model_io.ldamodel_full_topic_words) and [ldamodel_full_doc_topics](api.rst#tmtoolkit.topicmod.model_io.ldamodel_full_doc_topics). The output of both functions is naturally quite big, as long as you're not working with a \"toy dataset\"."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:50.575527Z",
     "iopub.status.busy": "2022-02-08T06:54:50.574073Z",
     "iopub.status.idle": "2022-02-08T06:54:50.586942Z",
     "shell.execute_reply": "2022-02-08T06:54:50.586516Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>care</th>\n",
       "      <th>carry</th>\n",
       "      <th>case</th>\n",
       "      <th>cause</th>\n",
       "      <th>center</th>\n",
       "      <th>central</th>\n",
       "      <th>centre</th>\n",
       "      <th>century</th>\n",
       "      <th>certain</th>\n",
       "      <th>chairman</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.000256</td>\n",
       "      <td>0.000256</td>\n",
       "      <td>0.002820</td>\n",
       "      <td>0.005383</td>\n",
       "      <td>0.000256</td>\n",
       "      <td>0.000256</td>\n",
       "      <td>0.000256</td>\n",
       "      <td>0.000256</td>\n",
       "      <td>0.000256</td>\n",
       "      <td>0.000256</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.000278</td>\n",
       "      <td>0.000278</td>\n",
       "      <td>0.000278</td>\n",
       "      <td>0.000278</td>\n",
       "      <td>0.000278</td>\n",
       "      <td>0.000278</td>\n",
       "      <td>0.000278</td>\n",
       "      <td>0.000278</td>\n",
       "      <td>0.000278</td>\n",
       "      <td>0.003063</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.000465</td>\n",
       "      <td>0.000465</td>\n",
       "      <td>0.000465</td>\n",
       "      <td>0.000465</td>\n",
       "      <td>0.000465</td>\n",
       "      <td>0.000465</td>\n",
       "      <td>0.000465</td>\n",
       "      <td>0.000465</td>\n",
       "      <td>0.000465</td>\n",
       "      <td>0.000465</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.000160</td>\n",
       "      <td>0.000160</td>\n",
       "      <td>0.000160</td>\n",
       "      <td>0.000160</td>\n",
       "      <td>0.000160</td>\n",
       "      <td>0.000160</td>\n",
       "      <td>0.000160</td>\n",
       "      <td>0.000160</td>\n",
       "      <td>0.000160</td>\n",
       "      <td>0.000160</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.000162</td>\n",
       "      <td>0.000162</td>\n",
       "      <td>0.003409</td>\n",
       "      <td>0.000162</td>\n",
       "      <td>0.000162</td>\n",
       "      <td>0.000162</td>\n",
       "      <td>0.000162</td>\n",
       "      <td>0.001785</td>\n",
       "      <td>0.000162</td>\n",
       "      <td>0.001785</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       care     carry      case     cause    center   central    centre  \\\n",
       "0  0.000256  0.000256  0.002820  0.005383  0.000256  0.000256  0.000256   \n",
       "1  0.000278  0.000278  0.000278  0.000278  0.000278  0.000278  0.000278   \n",
       "2  0.000465  0.000465  0.000465  0.000465  0.000465  0.000465  0.000465   \n",
       "3  0.000160  0.000160  0.000160  0.000160  0.000160  0.000160  0.000160   \n",
       "4  0.000162  0.000162  0.003409  0.000162  0.000162  0.000162  0.000162   \n",
       "\n",
       "    century   certain  chairman  \n",
       "0  0.000256  0.000256  0.000256  \n",
       "1  0.000278  0.000278  0.003063  \n",
       "2  0.000465  0.000465  0.000465  \n",
       "3  0.000160  0.000160  0.000160  \n",
       "4  0.001785  0.000162  0.001785  "
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.model_io import ldamodel_full_topic_words\n",
    "\n",
    "df_topic_word = ldamodel_full_topic_words(new_topic_word,\n",
    "                                          vocab_bg,\n",
    "                                          row_labels=new_topic_labels)\n",
    "# displaying only the first 5 topics with 10 tokens\n",
    "# from sorted vocabulary list (tokens 120 to 129)\n",
    "df_topic_word.iloc[:5, 120:130]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:50.596790Z",
     "iopub.status.busy": "2022-02-08T06:54:50.596200Z",
     "iopub.status.idle": "2022-02-08T06:54:50.599225Z",
     "shell.execute_reply": "2022-02-08T06:54:50.598811Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>_doc</th>\n",
       "      <th>2_germany_german</th>\n",
       "      <th>3_trump_president</th>\n",
       "      <th>5_flight_air</th>\n",
       "      <th>6_report_agency</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>NewsArticles-1032</td>\n",
       "      <td>0.138543</td>\n",
       "      <td>0.000126</td>\n",
       "      <td>0.000126</td>\n",
       "      <td>0.000126</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>NewsArticles-1036</td>\n",
       "      <td>0.149498</td>\n",
       "      <td>0.000083</td>\n",
       "      <td>0.000083</td>\n",
       "      <td>0.232506</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>NewsArticles-104</td>\n",
       "      <td>0.002519</td>\n",
       "      <td>0.002519</td>\n",
       "      <td>0.002519</td>\n",
       "      <td>0.002519</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                _doc  2_germany_german  3_trump_president  5_flight_air  \\\n",
       "0  NewsArticles-1032          0.138543           0.000126      0.000126   \n",
       "1  NewsArticles-1036          0.149498           0.000083      0.000083   \n",
       "2   NewsArticles-104          0.002519           0.002519      0.002519   \n",
       "\n",
       "   6_report_agency  \n",
       "0         0.000126  \n",
       "1         0.232506  \n",
       "2         0.002519  "
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.model_io import ldamodel_full_doc_topics\n",
    "\n",
    "df_doc_topic = ldamodel_full_doc_topics(new_doc_topic,\n",
    "                                        doc_labels_bg,\n",
    "                                        topic_labels=new_topic_labels)\n",
    "# displaying only the first 3 documents with the first\n",
    "# 5 topics\n",
    "df_doc_topic.iloc[:3, :5]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For quick inspection of topics there's also a pair of print functions. We already used [print_ldamodel_topic_words](api.rst#tmtoolkit.topicmod.model_io.print_ldamodel_topic_words), but we haven't tried [print_ldamodel_doc_topics](api.rst#tmtoolkit.topicmod.model_io.print_ldamodel_doc_topics) yet. This prints the `top_n` most probable topics for each document:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:50.607039Z",
     "iopub.status.busy": "2022-02-08T06:54:50.606399Z",
     "iopub.status.idle": "2022-02-08T06:54:50.609766Z",
     "shell.execute_reply": "2022-02-08T06:54:50.609300Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "NewsArticles-1032\n",
      "> #1. 9_party_vote (0.641877)\n",
      "> #2. 2_germany_german (0.138543)\n",
      "> #3. 18_minister_election (0.100793)\n",
      "NewsArticles-1036\n",
      "> #1. 21_attack_group (0.332116)\n",
      "> #2. 6_report_agency (0.232506)\n",
      "> #3. 2_germany_german (0.149498)\n",
      "NewsArticles-104\n",
      "> #1. 38_southern_deal (0.632242)\n",
      "> #2. 32_britain_democratic (0.254408)\n",
      "> #3. 24_industry_service (0.002519)\n",
      "NewsArticles-1043\n",
      "> #1. 42_win_anti (0.677856)\n",
      "> #2. 10_al_syria (0.185094)\n",
      "> #3. 6_report_agency (0.061903)\n",
      "NewsArticles-1048\n",
      "> #1. 50_officer_police (0.535578)\n",
      "> #2. 45_day_morning (0.260814)\n",
      "> #3. 6_report_agency (0.197407)\n"
     ]
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.model_io import print_ldamodel_doc_topics\n",
    "\n",
    "# subsetting new_doc_topic and doc_labels to get only the first\n",
    "# five documents\n",
    "print_ldamodel_doc_topics(new_doc_topic[:5, :], doc_labels_bg[:5],\n",
    "                          val_labels=new_topic_labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can also export the results of a topic model to an Excel file using [save_ldamodel_summary_to_excel](api.rst#tmtoolkit.topicmod.model_io.save_ldamodel_summary_to_excel). The resulting Excel file will contain the following sheets:\n",
    "\n",
    "- `top_doc_topics_vals`: document-topic distribution with probabilities of top topics per document\n",
    "- `top_doc_topics_labels`: document-topic distribution with labels of top topics per document\n",
    "- `top_doc_topics_labelled_vals`: document-topic distribution combining probabilities and labels of top topics per\n",
    "  document (e.g. `\"topic_12 (0.21)\"`)\n",
    "- `top_topic_word_vals`: topic-word distribution with probabilities of top words per topic\n",
    "- `top_topic_word_labels`: topic-word distribution with top words per (e.g. `\"politics\"`) topic\n",
    "- `top_topic_words_labelled_vals`: topic-word distribution combining probabilities and top words per topic\n",
    "  (e.g. `\"politics (0.08)\"`)\n",
    "- optional if `dtm` is given – `marginal_topic_distrib`: marginal topic distribution\n",
    "\n",
    "Additionally to saving the output to the specified Excel file, the function will also return a dict with the sheets and their data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:50.637242Z",
     "iopub.status.busy": "2022-02-08T06:54:50.636753Z",
     "iopub.status.idle": "2022-02-08T06:54:52.373170Z",
     "shell.execute_reply": "2022-02-08T06:54:52.373831Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [],
   "source": [
    "from tmtoolkit.topicmod.model_io import save_ldamodel_summary_to_excel\n",
    "\n",
    "sheets = save_ldamodel_summary_to_excel('data/news_articles_100.xlsx',\n",
    "                                        new_topic_word, new_doc_topic,\n",
    "                                        doc_labels_bg, vocab_bg,\n",
    "                                        dtm = dtm_bg,\n",
    "                                        topic_labels = new_topic_labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To quickly store a topic model to disk for sharing or loading again at a later point in time, there are [save_ldamodel_to_pickle](api.rst#tmtoolkit.topicmod.model_io.save_ldamodel_to_pickle) and [load_ldamodel_from_pickle](api.rst#tmtoolkit.topicmod.model_io.load_ldamodel_from_pickle). The function for saving takes a path to a pickle file to create (or update), a topic model object (such as an LDA instance as `best_tm`, but you could also pass a tuple like `(new_doc_topic, new_topic_word)`), the corresponding vocabulary and document labels, and optionally the DTM that was used to create the topic model. The function for loading the data will return the saved data as a dict. We will only show the dict's keys here, as the data itself is too large to be printed: "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:52.378925Z",
     "iopub.status.busy": "2022-02-08T06:54:52.376967Z",
     "iopub.status.idle": "2022-02-08T06:54:52.392064Z",
     "shell.execute_reply": "2022-02-08T06:54:52.393511Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['model', 'vocab', 'doc_labels', 'dtm'])"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.model_io import save_ldamodel_to_pickle, \\\n",
    "    load_ldamodel_from_pickle\n",
    "\n",
    "save_ldamodel_to_pickle('data/news_articles_100.pickle',\n",
    "                        best_tm, vocab_bg, doc_labels_bg,\n",
    "                        dtm = dtm_bg)\n",
    "\n",
    "loaded = load_ldamodel_from_pickle('data/news_articles_100.pickle')\n",
    "loaded.keys()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualizing topic models\n",
    "\n",
    "The [topicmod.visualize](api.rst#visualize-topic-models-and-topic-model-evaluation-results) module contains several functions to visualize topic models and evaluation results. We've already used [plot_eval_results](api.rst#tmtoolkit.topicmod.visualize.plot_eval_results) during [topic model evaluation](#Evaluation-of-topic-models) so we'll now focus on visualizing topic models.\n",
    "\n",
    "### Heatmaps\n",
    "\n",
    "Let's start with heatmap visualizations of document-topic or topic-word distributions from our topic model. This can be done with [plot_doc_topic_heatmap](api.rst#tmtoolkit.topicmod.visualize.plot_doc_topic_heatmap) and [plot_topic_word_heatmap](api.rst#tmtoolkit.topicmod.visualize.plot_topic_word_heatmap) respectively. Both functions draw on a [matplotlib](https://matplotlib.org/) figure and *Axes* object, which you must create before using these functions.\n",
    "\n",
    "Heatmap visualizations essentially shade cells in a 2D matrix (like the document-topic or topic-word distributions) according to their value, i.e. the respective probability for a topic in a given document or a word in a given topic. Since these matrices are usually quite large, i.e. with hundreds of rows and/or columns, it doesn't make sense to plot a heatmap of the whole matrix, but rather a certain subset of interest. When we want to visualize a document-topic distribution, we can optionally select a subset of the documents with the `which_documents` parameter and a subset of the topics with the `which_topics` parameter. Let's draw a heatmap of a subset of documents across all topics at first:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:52.402993Z",
     "iopub.status.busy": "2022-02-08T06:54:52.402077Z",
     "iopub.status.idle": "2022-02-08T06:54:53.925231Z",
     "shell.execute_reply": "2022-02-08T06:54:53.924860Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 2304x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "from tmtoolkit.topicmod.visualize import plot_doc_topic_heatmap\n",
    "\n",
    "# create a figure of certain size and\n",
    "# Axes object to draw on\n",
    "fig, ax = plt.subplots(figsize=(32, 8))\n",
    "\n",
    "# randomly selecting a subset of documents\n",
    "which_docs = random.sample(doc_labels_bg, 5)\n",
    "\n",
    "plot_doc_topic_heatmap(fig, ax, new_doc_topic, doc_labels_bg,\n",
    "                       topic_labels=new_topic_labels,\n",
    "                       which_documents=which_docs);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:53.938119Z",
     "iopub.status.busy": "2022-02-08T06:54:53.936335Z",
     "iopub.status.idle": "2022-02-08T06:54:54.341394Z",
     "shell.execute_reply": "2022-02-08T06:54:54.341774Z"
    },
    "pycharm": {
     "is_executing": false
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAeoAAAEVCAYAAAAvqab2AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/YYfK9AAAACXBIWXMAAAsTAAALEwEAmpwYAABgeElEQVR4nO2debhe0/XHP9+bAYmITKZIBCHEFEQMMUTMKqgpKKWNoipaqub+SpUaa4q2lJpqHlpDSUWJMUpoqqrmUILKUENiTLJ+f6z95p77ukkuec97N9bnec5z3zO8Z3/vOec9a++1115bZkYQBEEQBHnS0NoCgiAIgiCYO2GogyAIgiBjwlAHQRAEQcaEoQ6CIAiCjAlDHQRBEAQZE4Y6CIIgCDImDHUQBEEQZEwY6iAIgiDImDDUQfAlRZJaW0MQBOUThjoIWoikhuY+t4KOioFu21oagiCoH2Gog6AFSFoI2FTSwpJWB7ZorRatmZmkbwB/kvQDSRu1ho4gCOpDGOogaBmLAb2BW4BbgYnWSonyJS0HHALcBfQA9pW0RWtoCYKgfMJQB0ELMLPJwAxgK+BJ4G2onwu80nqX1B/YBBhrZqOAq4BngN0kbVMPLUEQ1BfF7FlBMHckqdhylrQpsD7euv61mf1bUnfgXTP7tGQtWwK/ByYA6wGrmdk0ScsDuwKrAkeZ2dQydQRBUF/CUAfBfJC0A7Aj8BZwNtAJ+D7QEfgHsBlwjJm9VaKGlYALcUM8QdLleGVhVzP7n6QVgNlm9kpZGoKgmuqKbFAO4foOgnkgaTXgJGA80AUYDUzHjeYkYCTwxzKMdMHdPQgYDCyX/mJm+wMTgbsldTGzl8NIB/VCUgdJ7VJgY5/W1tPaFH6rnco4fxjqIJgLkgYCxwKXm9nFZjYSeAT4M/CJmZ0JbGFmt5YRAZ5egpsCo4CXgF8BAyTtlvaPAJ4F+tW67CCYDxsC50naGThDUs9W1tNqVLwKaSTGcZKWKu6rRRlhqINg7swAegHrSeoCYGY/Bv4O3CupPfBe2l5z919yd58AnGJmDwJ3Ao8DW0raK5W7r5k9Wuuyg6A5JPVMhumvwArAH4DfmdkkSe1aWV6rkIz0lsDpwG1F71rat8DGOgx1ECQK7qv1JK0HfAzsAiwJfFfS4gBmdiiwh5l9YmazStSyEt4fvpekhczsNbw1/zRurJeK7GRfT1rxvh8DrJFGO/wNGAMcKqlT2cGUmTMc7w4bL2kvSaMkHQu1qcSHoQ4CPuO+uhjvC74N7xc+HNgSfyEtDmBmz9a6/EJFYUmgi5ndCRwFTAaOkNTezCYBfwJONLO3vqqBPF/X1llzFJ8LST2gHA9OS0jdP+8DlwK/MLNv4kGWNyaNK0jatTW01ZPCPemdNt2FG+v7gDWBF4EV0oiQBSYMdfC1RtIiktomI90D+DGwLfAu7tZ+08yeAX4EDAW6lqHDEpKG4QFrN0u6FjfSo4Gl8P6v9mb2empdfyVJ9+FHktZsbS05kJ6LHYDbgV9JuqXeGgqGqZ2ZTQT6Alem7T8A3pQ0Aa/cTqu3vnpTvCfJWN8LHAbsb2bH4rEs6wLta1FeGOrga4ukbsCJwJDUgpsBPAdsD4zAf3RvJuP5BvANM3u5xhpWk/Sd9HlF4KfAd8xsc7yycBhwP94/vQTeZ/5VZzmgDzAsJXj5WiNpHeB4fKz8WNz1vFgdy1fBMJ2Y1jcBFgeuN7OZZvYd4FRghJndVy9trYWkwXif9HfN7D+AgJfM7EVJQ4ArgZ+Z2Ru1KC8MdfB15j1gNt6C3sjMPsBrwKOAvczseUkb48a8l5l9WMvCU0XhOuCd1Oc3DR/yNRnAzA7Gk5iMBO7Bf/gv1VJDjpjZeOBRPJr926kC87Wiqg/6Y+B8vDvme8DWZvaepA3qoSUZ6W2AU4DRFbe7mW2XtN6V1m8ws7/VQ1NrUHVPFgWuBpaQdCj++/ylpM3wmJZDzez2iPoOggUgubs/Be4GBgAnpQCyU/Af4O/TD/BC4KTk/q41i+EGaaFUZi98jPbakjqnYy4GPjazWSmN6VeWgnt1e+CHwP+AzYF96tWyLmhYW9Ka8glY6k5laJ6knYBPgJ8BRwKbm9nEZBB+VtawKEn9JO2SPgvYCzgHD5b6pqTzJO1uZnsAbVKr/ytN4Z5sCPwLd///FK/wH423qmeZ2fVmdk/lO7UoOwx18LVDUoOZzUxDKs4FfgN8hLu7e+Et2DvwH+APzey2MqJsU19fJ+By4C9m9hQe1f094EhJP8Rb8/+qddk5UQyQkrQo8G3gaDP7IfAT3MW6t3y4WqkkDTsCF+ER/2dL2qrscudCd+AXwOv4GPqlgQ0ljcArkL9OwYVl0AaYKql7MjbjgJ3wKO/18d/LZgBmtrWZPVmSjtxYFQ/m7Gxm3wW2N7MrgdfwHPyzyyg05rMNvjZIWhlPm/tc2rQ1cIOZ3SzpNvyleBzwf8Co4tCrEqNsH8GjaLeW9ISZXSfpLaA/Pjzr+2Z2b0llt3oKyFQBukbSRDM70MymS/oYb0n/1cweSFHwZwGzJJ1pZtNL1LMEHji4NbA3nib275LalDUUr1B2k3thZrdIWhd3dV8saRHgG3jl7nAzG1PG/Uv/6zOS2gJTJB1lZhdJ+gfwvpn9K7ndR0laBg+4/EqOPqiQgjg/SddhNnCbpH3N7BFJ2+GeuBPN7JFSBJhZLLF8LRZgD2BTYJG0vh9wGd7/DO66ehFvrXQrSUMlv/56eA28f1o/GZ9Cs3+droUyuB9d099ueIXl/LQ+FM+pvmtaXwUPplu1Dpp6AL8DDgAeAlYqaOpVh/I3wVvPq+ENqb3wCkvxmIY66FgLWAOPXJ4MfC9tb4PPIPdvPLiyVZ+heizpOowCli1sOzBdgwHp+R2QtpfyuwrXd/C1wcxuwJOF/Df1qd2Fu6q2TK3tFfBUnZdZSTNQmZnJ0y7+Bm8djZL0DTP7KZ4O9OzKsKQy3O2V8yYdQyWdJU/QMKiMsuahYSHgVEnLp2u9PbCBpNPx7Gsv4fNs34nP/32hmf27RD39JPWzxjiAU/AI5hckbQ6cCSxcUtnF+/wMHqh0MF6JvAP4QNL3KweYWSnu1YKedniFYSczewJ3eZ8paYS5V6EPMNLM/lymjtakEKvQHXgBnwDncEnLSGqDZ2R7E69cf2RmE6BEz1tr11ZiiaXshcZW7KLp74/Tj2xFYB08SOY+PDXojiVr6Yonh1gU2BM3SksX9p8BrFOHa7INXmnZHXgYOA1oW+f7siiwPHBkWu+CT37yC9y70RXYDlir5OdiQ3yO8fF4ZW0QcB4eL3AgHiMwrGQNQ/BW/L5pfbH0XN6SnsvbgIXqeG+2AJ4C1kjrg4BZ+NDBuj0jrbngo0FuBjqn+3ED7ulZHtgY744ZUBctrX0xYomlzKXwItwx/eh6pPUf4i69fmm9F7BC8TslaKm4E3+Hjzl9CFgx7dsaWKJO16Qd3g/fH9gAeALomfZ1Kft+FK9vMgCP4X2uFWM9Dvhtna7FlngqzN3wpBXX4AFD3fAo6wOAoWU8F4Vncwu8Jf2j9L/fWDhmHTy4sJSKQpWefsBhhfUfAb8GFk/r6+P95aXfl9Ze0nP5PLBxYdti+Hzwl+EBfjvUTU9rX5BYYil7wfvU/o6PlQZol/5+F/gUGFhi2ZWX8Wp4q235ZACeBTZL+4bg/V1r1vGa/BgfGvY3YJm0bYdksOrRB7piodK0OvBXmrasn6AO/fU0zvFdWb8EeBBYrsQylwFWS5/b4uk4Dyjsvx24aW7PUkmaGvAYjmuAB9JvZg98WNiS9dLR2kvl2U+Vo9ML16byzmiLt7D71vNaRB918HVgAN4nPEU+69SfJP3QzH6PG81S0oLCnD7pQcAfgfPMh2SNw4d4HCfp58BvcSP1VBkaCv1tq6UhaeAJGqbhmaXekLQ+3g/7jpXQByqfQOTE9HkIboz+JOlgvD96JLCVpOPM7H/A+lbO2PVqngY6KeVwx1/QS+LR/zVHnthmZ+BCSWuZ2Ux8aE8x1eRe+NjkzsXvWrIMZWBms80TluyNd81sjPeT/xQfI1wXHa1FIU6g8i74D7CipOXStfk0jV3fxszeNbMXoX7XQl/Bax58zSkESzWY2WxJewDfxF17VwKL4AEx/2dm/y1+pyQ9bfB+zrfNbNO0rReevH9R4FUze7RkDdvi/a7v4i69XwAb4cOgeicdJ5rZ7SWVvyZwEp6KdQl8+s7l8LSYz+EVqRXx6NrvpApNrTVUnos1gQ+BD3B35ll4d8Tf8Bf1z4FlgZvN7IwSdCyNT+CwOV5RXAbvjz4Y97qsC1yAj9GdUuvy56GroVJJS0PB1sIrLBea2V/qpaPeFJ6LHXAPwg54q/lwvJ/+GRonIjnUzB6uu8Yw1MFXiaof3cZp83l40oz3zOfNXQVP3bl32a02eQa0mclYPw08aWbfKrPMQtmVa9EGOAQYa2b/lHQF8A6elnIS3j//kZm9VlZlIbVYBgAH4a3ltdP2LXF3+6v4fZKZzah1+QUd2+Oegz/gcQt74kb5+3hlpR9eeRgILGZmo0rSsSSwL/6MHpT+HoL3fa6BV5puK6PsKh1N7nfhman8XdjMPmrt8fZlIJ9g5NP0eXO8knhgxRCnFvRgvGtqNnCRmf2xVcTWw78eSyz1XPAxr08CK+Nu1d/gfUsN+DCgf1OH4JyCnrbpbxs8sviWOpa9E/AXfJzyiIoePCjmakrsi01lVRoDK5PGm+IR9mcXjtk26Vm+ZC2rAxPwyO4R+Jj5F2kcy7483rrdHm9JrVayniXxzGt/SuV2xceMVyKta97/WbgfHVp67FdxwbO+nUnTnAoj0vN5AD4a40d4JbYT0L01r0n0UQdfCarGog7E+9V6AW8Dp5rP8DMbrxkfaDVMmD8/TeYt6rbmY1A3APpJWqsO5ffD3amX48N7dpU0zLxf9Hv4teg89zMsOGZmKXPTnXgKzH/hgWxd05hpzGw08BMrwd1dxbt4trFewKF4RPUtwL2SNknlf4Bfs33NrNTUrebdLlfiwWtX4JWmZ83sn2l/zVuwhfvxe0m/lNRsfIY8O5lJaps8MjWj7Oe+hbyD59HvIamSP2E43uXQFk8tvBE+CuJ9S10QZdyTlhCu7+ArgzxH86rAFHy40zL4VJUvyaeS7GEl9DlWaVgLb63/opl9FTd46W7EZKQvAiaY2Y/k0yLuhLuZrzSzm8ssv6BjJXz86Ugzeyhta8D7P48C3jKzw8u4JgX3bU9ghpm9k7YfiU90coGk/fEW/W/NbGza397MPqmxlrmeU9JSwD5418T4WpbbTFnr45nPzsbHiD8H/MbMni0c08bMZknqgvfLfs9qkACocD8WscJMdPV0q6cuh3PxOIiPUjDnevj1nwW0N7O3JS0P3IRPY/mPemibF9GiDr4SSFobn7v5DtxttQY+1OZVeb7kI4B/llh+pZWwOB4s9RmSkW6XXlYNyWCVgnk+88eANSX1N7P38Mjz24DvSVqizPILfAI8bmYPSWqXDNZsvAVzJj4mtczW4zB8vPr5ki5Iu94DNpJ0CF5ZOMXMxhY8ILU20qvh/eHI82dX63wLuKBipMtqcUpaDvc03W1mt+CVtkWAg5VmJ0sBZbPkUfDXJ101ydJXaM3fIOkX8slFKtvr1cp+Dx+Seb2k9niCoUfxSm2PZKS/if9WTs7BSEMY6uArgHz2pQPxhCH/Mh/m9DP8RXQtXoM+3szuKvGFUHkBvwIMkrRbMzrbmA/z6IIHrnSoVeGV/0vSupK2ltTNzI7CI5lPlLSK+WQW1+N91W9byakoK9LwCUe+aWafmtkn8uFZhwBPWQlD0grXogMeWb8/Kcpc0nlmdjE+RG4V4LgyXc2J9YFjUhkzm9HbYGYfV4x4iToWxoMHt5E0KD0PP8L7aw+V1NF8lEQX3AtyspndV6vCJQ3EK9PX4ZkBN1MaslePFnVquX+IR/l3wbsdPgFOxyvxp0jqg4+KOMTM/lTL98UCnavWnd6xxFKPhabZrRrwALI78Yxb7dP2ZfEfZO/q79RYy8p4hWCttL4ZPtynS6VMoE36uzg+hnloCTp2xIOlfo2/aL+Rtp+MexpKn9SiuXuEd0O8ivdN74MHau1Uctlb4kGE19I4scYS+Pjti+b2LNVYQ/vC50vwbpjqZ7f4XNxAmqikxtd/Nbw/ful0DY7BK4rrpv0dKASw4ZW5zWt8LXoDL+PeC/BZydbBo+9LDdqr0rET7mE5GU/scgc+hn0R4Jd4S7p9CeX2IWUy4wskFIoWdfClxMxM0raSTgCOxVuOv8YjZw9PLubXzex/ZvafyndqVX5V7fhd4B/AWZLOwsdhtsGH91ihz29x/GV8ktV46srkuhyJG8VH8LG4u0jayXzCj3/hL6O6kf53mdndwLfw+IF18Uxgt5bo4l0XHwv9X/wFuaWknmb2Nh5Et5zSxCcVnSVoWAc4MMVGgBuHFYrlFZ6Lznh621FmNq1G5TdUfiN4d8ehScPa+CiASXgrej0z+8CaehX2txq2pNN5/4NPrnKQfCKWGeZzWHekxIRDRVJQ3N7Aaek3sSPwFnAV3j99Mv5s1rrrox/uXv8teHKZz/3s16smE0sstVzwXLwv4MMqxuKpIFfFW1KXAcfWQcOmeN/3qmm9Fx7VfTv+AriOxpSEbfFc0kNK0NGAuy/XTJqexPN4n4UPB9uzxGuwMGn6P9wofmbiCJppQVBeK7Yfnp5197S+He7iPJjG6Uxr3mKq0rA1MAavFNyZns0DcHfvdlXHdsE9LJvUqOwehc/dgbuBTdP6cNxID8Jb1z+lqjVbq/tCY2t+XTzTWoe0/jPczbwp3tJ/AVivzPtR0NQGDxCrTHzSFvfEvYIPkStjOFx/vOK8Oz764vgvdJ56XKBYYqnlQuPEFj8sbPst8If0efvqF1AJGjbGXbjX4u6zfUhzWOOJM4YkTb0K3+lew/IrL8L++PjPTmn9IODo9LmSu7k0lzc+89QRuFfjWVLe8Lkc26YOz0ZnvOL2WGHbNukFfSglz0CFt1j/SiFPOd4fPxJv4Z9cuX/JUFxMjdzM+GQrl1AYj45XUraj0cV+BHBH5Tkt+VoMxcep/zkZq7XT/30a8BHuRai43+syPhkYBkwEtkjrm+ABZYNKKKtTqgAclNb3TO+ESl6FFv/PpV+YWGKp9ZIM8V9xl1WfwvaH8ZZCqT96vCVwb+Elsx8eNboPhRmw8NbMTiXq2AJvjU1JBroT3j8+EzgeD4qpeV94MzquBGZQmHmpmWMqhmJRYHANy65UWJahMRahPe7N+GPhuFKmy6QwNSjQE58V7QWaSSiSjMTTVTq71VhPB9yzcUZaPyktFa/HYHzIVVkejcr9WASvKK6f1k9ORmvttH4Y7g6uWyKRgrY9cO/GeXhK21JmBMO9TX0K60vj3oQffN5zRR918KXDzO7EAz8qEcX9Ja2Ou/raWvpVlEHq5+qJB5DtkfRcgSet2BrYXlIHeT7nXngWtDJ0DMKTM/wY74/dAndx348bpUXxF0JN+8IL5Rf72C7DM4v1krS5pIXTMQ3pb7GPfjSeN7kmGszmDMG6HRgl6WLzPsYDgPck/QXAzO6yGg+1kdQO+IaklVKf9zeB+/F4heMldUrHVaK5b0/71k7rn1htxid3SEONAHrgLeutJR2O/06WB34h6dd4HMftZf1G0v3YEY+q/zGNv5Gf4v/7GSmGYBTu+biuoL1UKv+zmd2Ae1muxnMe3F1SeR+Z2SswJ4fCm/g12SJFl3+uk8USy5dmoWnE7DA8SvPB9HeH6mNqWSbeSqi4rbbFjcOhheP2B1YvrC9e4nXYF7i6sL4r3pL7Pildah3uxabALsCAtH487tpbB684HEjj9ICLU+gvXcByi9HUm+GR7svgfcIz8UpDG7yyci3J81HC/78QHhPxD3wGrD5p+1Z4QpH/wwMKwSuVvXBDvlKNdWyOR7jvgrv4l8CHnj2CD79SulcHABsWn+kSrkk/fPatb+B94n/Ck91U9p9avB/U3qPQrgXHfOZ/r8fvJZXTB/f2bPG5vlcPcbHEUsulylhvkR78QylnWEXFSA/D+6KvpzFQaRs8BeWRVd9pqNZZgq4BuOt/vcK2y5Oejcsqv3A9BuJ9fVcnHZU84sfj/a6TSG5/3CX/1xoZ6S54haTS7bB20rId3vXRF3cv34K7HkvtF8fjJf6TjOI6aVvbZDwvAE6h0e3fjhRLUIKOP+PdD9sUtq2CD0E6pcxrUCivV7oOV6b1hXAv003Vv5ESyl6y8Hk73HOwIwXXc9XxlXvStkxdcyn7BNzT1uIKdV0FxhJLS5f51XqrjPUO+NCPfUoy1tsATwAr4dMRzsJTEFbKvgOfsrFuCfuTwTobj6LdB89LfB+e3OWGlr4AvmDZWyZjXOlv/Abu/v5uWl8KWLnq/tQsshdvJb5J8l7gfb0X0ThO+XA889nKtSqzuWeTFJiGd7kMxyO8t03b+qRtpWio0rE4HqB1XXoWOxeOWRV3Ma9Sj+cT+EGqHGyW1tvhMSW3pWtSRuWxDV6B/l36fx/Bs95dincLrVp9fOG6nY7n8y71ulTdL1HwvLXou/UQGEssX2RJP/Cz8WCYfmlbm8J+0bTFW1M3WqGc/fHc1DviQWS74FGr3077axbNPR8dqlpfFvck/CG9jNfBW3KX0QIX4ALo+D4+occeab1zulfXAkeUWG7lBdsDd6NPKhjrI/HKw/eAuyjJ3V3QslMyzLcWNByUdB2Pu3zrYaR3wnN3L5PWL8RThIKP296RNENUmc8jTSvRP8RjESrDwtqV/RvBE6rcgHdDbJO2bYLPu35S4R5VnqHO6TezWZm6avb/tbaAWGIpLoUXUH+87/EwvF/rv1RN/1f40S2GB83UxK1YOH/xRdQlvXwqbuWb8JzBS9XhmvSch8ZiNPXOeMt/rZLuybI09jnvj+dNXjOtd8ZbzmuUfC22xyNn98RnnJqKt6J6AcfhXoUdS9bQH4+L2BKfpnIyaTggnrb2TlK8RMk6NgH+TmEMNu7uvwgfOvg05WTAW4zG6SHnZqxHpms0pMT/f1GS9wD3GPTGg9juLByzEV7Zn/N+wFvS91V+yzXQUfl9bIoPxfsGcxn6RlVXDC30MJT6IMUSyxdZ0o/rHuDgwrYf4Ia7R1ov1owfoEYJIwrlDQPOx1unPXH36rl48oZt8aCpNetwLZZJ5S6e1qtb1cUX5TEkd3QJOobh/cznAb9I2w7DE7tU+ovrMU765xRa7XgU7X9oTBNaSaxRZrrYq/HJKirbDsdd8QPS+sJlaiiUexhwQvq8UNWzsBNpaFSNy1wMrzgf0ty1rjLWh1NiMhM8h/oDwHfxIL0+6bd6N/DrwnEbk7wb6TpdW8L7Yis8j8BwfArNH1DV/VR4Z3Wp3LeWLjE8K8iRSbiB2grmDO+5EG89tAOwxuE+f8QnVniwVoXLZ+I6Fa8szMKHkqyBB4AMSeujLU0oUfLMPx/g0/DtBc2mu5wzBAr4lZn9vdYCUnrSk/E0oO3xSUc6mdn5eB/ffWkoUrW2MvgQbz1Xhn/9Bh9H/oikRYGPobS0oEvgWaz+i6ch3SANETsHr0w9IJ9OdGYZGpp5zmbjXg6AT83MJK2fJmC51cz+VsvyE+/jFaO+wJ6SOqRyKzOPza4MyzOzc8zs8RI0kM7/Nzyw8BLcML9iZpNww72cpCvScQ+Z2fPp88d4YFtN3hdyFsIN9HA8n/lE4BYrTHpTNUTxRrwfveWUVduJJZbPu+ARvAPS5yXwFvTJeE15I3wITKWvuj0ebbxZjTWsjkdPn1LYdgLen9WQyq30B5YZ1b0ksHT6vD7u5u1ddUzRq/BE5drVWEc3PMPU0XiE/aPACmlfpStiuTo+I73wilwl+9pgPJBqYMnlLolnedsdryyehQcsDSoc06cO//+GeBzC8njLbCKeFa5LuhYvUUJLOpU9ZzQD3vUxCjeKzbWsK8MYFy7ruuDR3VclHa8DfQv7euPxA2tU669BuR1oHIq3Ep6v/PtJy2OFfd+i4PqncYji527Nl/pQxRJLS5f08nkJ7zv6JbBiejk+nozQWcD2hePbk7It1VjHWnhqwzlBQmn7ndTB1Z3K6okP7RmDu9nXwIOEKv3BqjLSf6VG/W2V86e/m+F98asnA/08qU8+vSRvoekMYWVnu6r8zyum5+KyZKhK6Q+uMjw98Oj6K3C38kK4N2EUjdm3ShmWV/j/N8Arq9cno7AFXqG9F88ON440Y1qJ92AZGo3wLum5bGKsaRpVfRcFA1pDPYPwlmklZuREPMvYonh/9Y9IXRAllN0f7xY7Kf1/y+NBjE/SmJp0LeAZUowAXmF5mC/YsKj5PxFLLC1dCj/+RdIPbW08zd6JuGFeEW/R/Q04q/C9mveF4u7ldVJ5S6UX8nG4+30APk1jvzpci8rLvj0emPRHvEU/DR8ru2jhO53Ty7lmRrpw7i3wysKWaf0nNE4usQUeXVtK0FbhWvShapwrTSso3Wls3ZdVSdiUxjSXPfAW9dX4+OBF8OF6pU8fins1zsYnXmmHu1mvoTHCeSEa04SWdS22T7/F84Fr0rb98KjzgyikTcWN9Jgvapjmo6MrHiz3p6rn5Re4K/yflBjMh49/PhefNa/i2WmXfi+X44mQnij+PvBcAn2+cJllP2CxxDKvBW+dXIS7jCotk/74+OAL8ExHy+K15aNL0rAx8Hb6kd2EJ9DohbdaHk8vxO3TsWUmEdkR72+7isaZnpbF+wN/g7dg+6ftDXiUd02DYgqafon3gW6W1vsmfbfgySSGlXU90nk3x4c4fWaSj+bKpMaVt8I9GYVHlleM9VLJYD6CR7nXazKJ09L9qGQWWwJPz3kzjWP6y+yKWQefiW1lfEjgM8k4CfhOeiYqFYXOeFdRKc9mKmNnPDZhn6rt6wGrlFRm0cOyR/qN3AzsXNg+APeA9a/+zgKVXY+HLJZYmlvSQz0u/eguw1utXdO+1fAacmXYSy9S66lGZVdexF3x/raNkvH7Du7mXg9vsV2Jt6yXqFXZc9GzPV4LXxGPYH2aQgR30nYhPpf1nG0lXI+FC9t+hVeQulYd+7ln//mcWtbBo8vnmhKWxpb1QtRwnHDhOnQubPt5ejYrs6PtjVecBpT4PMztfrxW+I0sgQ9TW6ssHYWy10zGaQe8Vb182r5W+jsnbgP3vmxUwrVYH09LuxUefT4Ej2PZqw7/fzEj3zeSMV4E74e+Ha/s9wYOKaX8sv/BWGJpbsH7da4AflfYdh7uuqoMweqY/paS5i+9dG7G3bi7pW2dkuEei/cJ9k/HHEYNs54lg3xMYf209BLYMZV9Dt7/WsyLfBDeP1nTqRoLL6EdcC/GOaRsTXgXxAvUIakLjW7/s1OZ36GZljJN+0CvoSrIrgY6tkv3/BoaW6s/w431SDyWomYzgH3B+1H5jZSS2IaqyhFecf4vKfVl2rZJuk5LVB1b898r3tUwER+NcR9eeRqAdwm8BHyrDs/nNnjU/3l4RfrgpGFfPIbjFWCrUsou+5+LJZbqBW8JtMUnLRhNIQAGzy71Ou5WKzMN5iDgL+mHfine71gZa7kYPsfzoLS+JvOYZ/kLlt8TH0p0YlpvQ+OkDRUX4j/wcaKLpuv1PUpKKIIHrU3Ak4f8Ew+SWSvtq0TVtq1+gdeo7IphWqqw7QQ809TKVccW+6jvpfZR/2viQ2w2SS/iU4BT074ReGtx27Key895P0r5jdBYYdoG9+Icjwdo7YpH3G+GDxd8ipKTyyQd7fF+8YqHZY307jg8re9AuW52pd/gdRVDnJ6PS4DhaX0lSsphYBaGOpY6L7g7+f70Qxc+tOTs4suP1L9TooYe6cV3bWHbKDwV56ppvWIQyngRVs7dG0+U8fO0vlB6GQzDW/O/J/VJlngtGtIL/zc0Tm4xLr2EHqZxuFwp/X4FHdsDz6X7sE/adipegepfdWxnvAJTRhDdlsClhfX18QrDKmm91Aj3dO62eJ/vuvW8H7iHYtH0eVu8T3orPC7h+rT9u+meXEZjXvN65BA/Pz2jlcx4Q/A4gcULx5SdYOYiPFVtRcNu6Z40m4WspmWXXUAssdA0CGMhPIBsDF5Db4eP0f01qWVdhnGs0rMQHq36LLBfYfsl6aXcocSyKy/6yo+9N97veEpaPwKvMLxGfQLYilHmy+CVqMrUjC/j3ROLlayhO57pbRu8Avdr4Ptp36/wAL9KN0gb3IDXOv3jKnjg3gp4S7Y4FPAaUtdIHZ6LijelXdJzXz3uB95y/xONUfSH463EbfE+6d5VOsusqFTKWInGANN1cbf/vmm9H+6NK6VLpqChN41Z73ZPGjZJ66vjrv+OZT4bZkZbgqBkzMwkDQReMrP/SboL+AQ4Cs80dS7uUnw5HT97buf6IqTsUSZpfXzIz9t4K+Ed4DuSZprZ1WZ2gKTVzOyDWpZfJOkYAmwm6UV8LOi6wBOS/mdmZ6XsVj3N7N+V75Ss4zm8hfoOngltkKQ3cdf7r83svVqXX9CwCd7Pt5CZ/UVSBzz71XaSfmhmR0jqZ2Yz0ndmSTrdzN6toYZheJ/nAbg79wJgF0m98b7HNXGvT2kkHTsAJ0vazcxekvQBnk9+fUlvUNL9SFntfoxHck+T1AcPsvwDPvnMLmY2SdL2QB9Jl+K/31JI12I7/D68l57PH+JdANtJ2h8fRnmymU0pUcMO+LvpNUkvpfdDX+AQST/CR0L8vPJslkrZNYFYYsGHtFyHD8GqRKwugvczPYu3sMt2W22FB3ucjAeCHI5HF++M93XuV6drsSnwLzySfDReQ++Hjx9/H/hlK+k4HW89HYiP136JkpJnFDRsjAdG/RyfS3nPtH0h4Jt4vELvwvFldEOshBvjYtDekrjLeQzemv5mHe7H4HQ/KnNad8U9DQfgU0SWej/wytIT+FC0JfBENg8B56b9m+FdE1vW4Vr0w1uqle6G23C3d5f03lifxniSskYerJTeWevgXXQTgIvTvmXwfvEBZWpooqfsAmL5ei40uo42xIO22uJ9PGNpjGDdDQ9WKa0fNv3IFsZdhjumbWslLQel9d0pOQVlKmfFZBT3LqwfjbcMwOe03qIVdPRNOo7B0yEuw+ecL/cLaOibKgS7pPVv4B6VytSZC1HykLhUzvLphdwBdzdX4gfap2enMud02RXJrdP13waf0GE8cAZemVm6DvdjrWSkH6Ex29faeB/srXg+gVIrbqnMznh3xz9IQ7zSfbgF9z6V7mbGY1huwCvwxbSk44E/ll1+c0tMyhGUgpmZpC3xYU2jzGymmR2Ep6G8WdKx+IvoSjMbV+vyCxMYtDezj/CgrXUkLWJm/8BbSiMkdQRuMrPxtdbQDCvgFZZ9JHU2s5fw1sJ2kvqY2atm9teSJ/loTseL+FjQ3fCxwm+Y2dMla1ge72/eS9LiZvZnPJHGhZL2MrOPzeztkjWAX4fOeD/kp+au9Y3woKFFzCdxwNKbukRm4BWonwHT8aj3DvhQpzfLvh/pN7Ep7nE6TtIO5hO8bIznsd4p3aNSMe/SuBbvitlC0hrm7IJXuFeqg4bJePKjqcCmknqm7QOBFSWtU4ffaBOijzqoGZIWqrzYUj8reKrDeyvHmNmBkr6LB8T8wEqY4afQJz0M+IakY/A8vOvjL54x+JjQqXgLquyXcIV78D7g4cBRkk7B+/qEz9IF1MUoNKfj46SjntfiIzxf9I8k/crM7pT0Hdxo1QUze0HS3cBvJV2HX4cj8KE/pcUqNKPjYUnPAp+Y2fuS+uHP6tV11PAvSa/j/b/fSzNj3YAnvSkFSQ1WFZNiZo9Laos/GzummacmmNmwsnQU9ChVDO6U1A6vvM6W9Fcze83M1ixbQ7O66veOCr7KpBrmnvgQjxfwIIuNklG+EHfpfmZqt8oPowQ9m+MutCNTK7Ut3krqh/fBLY1HWt9c67LnoqdSeRA+dnsk7v59EzjTzO7+uugoaGhIGrbFKyqnphZVmc/FHMNQLEPScLxFuwyeQ/qeEsrugBvimVXbizra4LnUT8Oz0N1aax2Fcj9jJNP2xXAjuSfwbWByre+FpGXwqTknz+OYDfDMX1PwYL4ZJT0TbYv3pOp+DMNHiNyFB9d9UseKfaPGMNRBrUjG8EU8u9eOZvZw2v4d3M29m5ndXyctRwHvm9lvUsvgA0ntcQPdC3jHzJ4uyyDMRVPRSA7GX4Yzzeyo4v4alVX98vmMgSpbh6Tl8DHQdzWzr6hhGzx4a5SZvbCg5VaV0xcP/JkM3Gdmb1Ttn5uxqulzIWlVfGTDCdUamjl2TTxwbkKNn4lV8HHQHwOXp66XuR27GD4+uOataUkr48Fif8Rd/WruHqRjNwKmmdmzNdawEu7R+1Fab2Nmswr7i8Z6Z2Bi6h5oFaKPOqgJBcNwGfAW3iqovAgvw4di3SWpS2pJlU03fFgNBRfmasDH5hPJP5321fJl3EfSgLn1X1UMUyrzETyneDdJx6QXRa1eyCsDl0k6QdIBqew5L8J66JC0Oj4ut0dqJVa2q6Kh8PduPKCu1kZ6VTwWoRfe/zqyGR2zC9vm6Kzxc7EKHsz4UHMVherjzewpM5tQSx3JlX4lPuPTwrh3aa4a8EpuGUa6H94yfQmfTcrMbHb1b6Zwfx4pwUgLn7L0MPlQM8xjE9pXjilUIjGzP7Wmka4IiiWWL7zQ/IQJC+NDPc5M6xvgKTNLzxdd0LAMPi70uLS+Mf5yGFRSeX2BmXh07LrNXZfmrhnu+q1Z7m7ctf80XjE6GH85b15PHen+X08Lhrzh/eJNkq/USEN3PLJ9/7S+K3AmHlnfo7o8GqO9O5HmEK6Rjs74fOGXFbZthXsylpmHjs7UcFgYPgTu+PR5XXy40xAaE5y0aUZDra9FH3w4ZiWy/1FSeta5HF+KjnTOtfGhifcDt1c/k4XPlbzmHSkM4av3Ei3q4AuR3NxY5clOtU9J7cyjrLcDtkw11lvwHNVTisfWQENvSbtWnzO1Ct/AJ7jYQ9KVeCL9w8zssVqUXaWjDT7O9Mf4sI6fAevO4/9sSN/rgM9IVKtW08J4Yohfm9kZeMvlXeYeKVuKDry/WcDf0/mvkHSRpKsqrdZCK67BzEzS4sCvJS1SIw3v4n28l6f1H+HG8UfArZJWtMaugDbmLarFcS9ATZKpJKbjFYaXJe0h6U68AnUI8EdJK8xFxx/xgMcFonCd3wO6SloX/y2siI9Vf1RSf0tu37KuRfotTAGOMg9QA88h3l3SUs0cX3MdkjoVVhfBp5LdDOgs6W5J96TfQruChpkFDR8vqIYvTGvVEGL58i54usXz8YQlqzazvzi70R7AeiVpeA14pLBN1Z/xmnAPqlIglqBnOdLUiMBP8WFX61E1kxCNNfTF8Wj4TWusY30K04HiwUCXNnNcaTpwI30UPj79JDyorw/uYr9lLs/JX2uhAfconIyPSV4Br4xsjFfSKsecBRzbjIYxtboOBR1H4EkzjsRTcf4q7W/AW/g/mYuOBZ5kImk4CU/uszk+Dvl3wJjCMSekbSrxWqyCp4U9isIkK3h6zvupmvmKxrS2NdOBDwd8GDiwsO236W9PvCLz7FyezTHUePKXz62/NQuP5cu34DmBn8STMlwPXFG1vzqHdNF41sS1iU89+SA+HeW9eC29ueM+U16tNKRzrUJKVpLW2xc+/x8+NrnSP/qDwr4utXoZF3T8orn/EXf53po+DyK5HWutI2n4eWF9TzwT3dXAioXt9+A5rFWShn/hY/dvxvOHf2bKxWQwji2sL5pe4rU0TBUdf8IrBkvgE34U782xNJ3qtCPuDq6FYSpquBWfBaxj2n5i1X0aVeK1mN/7Yjc8kcgKVdtrrWN9vEvoNnwkCHjO+OPw7qof4x6gWwvf6YSP5y5tZq4W629tAbF8eRY8W9Ro4NC03g2fMGFP3JW2SNXxlVZbm+ZemF9QQye8Fn5wWh+Gp5rsQjOtZRprxu2AhWt4LfrgmbReAq4qXqPC50PSS+otYFja1j4ZkWb7jWuoo336uwEecd83adm81jqqNPyhsP0gfBTAPnjLZK2kYakSNCycnsXK1Idt8Sx4B1cdtw6eDnJoYdtQUjrIknQ8SOorLxy3bjM6VqcGGcjmouF+fJrO7vi8zr9Mv9u/03Sa2Vpei3m9L/riFYc2wFVUZT2rsQ7hFaVL8H75G/EK7Nb4mP0jCscOLnzekxK8gV/of2htAbF8uRZSQFj68T+VXrS/w5PX7572Fd1oXXA3eZcalS88WrSyvgrecqvMVVtswRfdV7+nMN9xDXQsj7u4O6aXT9FAVdJODgBmV7SlbYsBS9dDR9rfH+/f+xepslBrHc1oKE4f+l28ZXsRHmC4c2Ffpxpfi3XwHNmV+340TV3e2ycNlUpTKTNBtUDHJvjUlcPKKH8eGiqGuw9uHC+o/t2UcC3m9b7YLe37P2CDqu+VcU0uxCsA6+FpjacAP0z72leXTcmz+H0u7a0tIJYv70KaRCF9/inJhUaj27sz3vdYk9ZjM+VXytkfb7UsW9jXpkrDkBLK75D+dk8G6urCviWTkZrzMi7j5TMXHcVKw/p4UNLQio46aSga65XT9lVK1tCman0EaZITvL92e1ILqSwNLdCxCp7gZZ0ydcxHQy88B3+bMjU0o6n6ffHrOpVbMbynAhvhMSsTce/OiHpoWNAlor6Dz01hfOF1hc2PAR3lyUVmp0jJW/D+sPvK0GGNY2DvxGvrfSv6zCNGO+P9cz8zs7EllP9B+jsFj+RdWNKvU5KG7wN3mNnt1WOH66BjkaRjQ/yFvI2Z3VvrJB7z0dBe0m+Thu/i43KfTceUpaESuTwnzzvwcUogcifwlpk9XqaGFuj4M/C2mT1Zpo55aFgLj+uYXTmmzGtR1NDM+2IRSR3KzqtQ+P/uwLujHsNb9QcBwyQtX2b5NaG1awqxfPkXPLnJU8D2ab0dPrykphHN89FwLvBg+tyAu9p+S50DQfCXwGzqMDViC3XsnIGGVrkW6bn8O57UZVhraMhFRw4aqrTMeV/Usdyu+ExplVnz2pPG1Oe+tLqAWL68Cx4Isgoenblj1b7SpyhM5RT7pKv7ubrW+Xqsgw8Z+0x/+ddNRyYa1sMn/tiuNe5DTjoy0fCZ90W9nwugU2s+C190iVzfwQKRElgsYWZvJhdXgxVy5tZJQxOXruaSv7kOOvrh47XHlO3uzl1HJhra4kktJpbp9v8y6MhBQ9LR5H3RijparewvQhjqIAi+8uTyYs5BRw4ags9HGOogCIIgyJiI+g6CIAiCjAlDHQRBEAQZE4Y6aBGSDmxtDZCHjhw0QB46ctAAeejIQQPkoSMHDZCPjgUlDPXXBEnDFvAUC/zA10BDLjpy0JCLjhw05KIjBw256MhBQ046Fogw1F8fWv1hIw8NkIeOHDRAHjpy0AB56MhBA+ShIwcNkIGOiPr+ipNqg8M6der0vZVXXrm15QRBEHypeOKJJ94zs86tqSEM9deEgQMHWpcnWzel7ZjZN7JVw+6tqiEXHTloyEVHDhpy0ZGDhlx05KAB4B676QkzG9iaGsL1HQRBEAQZE4Y6CIIgCDImDHUQBEEQZEwY6iAIgiDImDDUQRAEQZAxYaiDIAiCIGPCUAdBEARBxoShDoIgCIKMCUMdBEEQBBkThjoIgiAIMiYMdRAEQRBkTBjqoAkDtxnA7/99Hpc/fwHDj975M/vbtW/L8dcezuXPX8D5405lyeV6fGV15KAhFx05aMhFRw4actGRg4acdJRFaYZakkk6u7B+pKQTSyxv51TmKvM4ZnFJhxTWl5F003zOO1bSAidkl3SKpNckTZ/L/l2T/oFV23tLmi7pyLTeT9KEwvKepB8tqD6AhoYGRo4awXHbn8IBqx3O5nsOpveqyzY5ZtsRQ5n+znT2X3kkt5x7Bwectk8tis5ORw4actGRg4ZcdOSgIRcdOWjISUeZlNmi/hjYRVL3EssoshfwUPr7GSS1BRYH5hhqM3vDzHarizq4HRjU3A5JnYAfAn9rZvevgLsqK2b2nJkNMLMBwLrAB8AfayGw36C+vPHiW7w18W1mfjqTsdc/zEY7Na2jbLTjetx9xf0APHDTo6y9xeq1KDo7HTloyEVHDhpy0ZGDhlx05KAhJx1lUqahnglcDBxevUNSD0k3S3o8LYPT9n+mVq8kTZX07bT9SklbSVpN0mOpJfmUpJXS/kWBjYERwJ6FcoZIelDSbcAzwGnAiun7Z0rqI+npdGwbSWdJejqde2QzureWNE7Sk5JuTOUi6TRJz6TvndXcxTCzR83szblcq5OB04GPqsrbGZgI/Gsu39sCeMnMXp3L/s9F955dmfz61DnrU16fRvee3Zoc061nVya/NgWA2bNmM+PdD1isW6daFJ+Vjhw05KIjBw256MhBQy46ctCQk44yKbuP+kLgW5KqJ90+DzjHzNYDdgUuSdsfBgYDqwEvA5uk7RsCjwAHA+el1uRA4PW0fydgtJk9D0yVtG6hrHWAH5rZysAxuGEbYGY/qdJ0INAHGGBmawJXF3cmz8AJwJZmtg4wHjhCUjfgm8Bq6Xu/aOnFSeddB+hlZn+u2r4ocDRw0jy+vidw7TzOfaCk8ZLGT548+fPICoIgCDKhVENtZu8BVwKHVe3aEhglaQJwG7BYMkwPApum5TfAGpJ6Av8zsxnAOOA4SUcDy5nZh+l8ewHXpc/X0dT9/ZiZTWyB3C2Bi8xsZtI+rWr/BkB/4OGkez9gOeBdvCV8qaRdcFd0i5DUgLu2f9zM7hPxyszc+rTbAzsCN87t/GZ2sZkNNLOBPXrMP3hiyqRp9Fi2sSbafdmuTJk0tckxUydNo0cv781oaNNAx84deG/q+/M99+chBx05aMhFRw4actGRg4ZcdOSgIScdZVKPqO9zcZd0x6pyN6j0tZpZz2SQHsBb0ZsAY4HJwG64AcfMrsGN04fAnZKGSuoKDAUukfQK8BNgD0lKZc2o0f8hYExBc38zG5EM+yDgJmAHYHRyo1eCvX4+j3N2AlYHxibtGwC3pYCy9YEz0vYf4RWUQwvf3Q540sz+W6P/j+cef5GeKy3NUn2WoG27tgwZPphxt41vcsy428ez9X6bAbDpbhsw4d6na1V8Vjpy0JCLjhw05KIjBw256MhBQ046yqRt2QWY2TRJN+DG+vdp893ASOBMAEkDzGyCmb2WXMztzexlSQ8BRwKHpuNWAF42s/Ml9QbWBPoCV5nZQZUyJd1Po9u8yPu4cWyOMcBBku4zs5mSula1qh8FLpTU18xelNQR6Am8AXQwszslPZz0zQIGtODavAvMCbaTNBY40szGF/WnaPnpZjaq8PW9mIfb+4swe9ZsRo28lF+OPp6GNg385bL7ePWZ19nvpOE8P/4lxt0+nrsuvZdjrhzJ5c9fwPvTpnPKXufUUkI2OnLQkIuOHDTkoiMHDbnoyEFDTjrKRGZWzoml6WZWCbZaEg+KOsPMTkzG+EJgVbyy8ICZHZyOvQpoY2Z7S9oIj+TuYWZTJR0D7At8CrwF7A3cDJxuZqMLZR+Wzn09bvh2KOy7BjfwdyUNd5jZ6ikq/Axg23T+35nZqKLxlDQUD/paKJ3uBOBx4FZgYbzVfZaZXdHM9Tgj6V0GN+6XmNmJVcfMKatq+4m4oT4rrXcE/gOskIz9fBk4cKB1eXL5lhxaGmNm38hWDbu3qoZcdOSgIRcdOWjIRUcOGnLRkYMGgHvspifMbIGH6C4IpbWoK0Y6ff4v0KGwPgUYPpfv7Vv4/AgF97yZnYZHbhfZvJlznF9YHVu1b++qw1dP22cCR6SlePyQwud7gfWakd3ssKuq8xwFHDWfY4bMZfuJVeszgG7NHRsEQRB8tYjMZEEQBEGQMWGogyAIgiBjwlAHQRAEQcaEoQ6CIAiCjAlDHQRBEAQZE4Y6CIIgCDImDHUQBEEQZEwY6iAIgiDImDDUQRAEQZAxYaiDIAiCIGPCUAdBEARBxpQ2KUeQB5KGAcP69u37vRdeeKG15QRBEHypkNTqk3KEof6aELNn5aUjBw256MhBQy46ctCQi44cNEAes2eF6zsIgiAIMiYMdRAEQRBkTBjqIAiCIMiYMNRBEARBkDFhqIMgCIIgY8JQB0EQBEHGhKEOgiAIgowJQx0EQRAEGROGOgiCIAgyJgx1EARBEGRMGOogCIIgyJgw1EETBm4zgN//+zwuf/4Chh+982f2t2vfluOvPZzLn7+A88edypLL9fjK6shBQy46ctCQi44cNOSiIwcNOekoi9IMtSSTdHZh/UhJJ5ZY3s6pzFXmcczikg4prC8j6ab5nHespAVOyC7pFEmvSZpetX1/SZMlTUjLAVX7F5P0uqRRVZqeK3xniQXVB9DQ0MDIUSM4bvtTOGC1w9l8z8H0XnXZJsdsO2Io09+Zzv4rj+SWc+/ggNP2qUXR2enIQUMuOnLQkIuOHDTkoiMHDTnpKJMyW9QfA7tI6l5iGUX2Ah5Kfz+DpLbA4sAcQ21mb5jZbnVRB7cDg+ay73ozG5CWS6r2nQw80Mx3vlX4ztu1ENhvUF/eePEt3pr4NjM/ncnY6x9mo52a1lE22nE97r7ifgAeuOlR1t5i9VoUnZ2OHDTkoiMHDbnoyEFDLjpy0JCTjjIp01DPBC4GDq/eIamHpJslPZ6WwWn7P1OrV5KmSvp22n6lpK0krSbpsdSKfErSSmn/osDGwAhgz0I5QyQ9KOk24BngNGDF9P0zJfWR9HQ6to2ksyQ9nc49shndW0saJ+lJSTemcpF0mqRn0vfOau5imNmjZvbm57mAktYFlgTu/jzf+6J079mVya9PnbM+5fVpdO/Zrckx3Xp2ZfJrUwCYPWs2M979gMW6dfrK6chBQy46ctCQi44cNOSiIwcNOekok7L7qC8EviWpc9X284BzzGw9YFeg0op8GBgMrAa8DGyStm8IPAIcDJxnZgOAgcDraf9OwGgzex6YmgxchXWAH5rZysAxwEupFfqTKk0HAn2AAWa2JnB1cWfyDJwAbGlm6wDjgSMkdQO+CayWvveLll6cArsmI3+TpF6pvAbgbODIuXznslTh+KkkNXeApAMljZc0fvLkyV9AVhAEQdDalGqozew94ErgsKpdWwKjJE0AbgMWS63TB4FN0/IbYA1JPYH/mdkMYBxwnKSjgeXM7MN0vr2A69Ln62jq/n7MzCa2QO6WwEVmNjNpn1a1fwOgP/Bw0r0fsBzwLvARcKmkXYAPWlBWkduBPsnIjwGuSNsPAe40s9eb+c63zGwNvCKzCbBvcyc2s4vNbKCZDezRY/7BE1MmTaPHso010e7LdmXKpKlNjpk6aRo9enlvRkObBjp27sB7U9+f77k/DznoyEFDLjpy0JCLjhw05KIjBw056SiTekR9n4u7pDtWlbtBoY+1p5lNx/tiK8ZnLDAZ2A034JjZNcCOwIfAnZKGSuoKDAUukfQK8BNgj0Irc0aN/g8BYwqa+5vZiGTYBwE3ATsAo5MbvRLo9fN5ndTMpprZx2n1EqDiDdgQODT9T2cB35Z0WvrOpPT3feAa5t73/bl47vEX6bnS0izVZwnatmvLkOGDGXfb+CbHjLt9PFvvtxkAm+62ARPufboWRWenIwcNuejIQUMuOnLQkIuOHDTkpKNM2pZdgJlNk3QDbqx/nzbfDYwEzgSQNMDMJpjZa8nF3N7MXpb0EO76PTQdtwLwspmdL6k3sCbQF7jKzA6qlCnpfhrd5kXeB+bWMTEGOEjSfWY2U1LXqlb1o8CFkvqa2YuSOgI9gTeADmZ2p6SHk75ZwICWXB9JSxf6rncE/p2u27cKx+wPDDSzYypBcWY2RVI7vHJwT0vKmh+zZ81m1MhL+eXo42lo08BfLruPV595nf1OGs7z419i3O3juevSeznmypFc/vwFvD9tOqfsdU4tis5ORw4actGRg4ZcdOSgIRcdOWjISUeZyMzKObE03cwqwVZLAhOBM8zsxGSMLwRWxSsLD5jZwenYq4A2Zra3pI3wSO4eZjZV0jG4m/dT4C1gb+Bm4HQzG10o+7B07uuBI81sh8K+a3ADf1fScIeZrZ4M4BnAtun8vzOzUZLGpnOMlzQUOB1YKJ3uBOBx4FZgYbzVfZaZVdzXxetxRtK7DG7cL0nX4pe4gZ4JTAO+b2bPVn13f9xQH5oqCA8A7YA2uJE+IlUO5srAgQOty5PLz+uQ0hkz+0a2ati9VTXkoiMHDbnoyEFDLjpy0JCLjhw0ANxjNz1hZgs8RHdBKK1FXTHS6fN/gQ6F9SnA8Ll8b9/C50couOfN7DQ8crvI5s2c4/zC6tiqfXtXHb562j4TOCItxeOHFD7fC6zXjOz5up7N7CjgqGa2HwscO5/vXg5cnj7PoNE9HgRBEHzFicxkQRAEQZAxYaiDIAiCIGPCUAdBEARBxoShDoIgCIKMaZGhlrRQS7YFQRAEQVBbWtqiHtfCbUEQBEEQ1JB5Ds+StBSe1GMRSWvj44QBFqMw3CoIgiAIgnKY3zjqbYD9gWWBXxW2vw8cV5KmIAiCIAgS8zTUKcPWFZJ2NbOb66QpCIIgCIJESzOT3SFpb3wayDnfMbN5TjgRBEEQBMGC0aJc35JG49M5PgHMySltZmeXJy2oBZKGAcP69u37vRdeeKG15QRBEHypkNTqub5baqifNrPV66AnKImYlCMvHTloyEVHDhoqOma/tVKramhY6oVsrkVr68hBA+QxKUdLh2c9ImmNUpUEQRAEQfAZWtpHvTGwv6SJwMf4MC0zszVLUxYEQRAEQYsN9XalqgiCIAiCoFla5Po2s1eBXsDQ9PmDln43CIIgCIIvTktzff8MOBo4Nm1qB/yhLFFBEARBEDgtbRV/E9gRmAFgZm8AncoSFQRBEASB01JD/Yn5OC4DkNSxPElBEARBEFRoqaG+QdJFwOKSvgfcA/yuPFlBEARBEEALo77N7CxJWwHvAf2A/zOzMaUqC4IgCIKgxcOzMLMxkv5W+Y6krmY2rTRlQRAEQRC0zFBLOgg4CfgImE1KeAKsUJ60IAiCIAha2kd9JLC6mfUxsxXMbHkzCyP9FWTgNgP4/b/P4/LnL2D40Tt/Zn+79m05/trDufz5Czh/3KksuVyPr6yOHDTkoiMHDa2pY/S9M1h141dZecNXOf2C/31m/6uvfcpWu09iwND/MHSX13n9jZlz9h198hTW2Ow/rLbJq/zwhMm0ZH6FlpDDPclBQ046yqKlhvolPMlJi5Fkks4urB8p6cTPc47PWd7OqcxV5nHM4pIOKawvI+mm+Zx3rKQFSsguqYOkP0t6VtK/JJ1W2LecpL9KeiqVtWxh336SXkjLflWanpM0IS1LLIi+Cg0NDYwcNYLjtj+FA1Y7nM33HEzvVZdtcsy2I4Yy/Z3p7L/ySG459w4OOG2fWhSdnY4cNOSiIwcNralj1ixj5HGT+fPVy/D0/b257k/v88xznzQ55ic/n8I+u3diwr29OeGIrhx36lQAHnn8Qx55/CMm3NuLp8b2ZvyEj7h/3IcLrCmHe5KDhpx0lElLDfWx+MQcF0k6v7LM5zsfA7tI6r5gElvMXsBD6e9nkNQWWByYY6jN7A0z260u6uAsM1sFWBsYLKmSlvUs4MqUN/3nwC+T3q7Az4D1gUHAzyR1KZzvW2Y2IC1v10Jgv0F9eePFt3hr4tvM/HQmY69/mI12alpH2WjH9bj7ivsBeOCmR1l7i9pPqpaDjhw05KIjBw2tqeOxv3/Ein3ascJy7WjfXgzfaVFu+8v0Jsf8+/lPGTp4EQA2H7zInP0SfPSR8cknxscfG59+Ckt2b3Fo0FzJ4Z7koCEnHWXSUkN9EXAv8Cg+J3VlmRczgYuBw6t3SOoh6WZJj6dlcNr+z9TqlaSpkr6dtl8paStJq0l6LLUin5K0Utq/KD5xyAhgz0I5QyQ9KOk24BngNGDF9P0zJfWR9HQ6to2ksyQ9nc49shndW0saJ+lJSTemcpF0mqRn0vfOqv6emX1gZvelz58ATwKVKl//dG0B7gN2Sp+3AcaY2TQz+x8wBth2Ptd8gejesyuTX586Z33K69Po3rNbk2O69ezK5NemADB71mxmvPsBi3Wrbe6bHHTkoCEXHTloaE0dk96aRa+e7eas91y6LZPemtXkmDVXa88f75wBwB/vnMH7042p02ax4cBFGDJ4EXoOeIWeA15h6yEdWHXl9gukB/K4JzloyElHmbS0atfOzI74Aue/EHhK0hlV288DzjGzhyT1Bv4CrAo8DAwGXgVeBjYBrgQ2BL6PG9rzzOxqSe2BNul8OwGjzez5ZODXNbNKRWIdvH99oqQ+6fMAgLRe4UCgDzDAzGamFu0ckmfgBGBLM5sh6WjgCEkX4pnbVjEzk7T4vC5I2j8sXQOAfwC7pPVvAp0kdQN6Aq8Vvvp62lbhMkmzgJuBX1gzHV+SDkz/F71796YLrTsfdRB8VTnz/7oz8rjJXHHDe2yy/iL0XLoNbdrAixM/4d8vfMJ/nuwDwDbDJ/Hgox3YZINFWldw8KWipS3quyQdKGlpSV0ry/y+ZGbv4Yb2sKpdWwKjJE0AbgMWS63TB4FN0/IbYA1JPYH/mdkMYBxwXDKSy5lZpbNnL+C69Pk6mrq/HzOziS34H7cELjKzmUl79dCzDfDW78NJ937AcsC7eDT8pZJ2YR59+cn9fi1wvpm9nDYfCWwm6e/AZsAkYNZcTlHhW2a2Bl6R2QTYt7mDzOxiMxtoZgN79Jh/8MSUSdPosWxjTbT7sl2ZMmlqk2OmTppGj17em9HQpoGOnTvw3tT353vuz0MOOnLQkIuOHDS0po6eS7XhtUmfzlmf9OZMei7VpskxyyzVlpt/vzRPjOnNL451jYt3bsOf7prBBusszKIdG1i0YwPbDu3Io098tEB6II97koOGnHSUSUsN9V6kfmoa3d7jW/jdc3GXdDHtaAOwQaGPtaeZTQceoNH4jAUmA7vhBhwzuwbPOf4hcKekoanCMBS4RNIrwE+APSQplTWjhTrnh3BXdEVzfzMbkQz7IOAmYAdgdHKjVwK9fl44x8XAC2Z2bmVD6iffxczWBo5P297BDXavwneXTdsws8rf94FrUvkLzHOPv0jPlZZmqT5L0LZdW4YMH8y425re5nG3j2fr/TYDYNPdNmDCvU/XoujsdOSgIRcdOWhoTR3rDViYFyd+ysT/fMonnxjX3zqdYds0zaI8ZeosZs92p9Zp5/+P7+y5GAC9erblgUc/ZOZM49NPjQfGfcgqKy246zuHe5KDhpx0lElLM5N9YZ+pmU2TdANurH+fNt8NjATOBJA0wMwmmNlrycXc3sxelvQQ3uI8NB23AvCymZ2fXOZrAn2Bq8zsoEqZku7HjX017zP3yUTGAAdJuq/i+q5qVT8KXCipr5m9KM933hN4A+hgZndKejjpmwUMKJ5c0i+AzsABVdu7A9PMbDZeGapco78ApxYCyLYGjq0ExZnZFEnt8MrBPXP5nz4Xs2fNZtTIS/nl6ONpaNPAXy67j1efeZ39ThrO8+NfYtzt47nr0ns55sqRXP78Bbw/bTqn7HVOLYrOTkcOGnLRkYOG1tTRtq04/9QebLfXG8yaZXxnz8VYrd9C/OyMqay71sLsuE1Hxo77kONPnYoEm2ywCKNOdQ/Wbjssyn0Pfcham/8HCbbZvAPDtl7wqRJyuCc5aMhJR5moJWP6KkFd1ZjZlfP4znQzqwRbLQlMBM4wsxOTcboQ75duCzxgZgenY68C2pjZ3pI2wiO5e5jZVEnH4G7eT4G3gL3xPtrTzWx0oezD0rmvB440sx0K+67BDfxdScMdZrZ6MoBn4AFbnwK/M7NRksamc4yXNBQ4HVgone4E4HHgVmBhvNV9lpldUXUtlsX7m5/Fo+EBRpnZJZJ2wyO9Dfco/MDMPk7f+y5wXDr+FDO7LFUQHsCnGm2DG+kjUuVgrgwcONC6PNm6fdRjZt/IVg27t6qGXHTkoCEXHTloqOiY/dZKraqhYakXsrkWra0jBw0A99hNT5jZAg3RXVBaGky2XuHzwsAWeOTyXA11xUinz/8FOhTWpwDD5/K9fQufH6Hgnjez0/CAsiKbN3OO4tCxsVX79q46fPW0fSZwRFqKxw8pfL6Xpteiwjxdz2b2Om7Em9t3E+42b27f72lsYVe2zQDWnVd5QRAEwVeHlrq+mwxVSpHL1zV/dBAEQRAEtaKlwWTVzIAY6xMEQRAEZdPSSTlux/tQwY17f+CGskQFQRAEQeC0tI+6mG1rJvBq6ncNgiAIgqBEWmqo/wO8aWYfAUhaRFIfM3ulNGVBEARBELS4j/pGfB7qCrPStiAIgiAISqSlhrptmkwCmDOxxIKn1wmCIAiCYJ601FBPlrRjZUXSTsCUciQFQRAEQVChpX3UBwNXp5miwLNsNTsRRBAEQRAEtaOlCU9eAjaozL+cJtAIgiAIgqBkWuT6ltRZ0q/wdJxjJZ0tqXOpyoIgCIIgaPGkHDcDTwOVySb2BdYys11K1BbUAEnDgGF9+/b93gsvvNDacoIgCL5USGr1STlaaqgnmNmA+W0L8iVmz8pLRw4actGRg4ZcdOSgIRcdOWiAPGbPamnU94eSNq6sSBoMfFiOpCAIgiAIKnyeqO8rC/3S/wP2K0dSEARBEAQV5mmoJRXnZr4S6Jg+zwC2BJ4qSVcQBEEQBMy/Rd0p/e0HrAfcCgjYB3isRF1BEARBEDAfQ21mJwFIegBYx8zeT+snAn8uXV0QBEEQfM1paTDZksAnhfVP0rYgCIIgCEqkpcFkVwKPSfpjWt8ZuLwMQUEQBEEQNNLSFKKnSLoL2CRt+o6Z/b08WUEQBEEQQMtb1JjZk8CTJWoJgiAIgqCKlvZRB0EQBEHQCoShDoIgCIKMCUMdNGHgNgP4/b/P4/LnL2D40Tt/Zn+79m05/trDufz5Czh/3KksuVyPr6yOHDTkoiMHDbnoyEFDLjpy0JCTjrJoFUMtySSdXVg/Mo3NLqu8nVOZq8zjmMUlHVJYX0bSTfM571hJC5SsXVInSRMKyxRJ56Z9vSXdJ+nvkp6StH3a3k7SFZL+Kenfko5dEA0VGhoaGDlqBMdtfwoHrHY4m+85mN6rLtvkmG1HDGX6O9PZf+WR3HLuHRxw2j61KDo7HTloyEVHDhpy0ZGDhlx05KAhJx1l0lot6o+BXSR1r1N5ewEPpb+fQVJbYHFgjqE2szfMbLeyhZnZ+2Y2oLIArwK3pN0nADeY2drAnsCv0/bdgYXMbA1gXeAgSX0WVEu/QX1548W3eGvi28z8dCZjr3+YjXZqWg/ZaMf1uPuK+wF44KZHWXuL1Re02Cx15KAhFx05aMhFRw4actGRg4acdJRJaxnqmcDFwOHVOyT1kHSzpMfTMjht/2dq9UrSVEnfTtuvlLSVpNUkPZZapU9JWintXxTYGBiBG7tKOUMkPSjpNuAZ4DRgxfT9MyX1kfR0OraNpLMkPZ3OPbIZ3VtLGifpSUk3pnKRdJqkZ9L3zprXRZG0MrAE8GDaZMBi6XNn4I3C9o6pgrEInoDmvXle8RbQvWdXJr8+dc76lNen0b1ntybHdOvZlcmvTQFg9qzZzHj3Axbr1olakoOOHDTkoiMHDbnoyEFDLjpy0JCTjjJp8fCsErgQeErSGVXbzwPOMbOHJPUG/gKsCjwMDMZbnC/jY7qvBDYEvo8b2vPM7GpJ7YE26Xw7AaPN7Plk4Nc1syfSvnWA1c1sYmqRrl6ZY7uqhXog0AcYYGYzJXUtCk6egROALc1shqSjgSMkXQh8E1jFzEzS4vO5JnsC11vjJOEnAnenikFHfCIUgJvS//Um0AE43MymVZ9M0oFJO71796YLrTsfdRAEQfD5abVgMjN7Dze0h1Xt2hIYJWkCcBuwWGqdPghsmpbfAGtI6gn8z8xmAOOA45KRXM7MKvNl7wVclz5fR1P392NmNrEFcrcELjKzmUl7tVHcAOgPPJx07wcsB7wLfARcKmkX4IP5lLMncG1hfS/gcjNbFtgeuEpSAzAImAUsAywP/FjSCtUnM7OLzWygmQ3s0WP+wRNTJk2jx7KNNdHuy3ZlyqSpTY6ZOmkaPXp5j0VDmwY6du7Ae1Pfn++5Pw856MhBQy46ctCQi44cNOSiIwcNOekok9aO+j4Xd0l3LGxrADYo9Nv2NLPpwAN4K3oTYCwwGdiN5CY2s2uAHYEPgTslDU0t36HAJZJeAX4C7CFJqawZNfo/BIwpaO5vZiOSYR+Et4B3AEYnN3olcOznc04grQW0LbT2SdfmhvT/jQMWBroDe+Negk/N7G3c27BAQW0Azz3+Ij1XWpql+ixB23ZtGTJ8MONuG9/kmHG3j2fr/TYDYNPdNmDCvU8vaLFZ6shBQy46ctCQi44cNOSiIwcNOekok1Y11KllegNukCrcDczpA5Y0IB37Gm6kVjKzl/HgsCNxA05qUb5sZufj03GuiRvyq8xsOTPrY2a9gIk0pkIt8j6N03pWMwYP2Gqbyupatf9RYLCkvml/R0krJ09AZzO7E++PX8vMZhUM+v8VzrEXTVvTAP8BtkjnXBU31JPT9qGVsvAW/bNz0d5iZs+azaiRl/LL0cdz6TPn8MCN43j1mdfZ76ThbDjM6wF3XXovi3XtxOXPX8Cuhw/jkmOvXtBis9SRg4ZcdOSgIRcdOWjIRUcOGnLSUSZq7A6tY6HSdDOrBFstiRvPM8zsxNTfeyHeL90WeMDMDk7HXgW0MbO9JW2EG+seZjZV0jHAvsCnwFt4q/Nm4HQzG10o+7B07uuBI81sh8K+a3ADf1fScIeZrZ4M9BnAtun8vzOzUZLGpnOMlzQUOB1YKJ3uBOBxvNKwMN7qPsvMrpjLNXkZ2N7Mni1s6w/8DlgUDyA7yszuThWAy3B3u4DLzOzMeV3zgQMHWpcnW7ePeszsG9mqYfdW1ZCLjhw05KIjBw256MhBQy46ctAAcI/d9ISZLbDHckFolWCyipFOn/+LB0RV1qcAw+fyvX0Lnx+h4BEws9PwgLIimzdzjvMLq2Or9u1ddfjqaftM4Ii0FI8fUvh8L7BeM7IHNbPtM5hZc33Mz+ABdNXbp+NDtIIgCIKvOK3dRx0EQRAEwTwIQx0EQRAEGROGOgiCIAgyJgx1EARBEGRMGOogCIIgyJgw1EEQBEGQMWGogyAIgiBjwlAHQRAEQcaEoQ6CIAiCjAlDHQRBEAQZ0yq5voP6IWkYMKxv377fe+GFF1pbThAEwZcKSa2e6zsM9deEmJQjLx05aMhFRw4actGRg4aKjs22O71VNdx/19FZXIscJuUI13cQBEEQZEwY6iAIgiDImDDUQRAEQZAxYaiDIAiCIGPCUAdBEARBxoShDoIgCIKMCUMdBEEQBBkThjoIgiAIMiYMdRAEQRBkTBjqIAiCIMiYMNRBEARBkDFtW1tAkBcDtxnAIed+h4Y2Ddx16V+5/vQ/Ndnfrn1bjrpiJCutuwLvTX2fU/Y8h/++OvkrqSMHDbnoyEFDLjpy0NCaOgatuzwjD96ChoYG/jz6H1xz49+a7N/jm+vxjW3XZNas2bzz7gecfs5d/Pft9wA44+Td6b/KMvzzX69z7Ik3L7CWCrnck7Koe4takkk6u7B+pKQTSyxv51TmKvM4ZnFJhxTWl5F003zOO1bSAidqlzRc0lOS/iXp9ML2cyRNSMvzkt5J2wdIGpeOf0rS8AXVUKGhoYGRo0Zw3PancMBqh7P5noPpveqyTY7ZdsRQpr8znf1XHskt597BAaftU6vis9KRg4ZcdOSgIRcdOWhoTR0NDeJHP9iKo356I/sddAlbDOnPcr27NTnmhZf+y4GHXcF3D7mM+x96joO/O2TOvutufoxTz7pjgXU01ZTHPSmT1nB9fwzsIql7ncrbC3go/f0MktoCiwNzDLWZvWFmu5UtTFI34ExgCzNbDVhK0hZJw+FmNsDMBgAXALekr30AfDsdvy1wrqTFa6Gn36C+vPHiW7w18W1mfjqTsdc/zEY7Na2LbLTjetx9xf0APHDTo6y9xeq1KDo7HTloyEVHDhpy0ZGDhtbUserKSzPpjXd48613mTlzNvfe/2823mClJsf8/an/8PHHMwF45tk36NG905x9T054lQ8++GSBdRTJ5Z6USWsY6pnAxcDh1Tsk9ZB0s6TH0zI4bf9navVK0lRJ307br5S0laTVJD2WWp9PSVop7V8U2BgYAexZKGeIpAcl3QY8A5wGrJi+f6akPpKeTse2kXSWpKfTuUc2o3vr1Mp9UtKNqVwknSbpmfS9s5q5FisAL5hZxQdzD7BrM8ftBVwLYGbPm9kL6fMbwNtAj3lf8pbRvWdXJr8+dc76lNen0b1n09pyt55dmfzaFABmz5rNjHc/YLFunaglOejIQUMuOnLQkIuOHDS0po7u3Tvx9uT35qxPnvI+3bstOtfjt996Tf42/uUFKnO+mjK5J2XSWn3UFwJPSTqjavt5wDlm9pCk3sBfgFWBh4HBwKvAy8AmwJXAhsD3cUN7npldLak90CadbydgtJk9nwz8umb2RNq3DrC6mU2U1Cd9HgCQ1iscCPQBBpjZTEldi4KTZ+AEYEszmyHpaOAISRcC3wRWMTObS6v3RaBfKu91YGegfdX5lwOWB+6t/rKkQen4l5o5N5IOTPrp3bs3XWjd+aiDIPj6sNXm/em38tL88KhrWlvKl55Wifo2s/dwQ3tY1a4tgVGSJgC3AYul1umDwKZp+Q2whqSewP/MbAYwDjguGcnlzOzDdL69gOvS5+to6v5+zMwmtkDulsBFZjYzaZ9WtX8DoD/wcNK9H7Ac8C7wEXCppF1wl3X1dfgfXtG4Pv2PrwCzqg7bE7jJzJpsl7Q0cBXwHTOb3ZxwM7vYzAaa2cAePebf6J4yaRo9lm2siXZftitTJk1tcszUSdPo0ct7LRraNNCxcwfem/r+fM/9echBRw4actGRg4ZcdOSgoTV1TJnyPkv0WGzOeo/unZgydfpnjlt3wHLsu+dGHHfizXz6afUrrbbkck/KpDWHZ52Lu6Q7FrY1ABtU+mbNrKeZTQcewFvRmwBjgcnAbrhxw8yuAXYEPgTulDQ0tXyHApdIegX4CbCHJKWyZtTo/xAwpqC5v5mNSIZ9EHATsAMwOrnRKwFiP0/abzez9c1sQ+A54Pmq8+9JcnvPKVBaDPgzcLyZPVqj/4PnHn+RnistzVJ9lqBtu7YMGT6YcbeNb3LMuNvHs/V+mwGw6W4bMOHep2tVfFY6ctCQi44cNOSiIwcNranj2effZNllurDUkp1p27aBoZutysOPvtjkmJVWXIIfH7YNx550M++8+5n2Sc3J5Z6USasNzzKzaZJuwI3179Pmu4GReIAVkgaY2QQzey25mNub2cuSHgKOBA5Nx60AvGxm5yeX+ZpAX+AqMzuoUqak+3FjX837wNw6LMYAB0m6r+L6rmpVPwpcKKmvmb0oqSPQE3gD6GBmd0p6OOmbBQwonlzSEmb2tqQueEDbHoV9qwBdcI9BZVt74I/AlWY2z8j0z8vsWbMZNfJSfjn6eBraNPCXy+7j1WdeZ7+ThvP8+JcYd/t47rr0Xo65ciSXP38B70+bzil7nVNLCdnoyEFDLjpy0JCLjhw0tKaOWbONc38zhrN+sQcNbcSdd/+TV/4zhe/uuzHPPv8Wj/ztRQ4esTmLLNyek47bCYC3J7/HcSd5LOwFZ+5N717dWGThdtx41SGccc5dPP5kSxyb+V2LeiIzq2+B0nQzqwRbLQlMBM4wsxOTMb4Q75duCzxgZgenY68C2pjZ3pI2wiO5e5jZVEnHAPsCnwJvAXsDNwOnm9noQtmHpXNfDxxpZjsU9l2DG/i7koY7zGz1FBV+Bh5h/SnwOzMbJWlsOsd4SUOB04GF0ulOAB4HbgUWxlvdZ5nZFc1cj2uBtdLqz83susK+E4GFzeyYwrZ9gMuAfxVOs7+ZTZjXdR84cKB1ebJ1+6jHzL6RrRp2b1UNuejIQUMuOnLQkIuOHDRUdGy23enzP7BE7r/r6CyuxT120xNmtsBDcReEureoK0Y6ff4v0KGwPgVodlywme1b+PwIBbe9mZ2GB5QV2byZc5xfWB1btW/vqsNXT9tnAkekpXj8kMLne4H1mpE9qJlt1ZqaHTaW9p3YzLY/AH+Y33mDIAiCrwaRQjQIgiAIMiYMdRAEQRBkTBjqIAiCIMiYMNRBEARBkDFhqIMgCIIgY8JQB0EQBEHGhKEOgiAIgowJQx0EQRAEGROGOgiCIAgyJgx1EARBEGRMGOogCIIgyJi6T8oR1BdJw4BhwLeAfy/AqboDUxZQTmd8nu4FIQcdOWjIRUcOGnLRkYOGXHTkoKFWOlYys84LeI4Fw8xi+RoswMUL+P3xra0hFx05aMhFRw4actGRg4ZcdOSgIScdC7qE6/vrw+2tLYA8NEAeOnLQAHnoyEED5KEjBw2Qh44cNEAGOsJQf00ws1Z/2HLQAHnoyEED5KEjBw2Qh44cNEAeOnLQAHnoCEMdtJSLW1tAIgcdOWiAPHTkoAHy0JGDBshDRw4aIB8dC0QEkwVBUFMkLQ7sbWa//oLfvzN9/51a6gqCLythqIMgqCmS+gB3mNnqra0lCL4KhOs7CIJacxqwoqQJks5My9OS/ilpOICkIZIekPRnSc9J+q2khrTvFUnd0+dvS3pK0j8kXdWK/1MQtBptW1tAEARfOY4BVjezAZJ2BQ4G1sLHtD4u6YF03CCgP/AqMBrYBbipchJJqwEnABuZ2RRJXev4PwRBNkSLOgiCMtkYuNbMZpnZf4H7gfXSvsfM7GUzmwVcm44tMhS40cymAJjZtHqJDoKcCEMdBEFrUR0gEwEzQdAMYaiDIKg17wOd0ucHgeGS2kjqAWwKPJb2DZK0fOqbHg48VHWee4HdJXUDCNd38HUlDHUQBDXFzKYCD0t6GtgQeAr4B254jzKzt9KhjwOj8Bz0E4E/Vp3nX8ApwP2S/gH8qj7/QRDkRQzPCoKg7kgaAhxpZju0spQgyJ5oUQdBEARBxkSLOgiCIAgyJlrUQRAEQZAxYaiDIAiCIGPCUAdBEARBxoShDoIgCIKMCUMdBEEQBBnz/0G+8teayBC0AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(figsize=(6, 8))\n",
    "\n",
    "# randomly selecting a subset of topics\n",
    "which_topics = random.sample(list(new_topic_labels), 10)\n",
    "\n",
    "plot_doc_topic_heatmap(fig, ax, new_doc_topic, doc_labels_bg,\n",
    "                       topic_labels=new_topic_labels,\n",
    "                       which_documents=which_docs,\n",
    "                       which_topics=which_topics);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%% md\n"
    }
   },
   "source": [
    "Similarily, we can work with [plot_topic_word_heatmap](api.rst#tmtoolkit.topicmod.visualize.plot_topic_word_heatmap) to visualize a topic-word distribution. We can also select a subset of topics and words from the vocabulary:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:54.354368Z",
     "iopub.status.busy": "2022-02-08T06:54:54.353361Z",
     "iopub.status.idle": "2022-02-08T06:54:54.668809Z",
     "shell.execute_reply": "2022-02-08T06:54:54.669528Z"
    },
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.visualize import plot_topic_word_heatmap\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(12, 8))\n",
    "\n",
    "which_words = ['may', 'european', 'referendum', 'brexit',\n",
    "               'eu', 'uk', 'britain', 'company', 'trade', 'growth']\n",
    "\n",
    "plot_topic_word_heatmap(fig, ax, new_topic_word, vocab_bg,\n",
    "                        topic_labels=new_topic_labels,\n",
    "                        which_topics=which_topics,\n",
    "                        which_words=which_words);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note that there's also a generic heatmap plotting function [plot_heatmap](api.rst#plot-heatmaps-for-topic-models) for any kind of 2D matrices. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Word clouds\n",
    "\n",
    "Thanks to the [wordlcloud package](https://pypi.org/project/wordcloud/), topic-word and document-topic distributions can also be visualized as \"word clouds\" with tmtoolkit. The function [generate_wordclouds_for_topic_words](api.rst#tmtoolkit.topicmod.visualize.generate_wordclouds_for_topic_words) generates a word cloud for each topic by scaling a topic's word by its probability (weight). You can choose to display only the top `top_n` words per topic. The result of this function will be a dictionary mapping topic labels to the respective word cloud image."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:54:54.674624Z",
     "iopub.status.busy": "2022-02-08T06:54:54.673570Z",
     "iopub.status.idle": "2022-02-08T06:55:01.083062Z",
     "shell.execute_reply": "2022-02-08T06:55:01.083410Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['2_germany_german', '3_trump_president', '5_flight_air', '6_report_agency', '7_company_reform', '8_intelligence_russia', '9_party_vote', '10_al_syria', '11_play_game', '12_parent_child', '13_economic_meeting', '14_say_committee', '15_northern_election', '16_mr_damage', '17_white_trump', '18_minister_election', '19_police_shoot', '20_percent_limit', '21_attack_group', '22_space_european', '23_country_love', '24_industry_service', '25_sanction_visit', '26_judge_victim', '27_man_new', '28_feel_mp', '29_club_sell', '30_china_chinese', '31_company_investment', '32_britain_democratic', '33_general_authority', '34_america_nation', '35_force_group', '37_north_discuss', '38_southern_deal', '39_russian_diplomat', '40_turkish_country', '41_bill_house', '42_win_anti', '43_bank_investor', '44_note_7', '45_day_morning', '46_support_provide', '47_family_refugee', '48_body_north', '49_fire_death', '50_officer_police'])"
      ]
     },
     "execution_count": 63,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.visualize import generate_wordclouds_for_topic_words\n",
    "\n",
    "# some options for wordcloud output\n",
    "img_w = 400   # image width\n",
    "img_h = 300   # image height\n",
    "\n",
    "topic_clouds = generate_wordclouds_for_topic_words(\n",
    "    new_topic_word, vocab_bg,\n",
    "    top_n=20, topic_labels=new_topic_labels,\n",
    "    width=img_w, height=img_h\n",
    ")\n",
    "\n",
    "# show all generated word clouds\n",
    "topic_clouds.keys()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's select specific topics and display their word cloud:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:55:01.087365Z",
     "iopub.status.busy": "2022-02-08T06:55:01.086727Z",
     "iopub.status.idle": "2022-02-08T06:55:01.101724Z",
     "shell.execute_reply": "2022-02-08T06:55:01.102072Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<PIL.Image.Image image mode=RGBA size=400x300 at 0x7FB69D9D9700>"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "topic_clouds['50_officer_police']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The same can be done for the document-topic distribution using [generate_wordclouds_for_document_topics](api.rst#tmtoolkit.topicmod.visualize.generate_wordclouds_for_document_topics). Here, a word cloud for each document will be generated that contains the `top_n` most probable topics for this document:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:55:01.119073Z",
     "iopub.status.busy": "2022-02-08T06:55:01.116242Z",
     "iopub.status.idle": "2022-02-08T06:55:12.344101Z",
     "shell.execute_reply": "2022-02-08T06:55:12.343647Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['NewsArticles-1032',\n",
       " 'NewsArticles-1036',\n",
       " 'NewsArticles-104',\n",
       " 'NewsArticles-1043',\n",
       " 'NewsArticles-1048']"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tmtoolkit.topicmod.visualize import generate_wordclouds_for_document_topics\n",
    "\n",
    "doc_clouds = generate_wordclouds_for_document_topics(\n",
    "    new_doc_topic, doc_labels_bg, topic_labels=new_topic_labels,\n",
    "    top_n=5, width=img_w, height=img_h)\n",
    "\n",
    "# show only the first 5 documents for\n",
    "# which word clouds were generated\n",
    "list(doc_clouds.keys())[:5]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To display a specific document's topic word cloud:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:55:12.347497Z",
     "iopub.status.busy": "2022-02-08T06:55:12.347064Z",
     "iopub.status.idle": "2022-02-08T06:55:12.358740Z",
     "shell.execute_reply": "2022-02-08T06:55:12.358373Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<PIL.Image.Image image mode=RGBA size=400x300 at 0x7FB69D9D9EE0>"
      ]
     },
     "execution_count": 66,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "doc_clouds['NewsArticles-1032']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can write the generated images as PNG files to a folder on disk. Here, we store all word clouds in `topic_clouds` to `'data/tm_wordclouds/'`:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:55:12.361930Z",
     "iopub.status.busy": "2022-02-08T06:55:12.361473Z",
     "iopub.status.idle": "2022-02-08T06:55:12.872692Z",
     "shell.execute_reply": "2022-02-08T06:55:12.873147Z"
    }
   },
   "outputs": [],
   "source": [
    "from tmtoolkit.topicmod.visualize import write_wordclouds_to_folder\n",
    "\n",
    "write_wordclouds_to_folder(topic_clouds, 'data/tm_wordclouds/')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Interactive visualization with pyLDAVis\n",
    "\n",
    "The [pyLDAVis package](https://pyldavis.readthedocs.io/) offers a great interactive tool to explore a topic model. The tmtoolkit function [parameters_for_ldavis](api.rst#tmtoolkit.topicmod.visualize.generate_wordclouds_for_document_topics) allows to prepare your topic model data for this package so that you can easily pass it on to pyLDAVis."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-08T06:55:12.877685Z",
     "iopub.status.busy": "2022-02-08T06:55:12.877243Z",
     "iopub.status.idle": "2022-02-08T06:55:12.878778Z",
     "shell.execute_reply": "2022-02-08T06:55:12.879157Z"
    }
   },
   "outputs": [],
   "source": [
    "from tmtoolkit.topicmod.visualize import parameters_for_ldavis\n",
    "\n",
    "ldavis_params = parameters_for_ldavis(new_topic_word,\n",
    "                                      new_doc_topic,\n",
    "                                      dtm_bg,\n",
    "                                      vocab_bg)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "If you have installed the package, you can now start the LDAVis explorer with the following lines of code in a Jupyter notebook:\n",
    "\n",
    "    import pyLDAvis\n",
    "    pyLDAVis.prepare(**ldavis_params)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.8.10"
  },
  "pycharm": {
   "stem_cell": {
    "cell_type": "raw",
    "metadata": {
     "collapsed": false
    },
    "source": []
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}