{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Gensim Tutorial on Online Non-Negative Matrix Factorization"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This notebooks explains basic ideas behind the open source NMF implementation in [Gensim](https://github.com/RaRe-Technologies/gensim), including code examples for applying NMF to text processing."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## What's in this tutorial?\n",
    "\n",
    "1. [Introduction: Why NMF?](#1.-Introduction-to-NMF)\n",
    "2. [Code example on 20 Newsgroups](#2.-Code-example:-NMF-on-20-Newsgroups)\n",
    "3. [Benchmarks against Sklearn's NMF and Gensim's LDA](#3.-Benchmarks)\n",
    "4. [Large-scale NMF training on the English Wikipedia (sparse text vectors)](#4.-NMF-on-English-Wikipedia)\n",
    "5. [NMF on face decomposition (dense image vectors)](#5.-And-now-for-something-completely-different:-Face-decomposition-from-images)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "lines_to_next_cell": 2
   },
   "source": [
    "# 1. Introduction to NMF"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "lines_to_next_cell": 2
   },
   "source": [
    "## What's in a name?\n",
    "\n",
    "Gensim's Online Non-Negative Matrix Factorization (NMF, NNMF, ONMF) implementation is based on [Renbo Zhao, Vincent Y. F. Tan: Online Nonnegative Matrix Factorization with Outliers, 2016](https://arxiv.org/abs/1604.02634) and is optimized for extremely large, sparse, streamed inputs. Such inputs happen in NLP with **unsupervised training** on massive text corpora.\n",
    "\n",
    "* Why **Online**? Because corpora and datasets in modern ML can be very large, and RAM is limited. Unlike batch algorithms, online algorithms learn iteratively, streaming through the available training examples, without loading the entire dataset into RAM or requiring random-access to the data examples.\n",
    "\n",
    "* Why **Non-Negative**? Because non-negativity leads to more interpretable, sparse \"human-friendly\" topics. This is in contrast to e.g. SVD (another popular matrix factorization method with [super-efficient implementation in Gensim](https://radimrehurek.com/gensim/models/lsimodel.html)), which produces dense negative factors and thus harder-to-interpret topics.\n",
    "\n",
    "* **Matrix factorizations** are the corner stone of modern machine learning. They can be used either directly (recommendation systems, bi-clustering, image compression, topic modeling…) or as internal routines in more complex deep learning algorithms."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "lines_to_next_cell": 2
   },
   "source": [
    "## How ONNMF works\n",
    "\n",
    "Terminology:\n",
    "- `corpus` is a stream of input documents = training examples\n",
    "- `batch` is a chunk of input corpus, a word-document matrix mini-batch that fits in RAM\n",
    "- `W` is a word-topic matrix (to be learned; stored in the resulting model)\n",
    "- `h` is a topic-document matrix (to be learned; not stored, but rather inferred for documents on-the-fly)\n",
    "- `A`, `B` - matrices that accumulate information from consecutive chunks. `A = h.dot(ht)`, `B = v.dot(ht)`.\n",
    "\n",
    "The idea behind the algorithm is as follows:\n",
    "\n",
    "```\n",
    "    Initialize W, A and B matrices\n",
    "\n",
    "    for batch in input corpus batches:\n",
    "        infer h:\n",
    "            do coordinate gradient descent step to find h that minimizes ||batch - Wh|| in L2 norm\n",
    "\n",
    "            bound h so that it is non-negative\n",
    "\n",
    "        update A and B:\n",
    "            A = h.dot(ht)\n",
    "            B = batch.dot(ht)\n",
    "\n",
    "        update W:\n",
    "            do gradient descent step to find W that minimizes ||0.5*trace(WtWA) - trace(WtB)|| in L2 norm\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 2. Code example: NMF on 20 Newsgroups"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Preprocessing"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's import the models we'll be using throughout this tutorial (`numpy==1.14.2`, `matplotlib==3.0.2`, `pandas==0.24.1`, `sklearn==0.19.1`, `gensim==3.7.1`) and set up logging at INFO level.\n",
    "\n",
    "Gensim uses logging generously to inform users what's going on. Eyeballing the logs is a good sanity check, to make sure everything is working as expected.\n",
    "\n",
    "Only `numpy` and `gensim` are actually needed to train and use NMF. The other imports are used only to make our life a little easier in this tutorial."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import logging\n",
    "import time\n",
    "from contextlib import contextmanager\n",
    "import os\n",
    "from multiprocessing import Process\n",
    "import psutil\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from numpy.random import RandomState\n",
    "from sklearn import decomposition\n",
    "from sklearn.cluster import MiniBatchKMeans\n",
    "from sklearn.datasets import fetch_olivetti_faces\n",
    "from sklearn.decomposition.nmf import NMF as SklearnNmf\n",
    "from sklearn.linear_model import LogisticRegressionCV\n",
    "from sklearn.metrics import f1_score\n",
    "\n",
    "import gensim.downloader\n",
    "from gensim import matutils, utils\n",
    "from gensim.corpora import Dictionary\n",
    "from gensim.models import CoherenceModel, LdaModel, TfidfModel, LsiModel\n",
    "from gensim.models.basemodel import BaseTopicModel\n",
    "from gensim.models.nmf import Nmf as GensimNmf\n",
    "from gensim.parsing.preprocessing import preprocess_string\n",
    "\n",
    "logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s', level=logging.INFO)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Dataset preparation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's load the notorious [20 Newsgroups dataset](http://qwone.com/~jason/20Newsgroups/) from Gensim's [repository of pre-trained models and corpora](https://github.com/RaRe-Technologies/gensim-data):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "newsgroups = gensim.downloader.load('20-newsgroups')\n",
    "\n",
    "categories = [\n",
    "    'alt.atheism',\n",
    "    'comp.graphics',\n",
    "    'rec.motorcycles',\n",
    "    'talk.politics.mideast',\n",
    "    'sci.space'\n",
    "]\n",
    "\n",
    "categories = {name: idx for idx, name in enumerate(categories)}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Create a train/test split:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "random_state = RandomState(42)\n",
    "\n",
    "trainset = np.array([\n",
    "    {\n",
    "        'data': doc['data'],\n",
    "        'target': categories[doc['topic']],\n",
    "    }\n",
    "    for doc in newsgroups\n",
    "    if doc['topic'] in categories and doc['set'] == 'train'\n",
    "])\n",
    "random_state.shuffle(trainset)\n",
    "\n",
    "testset = np.array([\n",
    "    {\n",
    "        'data': doc['data'],\n",
    "        'target': categories[doc['topic']],\n",
    "    }\n",
    "    for doc in newsgroups\n",
    "    if doc['topic'] in categories and doc['set'] == 'test'\n",
    "])\n",
    "random_state.shuffle(testset)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We'll use very [simple preprocessing with stemming](https://radimrehurek.com/gensim/parsing/preprocessing.html#gensim.parsing.preprocessing.preprocess_string) to tokenize each document. YMMV; in your application, use whatever preprocessing makes sense in your domain. Correctly preparing the input has [major impact](https://en.wikipedia.org/wiki/Garbage_in,_garbage_out) on any subsequent ML training."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_documents = [preprocess_string(doc['data']) for doc in trainset]\n",
    "test_documents = [preprocess_string(doc['data']) for doc in testset]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Dictionary compilation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's create a mapping between tokens and their ids. Another option would be a [HashDictionary](https://radimrehurek.com/gensim/corpora/hashdictionary.html), saving ourselves one pass over the training documents."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2019-05-06 15:57:16,471 : INFO : adding document #0 to Dictionary(0 unique tokens: [])\n",
      "2019-05-06 15:57:16,781 : INFO : built Dictionary(25279 unique tokens: ['sketch', 'addario', 'foyer', 'labratsat', 'reclaim']...) from 2819 documents (total 435328 corpus positions)\n",
      "2019-05-06 15:57:16,809 : INFO : discarding 18198 tokens: [('batka', 1), ('batkaj', 1), ('beatl', 1), ('ccmail', 3), ('dayton', 4), ('edu', 1785), ('inhibit', 1), ('jbatka', 1), ('line', 2748), ('organ', 2602)]...\n",
      "2019-05-06 15:57:16,810 : INFO : keeping 7081 tokens which were in no less than 5 and no more than 1409 (=50.0%) documents\n",
      "2019-05-06 15:57:16,821 : INFO : resulting dictionary: Dictionary(7081 unique tokens: ['colost', 'choke', 'editor', 'china', 'piss']...)\n"
     ]
    }
   ],
   "source": [
    "dictionary = Dictionary(train_documents)\n",
    "dictionary.filter_extremes(no_below=5, no_above=0.5, keep_n=20000)  # filter out too in/frequent tokens"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create training corpus"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's vectorize the training corpus into the bag-of-words format. We'll train LDA on a BOW and NMFs on an TF-IDF corpus:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "tfidf = TfidfModel(dictionary=dictionary)\n",
    "\n",
    "train_corpus = [\n",
    "    dictionary.doc2bow(document)\n",
    "    for document\n",
    "    in train_documents\n",
    "]\n",
    "\n",
    "test_corpus = [\n",
    "    dictionary.doc2bow(document)\n",
    "    for document\n",
    "    in test_documents\n",
    "]\n",
    "\n",
    "train_corpus_tfidf = list(tfidf[train_corpus])\n",
    "\n",
    "test_corpus_tfidf = list(tfidf[test_corpus])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here we simply stored the bag-of-words vectors into a `list`, but Gensim accepts [any iterable](https://radimrehurek.com/gensim/tut1.html#corpus-streaming-one-document-at-a-time) as input, including streamed ones. To learn more about memory-efficient input iterables, see our [Data Streaming in Python: Generators, Iterators, Iterables](https://rare-technologies.com/data-streaming-in-python-generators-iterators-iterables/) tutorial."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## NMF Model Training\n",
    "\n",
    "The API works in the same way as other Gensim models, such as [LdaModel](https://radimrehurek.com/gensim/models/ldamodel.html) or [LsiModel](https://radimrehurek.com/gensim/models/lsimodel.html)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Notable model parameters:\n",
    "\n",
    "- `kappa` float, optional\n",
    "\n",
    "    Gradient descent step size.\n",
    "    Larger value makes the model train faster, but could lead to non-convergence if set too large.\n",
    "    \n",
    "- `w_max_iter` int, optional\n",
    "\n",
    "    Maximum number of iterations to train W per each batch.\n",
    "    \n",
    "- `w_stop_condition` float, optional\n",
    "\n",
    "    If the error difference gets smaller than this, training of ``W`` stops for the current batch.\n",
    "    \n",
    "- `h_r_max_iter` int, optional\n",
    "\n",
    "    Maximum number of iterations to train h per each batch.\n",
    "    \n",
    "- `h_r_stop_condition` float, optional\n",
    "\n",
    "    If the error difference gets smaller than this, training of ``h`` stops for the current batch."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Learn an NMF model with 5 topics:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2019-05-06 15:57:18,565 : INFO : running NMF training, 5 topics, 5 passes over the supplied corpus of 2819 documents, evaluating l2 norm every 2819 documents\n",
      "2019-05-06 15:57:18,581 : INFO : PROGRESS: pass 0, at document #1000/2819\n",
      "2019-05-06 15:57:18,604 : INFO : W error: -11.552583824232896\n",
      "2019-05-06 15:57:18,619 : INFO : PROGRESS: pass 0, at document #2000/2819\n",
      "2019-05-06 15:57:18,627 : INFO : W error: -13.74803744073488\n",
      "2019-05-06 15:57:18,639 : INFO : PROGRESS: pass 0, at document #2819/2819\n",
      "2019-05-06 15:57:18,735 : INFO : L2 norm: 28.141396382337142\n",
      "2019-05-06 15:57:18,770 : INFO : topic #0 (0.395): 0.011*\"isra\" + 0.010*\"israel\" + 0.006*\"arab\" + 0.006*\"jew\" + 0.004*\"palestinian\" + 0.004*\"henri\" + 0.003*\"toronto\" + 0.003*\"question\" + 0.003*\"kill\" + 0.003*\"hernlem\"\n",
      "2019-05-06 15:57:18,771 : INFO : topic #1 (0.352): 0.008*\"space\" + 0.005*\"access\" + 0.005*\"nasa\" + 0.004*\"pat\" + 0.003*\"digex\" + 0.003*\"orbit\" + 0.003*\"shuttl\" + 0.003*\"data\" + 0.003*\"graphic\" + 0.003*\"com\"\n",
      "2019-05-06 15:57:18,772 : INFO : topic #2 (0.378): 0.012*\"armenian\" + 0.006*\"turkish\" + 0.004*\"greek\" + 0.004*\"peopl\" + 0.004*\"armenia\" + 0.004*\"turk\" + 0.004*\"argic\" + 0.004*\"bike\" + 0.003*\"serdar\" + 0.003*\"turkei\"\n",
      "2019-05-06 15:57:18,772 : INFO : topic #3 (0.412): 0.010*\"moral\" + 0.006*\"keith\" + 0.004*\"anim\" + 0.003*\"jake\" + 0.003*\"boni\" + 0.003*\"instinct\" + 0.003*\"act\" + 0.003*\"think\" + 0.003*\"object\" + 0.003*\"caltech\"\n",
      "2019-05-06 15:57:18,773 : INFO : topic #4 (0.428): 0.009*\"islam\" + 0.008*\"god\" + 0.006*\"livesei\" + 0.006*\"muslim\" + 0.005*\"imag\" + 0.005*\"sgi\" + 0.005*\"jaeger\" + 0.004*\"jon\" + 0.004*\"solntz\" + 0.004*\"wpd\"\n",
      "2019-05-06 15:57:18,776 : INFO : W error: -14.346700852834074\n",
      "2019-05-06 15:57:18,791 : INFO : PROGRESS: pass 1, at document #1000/2819\n",
      "2019-05-06 15:57:18,797 : INFO : W error: -15.92349117004869\n",
      "2019-05-06 15:57:18,811 : INFO : PROGRESS: pass 1, at document #2000/2819\n",
      "2019-05-06 15:57:18,816 : INFO : W error: -17.04437515427273\n",
      "2019-05-06 15:57:18,829 : INFO : PROGRESS: pass 1, at document #2819/2819\n",
      "2019-05-06 15:57:18,922 : INFO : L2 norm: 28.021861174968205\n",
      "2019-05-06 15:57:18,956 : INFO : topic #0 (0.341): 0.014*\"israel\" + 0.013*\"isra\" + 0.008*\"arab\" + 0.007*\"jew\" + 0.005*\"palestinian\" + 0.004*\"lebanes\" + 0.003*\"peac\" + 0.003*\"henri\" + 0.003*\"attack\" + 0.003*\"polici\"\n",
      "2019-05-06 15:57:18,956 : INFO : topic #1 (0.252): 0.008*\"space\" + 0.005*\"nasa\" + 0.004*\"access\" + 0.003*\"pat\" + 0.003*\"orbit\" + 0.003*\"digex\" + 0.003*\"launch\" + 0.003*\"shuttl\" + 0.003*\"graphic\" + 0.003*\"com\"\n",
      "2019-05-06 15:57:18,957 : INFO : topic #2 (0.295): 0.020*\"armenian\" + 0.010*\"turkish\" + 0.006*\"armenia\" + 0.006*\"turk\" + 0.006*\"argic\" + 0.006*\"serdar\" + 0.005*\"greek\" + 0.005*\"turkei\" + 0.004*\"peopl\" + 0.004*\"genocid\"\n",
      "2019-05-06 15:57:18,958 : INFO : topic #3 (0.345): 0.013*\"moral\" + 0.011*\"keith\" + 0.006*\"object\" + 0.005*\"caltech\" + 0.004*\"schneider\" + 0.004*\"anim\" + 0.004*\"jake\" + 0.004*\"allan\" + 0.004*\"boni\" + 0.004*\"cco\"\n",
      "2019-05-06 15:57:18,959 : INFO : topic #4 (0.375): 0.011*\"islam\" + 0.011*\"god\" + 0.006*\"livesei\" + 0.006*\"sgi\" + 0.006*\"jaeger\" + 0.005*\"muslim\" + 0.005*\"jon\" + 0.005*\"imag\" + 0.005*\"religion\" + 0.005*\"solntz\"\n",
      "2019-05-06 15:57:18,961 : INFO : W error: -17.08829704968913\n",
      "2019-05-06 15:57:18,975 : INFO : PROGRESS: pass 2, at document #1000/2819\n",
      "2019-05-06 15:57:18,981 : INFO : W error: -17.73961065930116\n",
      "2019-05-06 15:57:18,996 : INFO : PROGRESS: pass 2, at document #2000/2819\n",
      "2019-05-06 15:57:19,001 : INFO : W error: -18.289085863153712\n",
      "2019-05-06 15:57:19,013 : INFO : PROGRESS: pass 2, at document #2819/2819\n",
      "2019-05-06 15:57:19,107 : INFO : L2 norm: 28.001900396967052\n",
      "2019-05-06 15:57:19,141 : INFO : topic #0 (0.341): 0.014*\"israel\" + 0.014*\"isra\" + 0.009*\"arab\" + 0.008*\"jew\" + 0.005*\"palestinian\" + 0.004*\"lebanes\" + 0.004*\"peac\" + 0.003*\"attack\" + 0.003*\"polici\" + 0.003*\"lebanon\"\n",
      "2019-05-06 15:57:19,142 : INFO : topic #1 (0.230): 0.007*\"space\" + 0.005*\"nasa\" + 0.004*\"access\" + 0.003*\"orbit\" + 0.003*\"pat\" + 0.003*\"launch\" + 0.003*\"digex\" + 0.003*\"gov\" + 0.003*\"com\" + 0.003*\"graphic\"\n",
      "2019-05-06 15:57:19,142 : INFO : topic #2 (0.284): 0.021*\"armenian\" + 0.011*\"turkish\" + 0.007*\"armenia\" + 0.007*\"turk\" + 0.007*\"argic\" + 0.006*\"serdar\" + 0.006*\"turkei\" + 0.005*\"greek\" + 0.005*\"genocid\" + 0.004*\"soviet\"\n",
      "2019-05-06 15:57:19,143 : INFO : topic #3 (0.346): 0.015*\"moral\" + 0.013*\"keith\" + 0.006*\"object\" + 0.006*\"caltech\" + 0.005*\"schneider\" + 0.005*\"allan\" + 0.005*\"cco\" + 0.004*\"anim\" + 0.004*\"jake\" + 0.003*\"boni\"\n",
      "2019-05-06 15:57:19,144 : INFO : topic #4 (0.365): 0.012*\"islam\" + 0.011*\"god\" + 0.006*\"livesei\" + 0.006*\"sgi\" + 0.006*\"jaeger\" + 0.005*\"muslim\" + 0.005*\"jon\" + 0.005*\"religion\" + 0.004*\"solntz\" + 0.004*\"wpd\"\n",
      "2019-05-06 15:57:19,146 : INFO : W error: -18.220202313431095\n",
      "2019-05-06 15:57:19,160 : INFO : PROGRESS: pass 3, at document #1000/2819\n",
      "2019-05-06 15:57:19,165 : INFO : W error: -18.590446221955172\n",
      "2019-05-06 15:57:19,180 : INFO : PROGRESS: pass 3, at document #2000/2819\n",
      "2019-05-06 15:57:19,185 : INFO : W error: -18.936998738726114\n",
      "2019-05-06 15:57:19,197 : INFO : PROGRESS: pass 3, at document #2819/2819\n",
      "2019-05-06 15:57:19,291 : INFO : L2 norm: 27.993018072469805\n",
      "2019-05-06 15:57:19,324 : INFO : topic #0 (0.348): 0.015*\"israel\" + 0.014*\"isra\" + 0.009*\"arab\" + 0.008*\"jew\" + 0.005*\"palestinian\" + 0.004*\"lebanes\" + 0.004*\"peac\" + 0.003*\"attack\" + 0.003*\"lebanon\" + 0.003*\"polici\"\n",
      "2019-05-06 15:57:19,325 : INFO : topic #1 (0.221): 0.007*\"space\" + 0.005*\"nasa\" + 0.003*\"access\" + 0.003*\"orbit\" + 0.003*\"pat\" + 0.003*\"launch\" + 0.003*\"gov\" + 0.003*\"digex\" + 0.003*\"com\" + 0.002*\"graphic\"\n",
      "2019-05-06 15:57:19,325 : INFO : topic #2 (0.281): 0.022*\"armenian\" + 0.011*\"turkish\" + 0.007*\"armenia\" + 0.007*\"turk\" + 0.007*\"argic\" + 0.007*\"serdar\" + 0.006*\"turkei\" + 0.005*\"greek\" + 0.005*\"genocid\" + 0.005*\"soviet\"\n",
      "2019-05-06 15:57:19,326 : INFO : topic #3 (0.349): 0.016*\"moral\" + 0.014*\"keith\" + 0.007*\"object\" + 0.006*\"caltech\" + 0.006*\"schneider\" + 0.005*\"allan\" + 0.005*\"cco\" + 0.004*\"anim\" + 0.004*\"natur\" + 0.003*\"think\"\n",
      "2019-05-06 15:57:19,327 : INFO : topic #4 (0.365): 0.012*\"islam\" + 0.012*\"god\" + 0.006*\"sgi\" + 0.006*\"livesei\" + 0.006*\"jaeger\" + 0.005*\"muslim\" + 0.005*\"religion\" + 0.005*\"atheist\" + 0.005*\"jon\" + 0.005*\"atheism\"\n",
      "2019-05-06 15:57:19,328 : INFO : W error: -18.84541532652092\n",
      "2019-05-06 15:57:19,342 : INFO : PROGRESS: pass 4, at document #1000/2819\n",
      "2019-05-06 15:57:19,347 : INFO : W error: -19.091058241700402\n",
      "2019-05-06 15:57:19,362 : INFO : PROGRESS: pass 4, at document #2000/2819\n",
      "2019-05-06 15:57:19,367 : INFO : W error: -19.338453391122066\n",
      "2019-05-06 15:57:19,378 : INFO : PROGRESS: pass 4, at document #2819/2819\n",
      "2019-05-06 15:57:19,473 : INFO : L2 norm: 27.988158841345246\n",
      "2019-05-06 15:57:19,506 : INFO : topic #0 (0.352): 0.015*\"israel\" + 0.014*\"isra\" + 0.009*\"arab\" + 0.008*\"jew\" + 0.005*\"palestinian\" + 0.004*\"lebanes\" + 0.004*\"peac\" + 0.003*\"attack\" + 0.003*\"lebanon\" + 0.003*\"polici\"\n",
      "2019-05-06 15:57:19,507 : INFO : topic #1 (0.210): 0.007*\"space\" + 0.005*\"nasa\" + 0.003*\"access\" + 0.003*\"orbit\" + 0.003*\"launch\" + 0.003*\"pat\" + 0.003*\"gov\" + 0.003*\"com\" + 0.002*\"alaska\" + 0.002*\"graphic\"\n",
      "2019-05-06 15:57:19,508 : INFO : topic #2 (0.282): 0.023*\"armenian\" + 0.011*\"turkish\" + 0.008*\"armenia\" + 0.007*\"argic\" + 0.007*\"turk\" + 0.007*\"serdar\" + 0.006*\"turkei\" + 0.005*\"greek\" + 0.005*\"genocid\" + 0.005*\"soviet\"\n",
      "2019-05-06 15:57:19,509 : INFO : topic #3 (0.353): 0.016*\"moral\" + 0.015*\"keith\" + 0.007*\"object\" + 0.007*\"caltech\" + 0.006*\"schneider\" + 0.005*\"allan\" + 0.005*\"cco\" + 0.004*\"anim\" + 0.004*\"natur\" + 0.004*\"goal\"\n",
      "2019-05-06 15:57:19,509 : INFO : topic #4 (0.367): 0.012*\"god\" + 0.012*\"islam\" + 0.006*\"jaeger\" + 0.006*\"sgi\" + 0.006*\"livesei\" + 0.005*\"muslim\" + 0.005*\"atheist\" + 0.005*\"religion\" + 0.005*\"atheism\" + 0.004*\"jon\"\n",
      "2019-05-06 15:57:19,511 : INFO : W error: -19.245120389312117\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 1.52 s, sys: 1.84 s, total: 3.36 s\n",
      "Wall time: 947 ms\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "nmf = GensimNmf(\n",
    "    corpus=train_corpus_tfidf,\n",
    "    num_topics=5,\n",
    "    id2word=dictionary,\n",
    "    chunksize=1000,\n",
    "    passes=5,\n",
    "    eval_every=10,\n",
    "    minimum_probability=0,\n",
    "    random_state=0,\n",
    "    kappa=1,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "W = nmf.get_topics().T\n",
    "\n",
    "dense_test_corpus = matutils.corpus2dense(\n",
    "    test_corpus_tfidf,\n",
    "    num_terms=W.shape[0],\n",
    ")\n",
    "\n",
    "if isinstance(nmf, SklearnNmf):\n",
    "    H = nmf.transform(dense_test_corpus.T).T\n",
    "else:\n",
    "    H = np.zeros((nmf.num_topics, len(test_corpus_tfidf)))\n",
    "    for bow_id, bow in enumerate(test_corpus_tfidf):\n",
    "        for topic_id, word_count in nmf[bow]:\n",
    "            H[topic_id, bow_id] = word_count"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.105176733465657"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.linalg.norm(W.dot(H))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "43.312817"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.linalg.norm(dense_test_corpus)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### View the learned topics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(0,\n",
       "  '0.015*\"israel\" + 0.014*\"isra\" + 0.009*\"arab\" + 0.008*\"jew\" + 0.005*\"palestinian\" + 0.004*\"lebanes\" + 0.004*\"peac\" + 0.004*\"attack\" + 0.004*\"lebanon\" + 0.003*\"polici\"'),\n",
       " (1,\n",
       "  '0.007*\"space\" + 0.005*\"nasa\" + 0.003*\"access\" + 0.003*\"orbit\" + 0.003*\"launch\" + 0.003*\"pat\" + 0.003*\"gov\" + 0.003*\"com\" + 0.002*\"alaska\" + 0.002*\"moon\"'),\n",
       " (2,\n",
       "  '0.023*\"armenian\" + 0.011*\"turkish\" + 0.008*\"armenia\" + 0.007*\"argic\" + 0.007*\"turk\" + 0.007*\"serdar\" + 0.006*\"turkei\" + 0.005*\"greek\" + 0.005*\"genocid\" + 0.005*\"soviet\"'),\n",
       " (3,\n",
       "  '0.016*\"moral\" + 0.015*\"keith\" + 0.007*\"object\" + 0.007*\"caltech\" + 0.006*\"schneider\" + 0.005*\"allan\" + 0.005*\"cco\" + 0.004*\"anim\" + 0.004*\"natur\" + 0.004*\"goal\"'),\n",
       " (4,\n",
       "  '0.012*\"god\" + 0.012*\"islam\" + 0.006*\"jaeger\" + 0.006*\"sgi\" + 0.005*\"livesei\" + 0.005*\"muslim\" + 0.005*\"atheist\" + 0.005*\"religion\" + 0.005*\"atheism\" + 0.004*\"rushdi\"')]"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nmf.show_topics()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Evaluation measure: Coherence\n",
    "\n",
    "[Topic coherence](http://qpleple.com/topic-coherence-to-evaluate-topic-models/) measures how often do most frequent tokens from each topic co-occur in one document. Larger is better."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2019-05-06 15:57:20,590 : INFO : CorpusAccumulator accumulated stats from 1000 documents\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "-4.1310114675795875"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "CoherenceModel(\n",
    "    model=nmf,\n",
    "    corpus=test_corpus_tfidf,\n",
    "    coherence='u_mass'\n",
    ").get_coherence()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "lines_to_next_cell": 2
   },
   "source": [
    "## Topic inference on new documents\n",
    "\n",
    "With the NMF model trained, let's fetch one news document not seen during training, and infer its topic vector."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "From: spl@ivem.ucsd.edu (Steve Lamont)\n",
      "Subject: Re: RGB to HVS, and back\n",
      "Organization: University of Calif., San Diego/Microscopy and Imaging Resource\n",
      "Lines: 18\n",
      "Distribution: world\n",
      "NNTP-Posting-Host: ivem.ucsd.edu\n",
      "\n",
      "In article <ltu4buINNe7j@caspian.usc.edu> zyeh@caspian.usc.edu (zhenghao yeh) writes:\n",
      ">|> See Foley, van Dam, Feiner, and Hughes, _Computer Graphics: Principles\n",
      ">|> and Practice, Second Edition_.\n",
      ">|> \n",
      ">|> [If people would *read* this book, 75 percent of the questions in this\n",
      ">|> froup would disappear overnight...]\n",
      ">|> \n",
      ">\tNot really. I think it is less than 10%.\n",
      "\n",
      "Nah... I figure most people would be so busy reading that they wouldn't\n",
      "have *time* to post. :-) :-) :-)\n",
      "\n",
      "\t\t\t\t\t\t\tspl\n",
      "-- \n",
      "Steve Lamont, SciViGuy -- (619) 534-7968 -- spl@szechuan.ucsd.edu\n",
      "San Diego Microscopy and Imaging Resource/UC San Diego/La Jolla, CA 92093-0608\n",
      "\"Until I meet you, then, in Upper Hell\n",
      "Convulsed, foaming immortal blood: farewell\" - J. Berryman, \"A Professor's Song\"\n",
      "\n",
      "====================================================================================================\n",
      "Topics: [(0, 0.10199317206513686), (1, 0.39976628221371285), (2, 0.1428263926167706), (3, 0.0333080734922002), (4, 0.3221060796121796)]\n"
     ]
    }
   ],
   "source": [
    "print(testset[0]['data'])\n",
    "print('=' * 100)\n",
    "print(\"Topics: {}\".format(nmf[test_corpus[0]]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Word topic inference\n",
    "\n",
    "Similarly, we can inspect the topic distribution assigned to a vocabulary term:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "lines_to_next_cell": 2
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Word: actual\n",
      "Topics: [(0, 0.15401782844659068), (1, 0.2829834256007429), (2, 0.04354905106817273), (3, 0.25783766798021135), (4, 0.26161202690428226)]\n"
     ]
    }
   ],
   "source": [
    "word = dictionary[0]\n",
    "print(\"Word: {}\".format(word))\n",
    "print(\"Topics: {}\".format(nmf.get_term_topics(word)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Internal NMF state"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Density is a fraction of non-zero elements in a matrix."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "lines_to_next_cell": 2
   },
   "outputs": [],
   "source": [
    "def density(matrix):\n",
    "    return (matrix > 0).mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Term-topic matrix of shape `(words, topics)`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Density: 0.6872475639034035\n"
     ]
    }
   ],
   "source": [
    "print(\"Density: {}\".format(density(nmf._W)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Topic-document matrix for the last batch of shape `(topics, batch)`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Density: 0.6615384615384615\n"
     ]
    }
   ],
   "source": [
    "print(\"Density: {}\".format(density(nmf._h)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3. Benchmarks"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Gensim NMF vs Sklearn NMF vs Gensim LDA\n",
    "\n",
    "We'll run these three unsupervised models on the [20newsgroups](https://scikit-learn.org/0.19/datasets/twenty_newsgroups.html) dataset.\n",
    "\n",
    "20 Newsgroups also contains labels for each document, which will allow us to evaluate the trained models on an \"upstream\" classification task, using the unsupervised document topics as input features."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Metrics\n",
    "\n",
    "We'll track these metrics as we train and test NMF on the 20-newsgroups corpus we created above:\n",
    "- `train time` - time to train a model\n",
    "- `mean_ram` - mean RAM consumption during training\n",
    "- `max_ram` - maximum RAM consumption during training\n",
    "- `train time` - time to train a model.\n",
    "- `coherence` - coherence score (larger is better).\n",
    "- `l2_norm` - L2 norm of `v - Wh` (less is better, not defined for LDA).\n",
    "- `f1` - [F1 score](https://en.wikipedia.org/wiki/F1_score) on the task of news topic classification (larger is better)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "lines_to_next_cell": 2
   },
   "outputs": [],
   "source": [
    "fixed_params = dict(\n",
    "    chunksize=1000,\n",
    "    num_topics=5,\n",
    "    id2word=dictionary,\n",
    "    passes=5,\n",
    "    eval_every=10,\n",
    "    minimum_probability=0,\n",
    "    random_state=0,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "@contextmanager\n",
    "def measure_ram(output, tick=5):\n",
    "    def _measure_ram(pid, output, tick=tick):\n",
    "        py = psutil.Process(pid)\n",
    "        with open(output, 'w') as outfile:\n",
    "            while True:\n",
    "                memory = py.memory_info().rss\n",
    "                outfile.write(\"{}\\n\".format(memory))\n",
    "                outfile.flush()\n",
    "                time.sleep(tick)\n",
    "\n",
    "    pid = os.getpid()\n",
    "    p = Process(target=_measure_ram, args=(pid, output, tick))\n",
    "    p.start()\n",
    "    yield\n",
    "    p.terminate()\n",
    "\n",
    "\n",
    "def get_train_time_and_ram(func, name, tick=5):\n",
    "    memprof_filename = \"{}.memprof\".format(name)\n",
    "\n",
    "    start = time.time()\n",
    "\n",
    "    with measure_ram(memprof_filename, tick=tick):\n",
    "        result = func()\n",
    "\n",
    "    elapsed_time = pd.to_timedelta(time.time() - start, unit='s').round('ms')\n",
    "\n",
    "    memprof_df = pd.read_csv(memprof_filename, squeeze=True)\n",
    "\n",
    "    mean_ram = \"{} MB\".format(\n",
    "        int(memprof_df.mean() // 2 ** 20),\n",
    "    )\n",
    "\n",
    "    max_ram = \"{} MB\".format(int(memprof_df.max() // 2 ** 20))\n",
    "\n",
    "    return elapsed_time, mean_ram, max_ram, result\n",
    "\n",
    "\n",
    "def get_f1(model, train_corpus, X_test, y_train, y_test):\n",
    "    if isinstance(model, SklearnNmf):\n",
    "        dense_train_corpus = matutils.corpus2dense(\n",
    "            train_corpus,\n",
    "            num_terms=model.components_.shape[1],\n",
    "        )\n",
    "        X_train = model.transform(dense_train_corpus.T)\n",
    "    else:\n",
    "        X_train = np.zeros((len(train_corpus), model.num_topics))\n",
    "        for bow_id, bow in enumerate(train_corpus):\n",
    "            for topic_id, word_count in model[bow]:\n",
    "                X_train[bow_id, topic_id] = word_count\n",
    "\n",
    "    log_reg = LogisticRegressionCV(multi_class='multinomial', cv=5)\n",
    "    log_reg.fit(X_train, y_train)\n",
    "\n",
    "    pred_labels = log_reg.predict(X_test)\n",
    "\n",
    "    return f1_score(y_test, pred_labels, average='micro')\n",
    "\n",
    "def get_sklearn_topics(model, top_n=5):\n",
    "    topic_probas = model.components_.T\n",
    "    topic_probas = topic_probas / topic_probas.sum(axis=0)\n",
    "\n",
    "    sparsity = np.zeros(topic_probas.shape[1])\n",
    "\n",
    "    for row in topic_probas:\n",
    "        sparsity += (row == 0)\n",
    "\n",
    "    sparsity /= topic_probas.shape[1]\n",
    "\n",
    "    topic_probas = topic_probas[:, sparsity.argsort()[::-1]][:, :top_n]\n",
    "\n",
    "    token_indices = topic_probas.argsort(axis=0)[:-11:-1, :]\n",
    "    topic_probas.sort(axis=0)\n",
    "    topic_probas = topic_probas[:-11:-1, :]\n",
    "\n",
    "    topics = []\n",
    "\n",
    "    for topic_idx in range(topic_probas.shape[1]):\n",
    "        tokens = [\n",
    "            model.id2word[token_idx]\n",
    "            for token_idx\n",
    "            in token_indices[:, topic_idx]\n",
    "        ]\n",
    "        topic = (\n",
    "            '{}*\"{}\"'.format(round(proba, 3), token)\n",
    "            for proba, token\n",
    "            in zip(topic_probas[:, topic_idx], tokens)\n",
    "        )\n",
    "        topic = \" + \".join(topic)\n",
    "        topics.append((topic_idx, topic))\n",
    "\n",
    "    return topics\n",
    "\n",
    "def get_metrics(model, test_corpus, train_corpus=None, y_train=None, y_test=None, dictionary=None):\n",
    "    if isinstance(model, SklearnNmf):\n",
    "        model.get_topics = lambda: model.components_\n",
    "        model.show_topics = lambda top_n: get_sklearn_topics(model, top_n)\n",
    "        model.id2word = dictionary\n",
    "\n",
    "    W = model.get_topics().T\n",
    "\n",
    "    dense_test_corpus = matutils.corpus2dense(\n",
    "        test_corpus,\n",
    "        num_terms=W.shape[0],\n",
    "    )\n",
    "\n",
    "    if isinstance(model, SklearnNmf):\n",
    "        H = model.transform(dense_test_corpus.T).T\n",
    "    else:\n",
    "        H = np.zeros((model.num_topics, len(test_corpus)))\n",
    "        for bow_id, bow in enumerate(test_corpus):\n",
    "            for topic_id, word_count in model[bow]:\n",
    "                H[topic_id, bow_id] = word_count\n",
    "\n",
    "    l2_norm = None\n",
    "\n",
    "    if not isinstance(model, LdaModel):\n",
    "        pred_factors = W.dot(H)\n",
    "\n",
    "        l2_norm = np.linalg.norm(pred_factors - dense_test_corpus)\n",
    "        l2_norm = round(l2_norm, 4)\n",
    "\n",
    "    f1 = None\n",
    "\n",
    "    if train_corpus and y_train and y_test:\n",
    "        f1 = get_f1(model, train_corpus, H.T, y_train, y_test)\n",
    "        f1 = round(f1, 4)\n",
    "\n",
    "    model.normalize = True\n",
    "\n",
    "    coherence = CoherenceModel(\n",
    "        model=model,\n",
    "        corpus=test_corpus,\n",
    "        coherence='u_mass'\n",
    "    ).get_coherence()\n",
    "    coherence = round(coherence, 4)\n",
    "\n",
    "    topics = model.show_topics(5)\n",
    "\n",
    "    model.normalize = False\n",
    "\n",
    "    return dict(\n",
    "        coherence=coherence,\n",
    "        l2_norm=l2_norm,\n",
    "        f1=f1,\n",
    "        topics=topics,\n",
    "    )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Run the models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "tm_metrics = pd.DataFrame(columns=['model', 'train_time', 'coherence', 'l2_norm', 'f1', 'topics'])\n",
    "\n",
    "y_train = [doc['target'] for doc in trainset]\n",
    "y_test = [doc['target'] for doc in testset]\n",
    "\n",
    "# LDA metrics\n",
    "row = {}\n",
    "row['model'] = 'lda'\n",
    "row['train_time'], row['mean_ram'], row['max_ram'], lda = get_train_time_and_ram(\n",
    "    lambda: LdaModel(\n",
    "        corpus=train_corpus,\n",
    "        **fixed_params,\n",
    "    ),\n",
    "    'lda',\n",
    "    0.1,\n",
    ")\n",
    "row.update(get_metrics(\n",
    "    lda, test_corpus, train_corpus, y_train, y_test,\n",
    "))\n",
    "tm_metrics = tm_metrics.append(pd.Series(row), ignore_index=True)\n",
    "\n",
    "# LSI metrics\n",
    "row = {}\n",
    "row['model'] = 'lsi'\n",
    "row['train_time'], row['mean_ram'], row['max_ram'], lsi = get_train_time_and_ram(\n",
    "    lambda: LsiModel(\n",
    "        corpus=train_corpus_tfidf,\n",
    "        num_topics=5,\n",
    "        id2word=dictionary,\n",
    "        chunksize=2000,\n",
    "    ),\n",
    "    'lsi',\n",
    "    0.1,\n",
    ")\n",
    "row.update(get_metrics(\n",
    "    lsi, test_corpus_tfidf, train_corpus_tfidf, y_train, y_test,\n",
    "))\n",
    "tm_metrics = tm_metrics.append(pd.Series(row), ignore_index=True)\n",
    "\n",
    "# Sklearn NMF metrics\n",
    "row = {}\n",
    "row['model'] = 'sklearn_nmf'\n",
    "train_csc_corpus_tfidf = matutils.corpus2csc(train_corpus_tfidf, len(dictionary)).T\n",
    "row['train_time'], row['mean_ram'], row['max_ram'], sklearn_nmf = get_train_time_and_ram(\n",
    "    lambda: SklearnNmf(n_components=5, random_state=42).fit(train_csc_corpus_tfidf),\n",
    "    'sklearn_nmf',\n",
    "    0.1,\n",
    ")\n",
    "row.update(get_metrics(\n",
    "    sklearn_nmf, test_corpus_tfidf, train_corpus_tfidf, y_train, y_test, dictionary,\n",
    "))\n",
    "tm_metrics = tm_metrics.append(pd.Series(row), ignore_index=True)\n",
    "\n",
    "# Gensim NMF metrics\n",
    "row = {}\n",
    "row['model'] = 'gensim_nmf'\n",
    "row['train_time'], row['mean_ram'], row['max_ram'], gensim_nmf = get_train_time_and_ram(\n",
    "    lambda: GensimNmf(\n",
    "        normalize=False,\n",
    "        corpus=train_corpus_tfidf,\n",
    "        **fixed_params\n",
    "    ),\n",
    "    'gensim_nmf',\n",
    "    0.1,\n",
    ")\n",
    "row.update(get_metrics(\n",
    "    gensim_nmf, test_corpus_tfidf, train_corpus_tfidf, y_train, y_test,\n",
    "))\n",
    "tm_metrics = tm_metrics.append(pd.Series(row), ignore_index=True)\n",
    "tm_metrics.replace(np.nan, '-', inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Benchmark results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model</th>\n",
       "      <th>train_time</th>\n",
       "      <th>coherence</th>\n",
       "      <th>l2_norm</th>\n",
       "      <th>f1</th>\n",
       "      <th>max_ram</th>\n",
       "      <th>mean_ram</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>lda</td>\n",
       "      <td>00:00:08.862000</td>\n",
       "      <td>-2.1054</td>\n",
       "      <td>-</td>\n",
       "      <td>0.7511</td>\n",
       "      <td>366 MB</td>\n",
       "      <td>366 MB</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>lsi</td>\n",
       "      <td>00:00:00.332000</td>\n",
       "      <td>-5.7010</td>\n",
       "      <td>42.4642</td>\n",
       "      <td>0.8587</td>\n",
       "      <td>381 MB</td>\n",
       "      <td>379 MB</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>sklearn_nmf</td>\n",
       "      <td>00:00:00.166000</td>\n",
       "      <td>-3.1835</td>\n",
       "      <td>42.4759</td>\n",
       "      <td>0.7889</td>\n",
       "      <td>378 MB</td>\n",
       "      <td>378 MB</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>gensim_nmf</td>\n",
       "      <td>00:00:00.954000</td>\n",
       "      <td>-4.1310</td>\n",
       "      <td>42.5487</td>\n",
       "      <td>0.8065</td>\n",
       "      <td>379 MB</td>\n",
       "      <td>379 MB</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         model      train_time  coherence  l2_norm      f1 max_ram mean_ram\n",
       "0          lda 00:00:08.862000    -2.1054        -  0.7511  366 MB   366 MB\n",
       "1          lsi 00:00:00.332000    -5.7010  42.4642  0.8587  381 MB   379 MB\n",
       "2  sklearn_nmf 00:00:00.166000    -3.1835  42.4759  0.7889  378 MB   378 MB\n",
       "3   gensim_nmf 00:00:00.954000    -4.1310  42.5487  0.8065  379 MB   379 MB"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tm_metrics.drop('topics', axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Main insights\n",
    "\n",
    "- LDA has the best coherence of all models.\n",
    "- LSI has the best l2 norm and f1 performance on downstream task (it's factors aren't non-negative though).\n",
    "- Gensim NMF, Sklearn NMF and LSI has a bit larger memory footprint than that of LDA.\n",
    "- Gensim NMF, Sklearn NMF and LSI are much faster than LDA."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Learned topics"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's inspect the 5 topics learned by each of the three models:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "lda:\n",
      "(0, '0.013*\"space\" + 0.008*\"imag\" + 0.007*\"nasa\" + 0.006*\"graphic\" + 0.006*\"program\" + 0.005*\"launch\" + 0.005*\"file\" + 0.005*\"com\" + 0.005*\"new\" + 0.004*\"orbit\"')\n",
      "(1, '0.015*\"com\" + 0.007*\"like\" + 0.007*\"nntp\" + 0.007*\"host\" + 0.006*\"know\" + 0.006*\"univers\" + 0.005*\"henri\" + 0.005*\"work\" + 0.005*\"bit\" + 0.005*\"think\"')\n",
      "(2, '0.014*\"armenian\" + 0.011*\"peopl\" + 0.009*\"turkish\" + 0.007*\"jew\" + 0.007*\"said\" + 0.006*\"right\" + 0.005*\"know\" + 0.005*\"kill\" + 0.005*\"isra\" + 0.005*\"turkei\"')\n",
      "(3, '0.012*\"com\" + 0.010*\"israel\" + 0.009*\"bike\" + 0.006*\"isra\" + 0.006*\"dod\" + 0.005*\"like\" + 0.005*\"ride\" + 0.005*\"host\" + 0.005*\"nntp\" + 0.005*\"motorcycl\"')\n",
      "(4, '0.011*\"god\" + 0.008*\"peopl\" + 0.007*\"think\" + 0.006*\"exist\" + 0.006*\"univers\" + 0.005*\"com\" + 0.005*\"believ\" + 0.005*\"islam\" + 0.005*\"moral\" + 0.005*\"christian\"')\n",
      "\n",
      "lsi:\n",
      "(0, '0.157*\"armenian\" + 0.118*\"israel\" + 0.110*\"peopl\" + 0.110*\"isra\" + 0.105*\"space\" + 0.098*\"com\" + 0.097*\"god\" + 0.084*\"jew\" + 0.082*\"think\" + 0.081*\"turkish\"')\n",
      "(1, '-0.476*\"armenian\" + -0.231*\"turkish\" + -0.157*\"armenia\" + -0.151*\"argic\" + -0.149*\"serdar\" + -0.145*\"turk\" + 0.128*\"space\" + -0.117*\"turkei\" + -0.111*\"genocid\" + 0.107*\"nasa\"')\n",
      "(2, '0.295*\"israel\" + -0.291*\"armenian\" + 0.273*\"isra\" + 0.182*\"arab\" + 0.157*\"jew\" + -0.143*\"space\" + -0.134*\"turkish\" + -0.112*\"nasa\" + 0.106*\"jake\" + 0.103*\"boni\"')\n",
      "(3, '0.274*\"keith\" + 0.252*\"moral\" + -0.235*\"israel\" + 0.213*\"god\" + -0.213*\"isra\" + 0.165*\"livesei\" + -0.143*\"arab\" + 0.123*\"sgi\" + 0.118*\"caltech\" + 0.114*\"islam\"')\n",
      "(4, '0.240*\"henri\" + -0.215*\"bike\" + 0.210*\"space\" + 0.167*\"toronto\" + 0.158*\"nasa\" + 0.148*\"moral\" + 0.143*\"keith\" + -0.142*\"graphic\" + 0.128*\"alaska\" + 0.125*\"orbit\"')\n",
      "\n",
      "sklearn_nmf:\n",
      "(0, '0.027*\"armenian\" + 0.013*\"turkish\" + 0.009*\"armenia\" + 0.009*\"argic\" + 0.009*\"serdar\" + 0.008*\"turk\" + 0.007*\"turkei\" + 0.006*\"genocid\" + 0.006*\"soviet\" + 0.006*\"zuma\"')\n",
      "(1, '0.015*\"israel\" + 0.014*\"isra\" + 0.01*\"arab\" + 0.008*\"jew\" + 0.005*\"palestinian\" + 0.005*\"jake\" + 0.005*\"boni\" + 0.004*\"lebanes\" + 0.004*\"peac\" + 0.004*\"adam\"')\n",
      "(2, '0.011*\"god\" + 0.01*\"keith\" + 0.01*\"moral\" + 0.006*\"islam\" + 0.006*\"livesei\" + 0.006*\"atheist\" + 0.005*\"atheism\" + 0.005*\"caltech\" + 0.004*\"religion\" + 0.004*\"object\"')\n",
      "(3, '0.011*\"space\" + 0.008*\"nasa\" + 0.008*\"henri\" + 0.006*\"orbit\" + 0.005*\"toronto\" + 0.005*\"alaska\" + 0.005*\"launch\" + 0.005*\"moon\" + 0.004*\"gov\" + 0.004*\"access\"')\n",
      "(4, '0.005*\"bike\" + 0.005*\"graphic\" + 0.004*\"file\" + 0.004*\"imag\" + 0.003*\"com\" + 0.003*\"ride\" + 0.003*\"thank\" + 0.003*\"program\" + 0.003*\"motorcycl\" + 0.002*\"look\"')\n",
      "\n",
      "gensim_nmf:\n",
      "(0, '0.015*\"israel\" + 0.014*\"isra\" + 0.009*\"arab\" + 0.008*\"jew\" + 0.005*\"palestinian\" + 0.004*\"lebanes\" + 0.004*\"peac\" + 0.004*\"attack\" + 0.004*\"lebanon\" + 0.003*\"polici\"')\n",
      "(1, '0.007*\"space\" + 0.005*\"nasa\" + 0.003*\"access\" + 0.003*\"orbit\" + 0.003*\"launch\" + 0.003*\"pat\" + 0.003*\"gov\" + 0.003*\"com\" + 0.002*\"alaska\" + 0.002*\"moon\"')\n",
      "(2, '0.023*\"armenian\" + 0.011*\"turkish\" + 0.008*\"armenia\" + 0.007*\"argic\" + 0.007*\"turk\" + 0.007*\"serdar\" + 0.006*\"turkei\" + 0.005*\"greek\" + 0.005*\"genocid\" + 0.005*\"soviet\"')\n",
      "(3, '0.016*\"moral\" + 0.015*\"keith\" + 0.007*\"object\" + 0.007*\"caltech\" + 0.006*\"schneider\" + 0.005*\"allan\" + 0.005*\"cco\" + 0.004*\"anim\" + 0.004*\"natur\" + 0.004*\"goal\"')\n",
      "(4, '0.012*\"god\" + 0.012*\"islam\" + 0.006*\"jaeger\" + 0.006*\"sgi\" + 0.005*\"livesei\" + 0.005*\"muslim\" + 0.005*\"atheist\" + 0.005*\"religion\" + 0.005*\"atheism\" + 0.004*\"rushdi\"')\n"
     ]
    }
   ],
   "source": [
    "def compare_topics(tm_metrics):\n",
    "    for _, row in tm_metrics.iterrows():\n",
    "        print('\\n{}:'.format(row.model))\n",
    "        print(\"\\n\".join(str(topic) for topic in row.topics))\n",
    "        \n",
    "compare_topics(tm_metrics)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Subjectively, Gensim and Sklearn NMFs are on par with each other, LDA and LSI look a bit worse."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 4. NMF on English Wikipedia"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This section shows how to train an NMF model on a large text corpus, the entire English Wikipedia: **2.6 billion words, in 23.1 million article sections across 5 million Wikipedia articles**.\n",
    "\n",
    "The data preprocessing takes a while, and we'll be comparing multiple models, so **reserve about 3 hours** and some **20 GB of disk space** to go through the following notebook cells in full. You'll need `gensim>=3.7.1`, `numpy`, `tqdm`, `pandas`, `psutils`, `joblib` and `sklearn`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Re-import modules from scratch, so that this Section doesn't rely on any previous cells.\n",
    "import itertools\n",
    "import json\n",
    "import logging\n",
    "import time\n",
    "import os\n",
    "\n",
    "from smart_open import smart_open\n",
    "import psutil\n",
    "import numpy as np\n",
    "import scipy.sparse\n",
    "from contextlib import contextmanager, contextmanager, contextmanager\n",
    "from multiprocessing import Process\n",
    "from tqdm import tqdm, tqdm_notebook\n",
    "import joblib\n",
    "import pandas as pd\n",
    "from sklearn.decomposition.nmf import NMF as SklearnNmf\n",
    "\n",
    "import gensim.downloader\n",
    "from gensim import matutils\n",
    "from gensim.corpora import MmCorpus, Dictionary\n",
    "from gensim.models import LdaModel, LdaMulticore, CoherenceModel\n",
    "from gensim.models.nmf import Nmf as GensimNmf\n",
    "from gensim.utils import simple_preprocess\n",
    "\n",
    "tqdm.pandas()\n",
    "\n",
    "logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s', level=logging.INFO)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load the Wikipedia dump"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We'll use the [gensim.downloader](https://github.com/RaRe-Technologies/gensim-data) to download a parsed Wikipedia dump (6.1 GB disk space):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = gensim.downloader.load(\"wiki-english-20171001\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Print the titles and sections of the first Wikipedia article, as a little sanity check:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Article: 'Anarchism'\n",
      "\n",
      "Section title: 'Introduction'\n",
      "Section text: '''Anarchism''' is a political philosophy that advocates self-governed societies based on volun…\n",
      "\n",
      "Section title: 'Etymology and terminology'\n",
      "Section text: The word ''anarchism'' is composed from the word ''anarchy'' and the suffix ''-ism'', themselves d…\n",
      "\n",
      "Section title: 'History'\n",
      "Section text: ===Origins=== Woodcut from a Diggers document by William Everard  The earliest anarchist themes ca…\n",
      "\n",
      "Section title: 'Anarchist schools of thought'\n",
      "Section text: Portrait of philosopher Pierre-Joseph Proudhon (1809–1865) by Gustave Courbet. Proudhon was the pri…\n",
      "\n",
      "Section title: 'Internal issues and debates'\n",
      "Section text: consistent with anarchist values is a controversial subject among anarchists.  Anarchism is a philo…\n",
      "\n",
      "Section title: 'Topics of interest'\n",
      "Section text: Intersecting and overlapping between various schools of thought, certain topics of interest and inte…\n",
      "\n",
      "Section title: 'Criticisms'\n",
      "Section text: Criticisms of anarchism include moral criticisms and pragmatic criticisms. Anarchism is often evalu…\n",
      "\n",
      "Section title: 'See also'\n",
      "Section text: * Anarchism by country…\n",
      "\n",
      "Section title: 'References'\n",
      "Section text: …\n",
      "\n",
      "Section title: 'Further reading'\n",
      "Section text: * Barclay, Harold, ''People Without Government: An Anthropology of Anarchy'' (2nd ed.), Left Bank Bo…\n",
      "\n",
      "Section title: 'External links'\n",
      "Section text: *  *…\n",
      "\n"
     ]
    }
   ],
   "source": [
    "data = gensim.downloader.load(\"wiki-english-20171001\")\n",
    "article = next(iter(data))\n",
    "\n",
    "print(\"Article: %r\\n\" % article['title'])\n",
    "for section_title, section_text in zip(article['section_titles'], article['section_texts']):\n",
    "    print(\"Section title: %r\" % section_title)\n",
    "    print(\"Section text: %s…\\n\" % section_text[:100].replace('\\n', ' ').strip())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's create a Python generator function that streams through the downloaded Wikipedia dump and preprocesses (tokenizes, lower-cases) each article:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "def wikidump2tokens(articles):\n",
    "    \"\"\"Stream through the Wikipedia dump, yielding a list of tokens for each article.\"\"\"\n",
    "    for article in articles:\n",
    "        article_section_texts = [\n",
    "            \" \".join([title, text])\n",
    "            for title, text\n",
    "            in zip(article['section_titles'], article['section_texts'])\n",
    "        ]\n",
    "        article_tokens = simple_preprocess(\" \".join(article_section_texts))\n",
    "        yield article_tokens"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Create a word-to-id mapping, in order to vectorize texts. Makes a full pass over the Wikipedia corpus, takes **~3.5 hours**:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2019-05-06 15:57:45,000 : INFO : loading Dictionary object from wiki.dict\n",
      "2019-05-06 15:57:45,031 : INFO : loaded wiki.dict\n"
     ]
    }
   ],
   "source": [
    "if os.path.exists('wiki.dict'):\n",
    "    # If we already stored the Dictionary in a previous run, simply load it, to save time.\n",
    "    dictionary = Dictionary.load('wiki.dict')\n",
    "else:\n",
    "    dictionary = Dictionary(wikidump2tokens(data))\n",
    "    # Keep only the 30,000 most frequent vocabulary terms, after filtering away terms\n",
    "    # that are too frequent/too infrequent.\n",
    "    dictionary.filter_extremes(no_below=5, no_above=0.5, keep_n=30000)\n",
    "    dictionary.save('wiki.dict')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Store preprocessed Wikipedia as bag-of-words sparse matrix in MatrixMarket format"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "When training NMF with a single pass over the input corpus (\"online\"), we simply vectorize each raw text straight from the input storage:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "vector_stream = (dictionary.doc2bow(article) for article in wikidump2tokens(data))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For the purposes of this tutorial though, we'll serialize (\"cache\") the vectorized bag-of-words vectors to disk, to `wiki.mm` file in MatrixMarket format. The reason is, we'll be re-using the vectorized articles multiple times, for different models for our benchmarks, and also shuffling them, so it makes sense to amortize the vectorization time by persisting the resulting vectors to disk."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So, let's stream through the preprocessed sparse Wikipedia bag-of-words matrix while storing it to disk. **This step takes about 3 hours** and needs **38 GB of disk space**:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "class RandomSplitCorpus(MmCorpus):\n",
    "    \"\"\"\n",
    "    Use the fact that MmCorpus supports random indexing, and create a streamed\n",
    "    corpus in shuffled order, including a train/test split for evaluation.\n",
    "    \"\"\"\n",
    "    def __init__(self, random_seed=42, testset=False, testsize=1000, *args, **kwargs):\n",
    "        super().__init__(*args, **kwargs)\n",
    "\n",
    "        random_state = np.random.RandomState(random_seed)\n",
    "        \n",
    "        self.indices = random_state.permutation(range(self.num_docs))\n",
    "        test_nnz = sum(len(self[doc_idx]) for doc_idx in self.indices[:testsize])\n",
    "        \n",
    "        if testset:\n",
    "            self.indices = self.indices[:testsize]\n",
    "            self.num_docs = testsize\n",
    "            self.num_nnz = test_nnz\n",
    "        else:\n",
    "            self.indices = self.indices[testsize:]\n",
    "            self.num_docs -= testsize\n",
    "            self.num_nnz -= test_nnz\n",
    "\n",
    "    def __iter__(self):\n",
    "        for doc_id in self.indices:\n",
    "            yield self[doc_id]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "if not os.path.exists('wiki.mm'):\n",
    "    MmCorpus.serialize('wiki.mm', vector_stream, progress_cnt=100000)\n",
    "\n",
    "if not os.path.exists('wiki_tfidf.mm'):\n",
    "    MmCorpus.serialize('wiki_tfidf.mm', tfidf[MmCorpus('wiki.mm')], progress_cnt=100000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2019-05-06 15:57:45,584 : INFO : loaded corpus index from wiki.mm.index\n",
      "2019-05-06 15:57:45,585 : INFO : initializing cython corpus reader from wiki.mm\n",
      "2019-05-06 15:57:45,586 : INFO : accepted corpus with 4924894 documents, 30000 features, 820242695 non-zero entries\n",
      "2019-05-06 15:57:49,102 : INFO : loaded corpus index from wiki.mm.index\n",
      "2019-05-06 15:57:49,103 : INFO : initializing cython corpus reader from wiki.mm\n",
      "2019-05-06 15:57:49,103 : INFO : accepted corpus with 4924894 documents, 30000 features, 820242695 non-zero entries\n",
      "2019-05-06 15:57:51,552 : INFO : loaded corpus index from wiki_tfidf.mm.index\n",
      "2019-05-06 15:57:51,553 : INFO : initializing cython corpus reader from wiki_tfidf.mm\n",
      "2019-05-06 15:57:51,554 : INFO : accepted corpus with 4924661 documents, 30000 features, 820007548 non-zero entries\n",
      "2019-05-06 15:57:55,680 : INFO : loaded corpus index from wiki_tfidf.mm.index\n",
      "2019-05-06 15:57:55,681 : INFO : initializing cython corpus reader from wiki_tfidf.mm\n",
      "2019-05-06 15:57:55,682 : INFO : accepted corpus with 4924661 documents, 30000 features, 820007548 non-zero entries\n"
     ]
    }
   ],
   "source": [
    "# Load back the vectors as two lazily-streamed train/test iterables.\n",
    "train_corpus = RandomSplitCorpus(\n",
    "    random_seed=42, testset=False, testsize=10000, fname='wiki.mm',\n",
    ")\n",
    "test_corpus = RandomSplitCorpus(\n",
    "    random_seed=42, testset=True, testsize=10000, fname='wiki.mm',\n",
    ")\n",
    "\n",
    "train_corpus_tfidf = RandomSplitCorpus(\n",
    "    random_seed=42, testset=False, testsize=10000, fname='wiki_tfidf.mm',\n",
    ")\n",
    "test_corpus_tfidf = RandomSplitCorpus(\n",
    "    random_seed=42, testset=True, testsize=10000, fname='wiki_tfidf.mm',\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Save preprocessed Wikipedia in scipy.sparse format"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This is only needed to run the Sklearn NMF on Wikipedia, for comparison in the benchmarks below. Sklearn expects in-memory scipy sparse input, not on-the-fly vector streams. Needs additional ~2 GB of disk space.\n",
    "\n",
    "\n",
    "**Skip this step if you don't need the Sklearn's NMF benchmark, and only want to run Gensim's NMF.**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "if not os.path.exists('wiki_train_csr.npz'):\n",
    "    scipy.sparse.save_npz(\n",
    "        'wiki_train_csr.npz',\n",
    "        matutils.corpus2csc(train_corpus_tfidf, len(dictionary)).T,\n",
    "    )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Metrics\n",
    "\n",
    "We'll track these metrics as we train and test NMF on the Wikipedia corpus we created above:\n",
    "- `train time` - time to train a model\n",
    "- `mean_ram` - mean RAM consumption during training\n",
    "- `max_ram` - maximum RAM consumption during training\n",
    "- `train time` - time to train a model.\n",
    "- `coherence` - coherence score (larger is better).\n",
    "- `l2_norm` - L2 norm of `v - Wh` (less is better, not defined for LDA)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Define a dataframe in which we'll store the recorded metrics:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "tm_metrics = pd.DataFrame(columns=[\n",
    "    'model', 'train_time', 'mean_ram', 'max_ram', 'coherence', 'l2_norm', 'topics',\n",
    "])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Define common parameters, to be shared by all evaluated models:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "params = dict(\n",
    "    chunksize=2000,\n",
    "    num_topics=50,\n",
    "    id2word=dictionary,\n",
    "    passes=1,\n",
    "    eval_every=10,\n",
    "    minimum_probability=0,\n",
    "    random_state=42,\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train Gensim NMF model and record its metrics"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Wikipedia training"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train Gensim NMF model and record its metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "row = {}\n",
    "row['model'] = 'gensim_nmf'\n",
    "row['train_time'], row['mean_ram'], row['max_ram'], nmf = get_train_time_and_ram(\n",
    "    lambda: GensimNmf(normalize=False, corpus=train_corpus_tfidf, **params),\n",
    "    'gensim_nmf',\n",
    "    1,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2019-05-06 16:23:44,657 : INFO : saving Nmf object under gensim_nmf.model, separately None\n",
      "2019-05-06 16:23:44,767 : INFO : saved gensim_nmf.model\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'max_ram': '774 MB', 'train_time': Timedelta('0 days 00:25:46.617000'), 'model': 'gensim_nmf', 'mean_ram': '771 MB'}\n"
     ]
    }
   ],
   "source": [
    "print(row)\n",
    "nmf.save('gensim_nmf.model')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2019-05-06 16:23:44,772 : INFO : loading Nmf object from gensim_nmf.model\n",
      "2019-05-06 16:23:44,871 : INFO : loading id2word recursively from gensim_nmf.model.id2word.* with mmap=None\n",
      "2019-05-06 16:23:44,872 : INFO : loaded gensim_nmf.model\n",
      "2019-05-06 16:24:47,126 : INFO : CorpusAccumulator accumulated stats from 1000 documents\n",
      "2019-05-06 16:24:47,272 : INFO : CorpusAccumulator accumulated stats from 2000 documents\n",
      "2019-05-06 16:24:47,424 : INFO : CorpusAccumulator accumulated stats from 3000 documents\n",
      "2019-05-06 16:24:47,573 : INFO : CorpusAccumulator accumulated stats from 4000 documents\n",
      "2019-05-06 16:24:47,726 : INFO : CorpusAccumulator accumulated stats from 5000 documents\n",
      "2019-05-06 16:24:47,880 : INFO : CorpusAccumulator accumulated stats from 6000 documents\n",
      "2019-05-06 16:24:48,027 : INFO : CorpusAccumulator accumulated stats from 7000 documents\n",
      "2019-05-06 16:24:48,168 : INFO : CorpusAccumulator accumulated stats from 8000 documents\n",
      "2019-05-06 16:24:48,319 : INFO : CorpusAccumulator accumulated stats from 9000 documents\n",
      "2019-05-06 16:24:48,472 : INFO : CorpusAccumulator accumulated stats from 10000 documents\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'mean_ram': '771 MB', 'topics': [(21, '0.009*\"his\" + 0.005*\"that\" + 0.005*\"him\" + 0.004*\"had\" + 0.003*\"they\" + 0.003*\"who\" + 0.003*\"her\" + 0.003*\"but\" + 0.003*\"king\" + 0.003*\"were\"'), (39, '0.005*\"are\" + 0.005*\"or\" + 0.004*\"be\" + 0.004*\"that\" + 0.003*\"can\" + 0.003*\"used\" + 0.003*\"this\" + 0.002*\"have\" + 0.002*\"such\" + 0.002*\"which\"'), (45, '0.092*\"apelor\" + 0.087*\"bucurești\" + 0.050*\"river\" + 0.046*\"cadastrul\" + 0.046*\"hidrologie\" + 0.046*\"meteorologie\" + 0.046*\"institutul\" + 0.046*\"române\" + 0.045*\"româniei\" + 0.045*\"rîurile\"'), (28, '0.066*\"gmina\" + 0.065*\"poland\" + 0.065*\"voivodeship\" + 0.046*\"village\" + 0.045*\"administrative\" + 0.042*\"lies\" + 0.037*\"approximately\" + 0.036*\"east\" + 0.031*\"west\" + 0.030*\"county\"'), (34, '0.087*\"romanized\" + 0.084*\"iran\" + 0.067*\"rural\" + 0.067*\"province\" + 0.066*\"census\" + 0.060*\"families\" + 0.055*\"village\" + 0.049*\"county\" + 0.047*\"population\" + 0.043*\"district\"')], 'model': 'gensim_nmf', 'coherence': -2.1071, 'l2_norm': 94.9686, 'train_time': Timedelta('0 days 00:25:46.617000'), 'max_ram': '774 MB', 'f1': None}\n"
     ]
    }
   ],
   "source": [
    "nmf = GensimNmf.load('gensim_nmf.model')\n",
    "row.update(get_metrics(nmf, test_corpus_tfidf))\n",
    "print(row)\n",
    "tm_metrics = tm_metrics.append(pd.Series(row), ignore_index=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train Gensim LSI model and record its metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "row = {}\n",
    "row['model'] = 'lsi'\n",
    "row['train_time'], row['mean_ram'], row['max_ram'], lsi = get_train_time_and_ram(\n",
    "    lambda: LsiModel(\n",
    "        corpus=train_corpus_tfidf,\n",
    "        chunksize=2000,\n",
    "        num_topics=50,\n",
    "        id2word=dictionary,\n",
    "    ),\n",
    "    'lsi',\n",
    "    1,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2019-05-06 17:34:51,929 : INFO : saving Projection object under lsi.model.projection, separately None\n",
      "2019-05-06 17:34:52,010 : INFO : saved lsi.model.projection\n",
      "2019-05-06 17:34:52,011 : INFO : saving LsiModel object under lsi.model, separately None\n",
      "2019-05-06 17:34:52,012 : INFO : not storing attribute projection\n",
      "2019-05-06 17:34:52,012 : INFO : not storing attribute dispatcher\n",
      "2019-05-06 17:34:52,022 : INFO : saved lsi.model\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'max_ram': '896 MB', 'train_time': Timedelta('0 days 01:10:03.206000'), 'model': 'lsi', 'mean_ram': '882 MB'}\n"
     ]
    }
   ],
   "source": [
    "print(row)\n",
    "lsi.save('lsi.model')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2019-05-06 17:34:52,027 : INFO : loading LsiModel object from lsi.model\n",
      "2019-05-06 17:34:52,039 : INFO : loading id2word recursively from lsi.model.id2word.* with mmap=None\n",
      "2019-05-06 17:34:52,039 : INFO : setting ignored attribute projection to None\n",
      "2019-05-06 17:34:52,040 : INFO : setting ignored attribute dispatcher to None\n",
      "2019-05-06 17:34:52,040 : INFO : loaded lsi.model\n",
      "2019-05-06 17:34:52,041 : INFO : loading LsiModel object from lsi.model.projection\n",
      "2019-05-06 17:34:52,115 : INFO : loaded lsi.model.projection\n",
      "2019-05-06 17:35:03,515 : INFO : CorpusAccumulator accumulated stats from 1000 documents\n",
      "2019-05-06 17:35:03,650 : INFO : CorpusAccumulator accumulated stats from 2000 documents\n",
      "2019-05-06 17:35:03,791 : INFO : CorpusAccumulator accumulated stats from 3000 documents\n",
      "2019-05-06 17:35:03,929 : INFO : CorpusAccumulator accumulated stats from 4000 documents\n",
      "2019-05-06 17:35:04,071 : INFO : CorpusAccumulator accumulated stats from 5000 documents\n",
      "2019-05-06 17:35:04,211 : INFO : CorpusAccumulator accumulated stats from 6000 documents\n",
      "2019-05-06 17:35:04,347 : INFO : CorpusAccumulator accumulated stats from 7000 documents\n",
      "2019-05-06 17:35:04,478 : INFO : CorpusAccumulator accumulated stats from 8000 documents\n",
      "2019-05-06 17:35:04,622 : INFO : CorpusAccumulator accumulated stats from 9000 documents\n",
      "2019-05-06 17:35:04,758 : INFO : CorpusAccumulator accumulated stats from 10000 documents\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'mean_ram': '882 MB', 'topics': [(0, '0.260*\"he\" + 0.172*\"his\" + 0.131*\"she\" + 0.111*\"her\" + 0.105*\"that\" + 0.094*\"district\" + 0.092*\"were\" + 0.087*\"school\" + 0.077*\"had\" + 0.077*\"film\"'), (1, '-0.430*\"district\" + -0.308*\"village\" + -0.270*\"population\" + -0.264*\"census\" + -0.263*\"romanized\" + -0.255*\"iran\" + -0.247*\"rural\" + -0.232*\"county\" + -0.221*\"province\" + -0.204*\"families\"'), (2, '-0.287*\"league\" + -0.275*\"he\" + -0.215*\"football\" + 0.212*\"album\" + -0.183*\"season\" + -0.180*\"team\" + -0.169*\"club\" + -0.161*\"cup\" + -0.136*\"played\" + 0.128*\"song\"'), (3, '-0.216*\"poland\" + -0.214*\"gmina\" + 0.213*\"romanized\" + -0.213*\"voivodeship\" + 0.209*\"iran\" + 0.208*\"album\" + 0.180*\"rural\" + -0.156*\"administrative\" + -0.155*\"east\" + -0.154*\"lies\"'), (4, '-0.260*\"album\" + -0.252*\"gmina\" + -0.252*\"poland\" + -0.250*\"voivodeship\" + -0.171*\"administrative\" + -0.161*\"lies\" + -0.152*\"song\" + -0.136*\"chart\" + -0.135*\"approximately\" + -0.129*\"village\"')], 'model': 'lsi', 'coherence': -3.9198, 'l2_norm': 94.5983, 'train_time': Timedelta('0 days 01:10:03.206000'), 'max_ram': '896 MB', 'f1': None}\n"
     ]
    }
   ],
   "source": [
    "lsi = LsiModel.load('lsi.model')\n",
    "row.update(get_metrics(lsi, test_corpus_tfidf))\n",
    "print(row)\n",
    "tm_metrics = tm_metrics.append(pd.Series(row), ignore_index=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train Gensim LDA and record its metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "row = {}\n",
    "row['model'] = 'lda'\n",
    "row['train_time'], row['mean_ram'], row['max_ram'], lda = get_train_time_and_ram(\n",
    "    lambda: LdaModel(corpus=train_corpus, **params),\n",
    "    'lda',\n",
    "    1,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2019-05-06 18:56:04,893 : INFO : saving LdaState object under lda.model.state, separately None\n",
      "2019-05-06 18:56:04,956 : INFO : saved lda.model.state\n",
      "2019-05-06 18:56:04,977 : INFO : saving LdaModel object under lda.model, separately ['expElogbeta', 'sstats']\n",
      "2019-05-06 18:56:04,979 : INFO : not storing attribute state\n",
      "2019-05-06 18:56:04,980 : INFO : not storing attribute id2word\n",
      "2019-05-06 18:56:04,981 : INFO : storing np array 'expElogbeta' to lda.model.expElogbeta.npy\n",
      "2019-05-06 18:56:04,992 : INFO : not storing attribute dispatcher\n",
      "2019-05-06 18:56:04,995 : INFO : saved lda.model\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'max_ram': '864 MB', 'train_time': Timedelta('0 days 01:20:59.941000'), 'model': 'lda', 'mean_ram': '862 MB'}\n"
     ]
    }
   ],
   "source": [
    "print(row)\n",
    "lda.save('lda.model')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2019-05-06 18:56:05,002 : INFO : loading LdaModel object from lda.model\n",
      "2019-05-06 18:56:05,005 : INFO : loading expElogbeta from lda.model.expElogbeta.npy with mmap=None\n",
      "2019-05-06 18:56:05,007 : INFO : setting ignored attribute state to None\n",
      "2019-05-06 18:56:05,008 : INFO : setting ignored attribute id2word to None\n",
      "2019-05-06 18:56:05,008 : INFO : setting ignored attribute dispatcher to None\n",
      "2019-05-06 18:56:05,009 : INFO : loaded lda.model\n",
      "2019-05-06 18:56:05,009 : INFO : loading LdaState object from lda.model.state\n",
      "2019-05-06 18:56:05,039 : INFO : loaded lda.model.state\n",
      "2019-05-06 18:56:20,341 : INFO : CorpusAccumulator accumulated stats from 1000 documents\n",
      "2019-05-06 18:56:20,466 : INFO : CorpusAccumulator accumulated stats from 2000 documents\n",
      "2019-05-06 18:56:20,595 : INFO : CorpusAccumulator accumulated stats from 3000 documents\n",
      "2019-05-06 18:56:20,715 : INFO : CorpusAccumulator accumulated stats from 4000 documents\n",
      "2019-05-06 18:56:20,848 : INFO : CorpusAccumulator accumulated stats from 5000 documents\n",
      "2019-05-06 18:56:20,973 : INFO : CorpusAccumulator accumulated stats from 6000 documents\n",
      "2019-05-06 18:56:21,100 : INFO : CorpusAccumulator accumulated stats from 7000 documents\n",
      "2019-05-06 18:56:21,226 : INFO : CorpusAccumulator accumulated stats from 8000 documents\n",
      "2019-05-06 18:56:21,351 : INFO : CorpusAccumulator accumulated stats from 9000 documents\n",
      "2019-05-06 18:56:21,483 : INFO : CorpusAccumulator accumulated stats from 10000 documents\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'mean_ram': '862 MB', 'topics': [(11, '0.066*\"de\" + 0.034*\"art\" + 0.030*\"french\" + 0.028*\"la\" + 0.022*\"france\" + 0.019*\"paris\" + 0.017*\"le\" + 0.016*\"museum\" + 0.013*\"van\" + 0.013*\"saint\"'), (45, '0.033*\"new\" + 0.027*\"states\" + 0.025*\"united\" + 0.023*\"york\" + 0.023*\"american\" + 0.023*\"county\" + 0.021*\"state\" + 0.017*\"city\" + 0.014*\"california\" + 0.012*\"washington\"'), (40, '0.028*\"radio\" + 0.025*\"show\" + 0.021*\"tv\" + 0.020*\"television\" + 0.016*\"news\" + 0.015*\"station\" + 0.014*\"channel\" + 0.012*\"fm\" + 0.012*\"network\" + 0.011*\"media\"'), (28, '0.064*\"university\" + 0.018*\"research\" + 0.015*\"college\" + 0.014*\"institute\" + 0.013*\"science\" + 0.011*\"professor\" + 0.010*\"has\" + 0.010*\"international\" + 0.009*\"national\" + 0.009*\"society\"'), (20, '0.179*\"he\" + 0.123*\"his\" + 0.015*\"born\" + 0.014*\"after\" + 0.013*\"him\" + 0.011*\"who\" + 0.011*\"career\" + 0.010*\"had\" + 0.010*\"later\" + 0.009*\"where\"')], 'model': 'lda', 'coherence': -1.7641, 'l2_norm': None, 'train_time': Timedelta('0 days 01:20:59.941000'), 'max_ram': '864 MB', 'f1': None}\n"
     ]
    }
   ],
   "source": [
    "lda = LdaModel.load('lda.model')\n",
    "row.update(get_metrics(lda, test_corpus))\n",
    "print(row)\n",
    "tm_metrics = tm_metrics.append(pd.Series(row), ignore_index=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train Sklearn NMF and record its metrics"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Careful!** Sklearn loads the entire input Wikipedia matrix into RAM. Even though the matrix is sparse, **you'll need FIXME GB of free RAM to run the cell below**."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'max_ram': '18177 MB', 'train_time': Timedelta('0 days 00:47:41.026000'), 'model': 'sklearn_nmf', 'mean_ram': '12872 MB'}\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "['sklearn_nmf.joblib']"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "row = {}\n",
    "row['model'] = 'sklearn_nmf'\n",
    "sklearn_nmf = SklearnNmf(n_components=50, tol=1e-2, random_state=42)\n",
    "row['train_time'], row['mean_ram'], row['max_ram'], sklearn_nmf = get_train_time_and_ram(\n",
    "    lambda: sklearn_nmf.fit(scipy.sparse.load_npz('wiki_train_csr.npz')),\n",
    "    'sklearn_nmf',\n",
    "    10,\n",
    ")\n",
    "print(row)\n",
    "joblib.dump(sklearn_nmf, 'sklearn_nmf.joblib')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2019-05-06 19:44:22,696 : INFO : CorpusAccumulator accumulated stats from 1000 documents\n",
      "2019-05-06 19:44:22,841 : INFO : CorpusAccumulator accumulated stats from 2000 documents\n",
      "2019-05-06 19:44:22,994 : INFO : CorpusAccumulator accumulated stats from 3000 documents\n",
      "2019-05-06 19:44:23,142 : INFO : CorpusAccumulator accumulated stats from 4000 documents\n",
      "2019-05-06 19:44:23,297 : INFO : CorpusAccumulator accumulated stats from 5000 documents\n",
      "2019-05-06 19:44:23,448 : INFO : CorpusAccumulator accumulated stats from 6000 documents\n",
      "2019-05-06 19:44:23,595 : INFO : CorpusAccumulator accumulated stats from 7000 documents\n",
      "2019-05-06 19:44:23,737 : INFO : CorpusAccumulator accumulated stats from 8000 documents\n",
      "2019-05-06 19:44:23,889 : INFO : CorpusAccumulator accumulated stats from 9000 documents\n",
      "2019-05-06 19:44:24,038 : INFO : CorpusAccumulator accumulated stats from 10000 documents\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'mean_ram': '12872 MB', 'topics': [(0, '0.067*\"gmina\" + 0.067*\"poland\" + 0.066*\"voivodeship\" + 0.047*\"administrative\" + 0.043*\"lies\" + 0.038*\"approximately\" + 0.037*\"east\" + 0.032*\"west\" + 0.031*\"county\" + 0.03*\"regional\"'), (1, '0.098*\"district\" + 0.077*\"romanized\" + 0.075*\"iran\" + 0.071*\"rural\" + 0.062*\"census\" + 0.061*\"province\" + 0.054*\"families\" + 0.048*\"population\" + 0.043*\"county\" + 0.031*\"village\"'), (2, '0.095*\"apelor\" + 0.09*\"bucurești\" + 0.047*\"cadastrul\" + 0.047*\"hidrologie\" + 0.047*\"meteorologie\" + 0.047*\"institutul\" + 0.047*\"române\" + 0.047*\"româniei\" + 0.047*\"rîurile\" + 0.046*\"river\"'), (3, '0.097*\"commune\" + 0.05*\"department\" + 0.045*\"communes\" + 0.03*\"insee\" + 0.029*\"france\" + 0.018*\"population\" + 0.015*\"saint\" + 0.014*\"region\" + 0.01*\"town\" + 0.01*\"french\"'), (4, '0.148*\"township\" + 0.05*\"county\" + 0.018*\"townships\" + 0.018*\"unincorporated\" + 0.015*\"community\" + 0.011*\"indiana\" + 0.01*\"census\" + 0.01*\"creek\" + 0.009*\"pennsylvania\" + 0.009*\"illinois\"')], 'model': 'sklearn_nmf', 'coherence': -2.0476, 'l2_norm': 94.8459, 'train_time': Timedelta('0 days 00:47:41.026000'), 'max_ram': '18177 MB', 'f1': None}\n"
     ]
    }
   ],
   "source": [
    "sklearn_nmf = joblib.load('sklearn_nmf.joblib')\n",
    "row.update(get_metrics(\n",
    "    sklearn_nmf, test_corpus_tfidf, dictionary=dictionary,\n",
    "))\n",
    "print(row)\n",
    "tm_metrics = tm_metrics.append(pd.Series(row), ignore_index=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Wikipedia results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model</th>\n",
       "      <th>train_time</th>\n",
       "      <th>mean_ram</th>\n",
       "      <th>max_ram</th>\n",
       "      <th>coherence</th>\n",
       "      <th>l2_norm</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>gensim_nmf</td>\n",
       "      <td>00:25:46.617000</td>\n",
       "      <td>771 MB</td>\n",
       "      <td>774 MB</td>\n",
       "      <td>-2.1071</td>\n",
       "      <td>94.9686</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>lsi</td>\n",
       "      <td>01:10:03.206000</td>\n",
       "      <td>882 MB</td>\n",
       "      <td>896 MB</td>\n",
       "      <td>-3.9198</td>\n",
       "      <td>94.5983</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>lda</td>\n",
       "      <td>01:20:59.941000</td>\n",
       "      <td>862 MB</td>\n",
       "      <td>864 MB</td>\n",
       "      <td>-1.7641</td>\n",
       "      <td>-</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>sklearn_nmf</td>\n",
       "      <td>00:47:41.026000</td>\n",
       "      <td>12872 MB</td>\n",
       "      <td>18177 MB</td>\n",
       "      <td>-2.0476</td>\n",
       "      <td>94.8459</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         model      train_time  mean_ram   max_ram  coherence  l2_norm\n",
       "0   gensim_nmf 00:25:46.617000    771 MB    774 MB    -2.1071  94.9686\n",
       "1          lsi 01:10:03.206000    882 MB    896 MB    -3.9198  94.5983\n",
       "2          lda 01:20:59.941000    862 MB    864 MB    -1.7641        -\n",
       "3  sklearn_nmf 00:47:41.026000  12872 MB  18177 MB    -2.0476  94.8459"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tm_metrics.replace(np.nan, '-', inplace=True)\n",
    "tm_metrics.drop(['topics', 'f1'], axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Insights\n",
    "\n",
    "Gensim's online NMF outperforms all other models in terms of speed and memory foorprint size.\n",
    "\n",
    "Compared to Sklearn's NMF:\n",
    "\n",
    "- **2x** faster.\n",
    "\n",
    "- Uses **~20x** less memory.\n",
    "\n",
    "    About **8GB** of Sklearn's RAM comes from the in-memory input matrices, which, in contrast to Gensim NMF, cannot be streamed iteratively. But even if we forget about the huge input size, Sklearn NMF uses about **2-8 GB** of RAM – significantly more than Gensim NMF or LDA.\n",
    "\n",
    "- L2 norm and coherence are a bit worse.\n",
    "\n",
    "Compared to Gensim's LSI:\n",
    "\n",
    "- **3x** faster\n",
    "- Better coherence but slightly worse l2 norm.\n",
    "\n",
    "Compared to Gensim's LDA, Gensim NMF also gives superior results:\n",
    "\n",
    "- **3x** faster\n",
    "- Coherence is worse than LDA's."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Learned Wikipedia topics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "gensim_nmf:\n",
      "(21, '0.009*\"his\" + 0.005*\"that\" + 0.005*\"him\" + 0.004*\"had\" + 0.003*\"they\" + 0.003*\"who\" + 0.003*\"her\" + 0.003*\"but\" + 0.003*\"king\" + 0.003*\"were\"')\n",
      "(39, '0.005*\"are\" + 0.005*\"or\" + 0.004*\"be\" + 0.004*\"that\" + 0.003*\"can\" + 0.003*\"used\" + 0.003*\"this\" + 0.002*\"have\" + 0.002*\"such\" + 0.002*\"which\"')\n",
      "(45, '0.092*\"apelor\" + 0.087*\"bucurești\" + 0.050*\"river\" + 0.046*\"cadastrul\" + 0.046*\"hidrologie\" + 0.046*\"meteorologie\" + 0.046*\"institutul\" + 0.046*\"române\" + 0.045*\"româniei\" + 0.045*\"rîurile\"')\n",
      "(28, '0.066*\"gmina\" + 0.065*\"poland\" + 0.065*\"voivodeship\" + 0.046*\"village\" + 0.045*\"administrative\" + 0.042*\"lies\" + 0.037*\"approximately\" + 0.036*\"east\" + 0.031*\"west\" + 0.030*\"county\"')\n",
      "(34, '0.087*\"romanized\" + 0.084*\"iran\" + 0.067*\"rural\" + 0.067*\"province\" + 0.066*\"census\" + 0.060*\"families\" + 0.055*\"village\" + 0.049*\"county\" + 0.047*\"population\" + 0.043*\"district\"')\n",
      "\n",
      "lsi:\n",
      "(0, '0.260*\"he\" + 0.172*\"his\" + 0.131*\"she\" + 0.111*\"her\" + 0.105*\"that\" + 0.094*\"district\" + 0.092*\"were\" + 0.087*\"school\" + 0.077*\"had\" + 0.077*\"film\"')\n",
      "(1, '-0.430*\"district\" + -0.308*\"village\" + -0.270*\"population\" + -0.264*\"census\" + -0.263*\"romanized\" + -0.255*\"iran\" + -0.247*\"rural\" + -0.232*\"county\" + -0.221*\"province\" + -0.204*\"families\"')\n",
      "(2, '-0.287*\"league\" + -0.275*\"he\" + -0.215*\"football\" + 0.212*\"album\" + -0.183*\"season\" + -0.180*\"team\" + -0.169*\"club\" + -0.161*\"cup\" + -0.136*\"played\" + 0.128*\"song\"')\n",
      "(3, '-0.216*\"poland\" + -0.214*\"gmina\" + 0.213*\"romanized\" + -0.213*\"voivodeship\" + 0.209*\"iran\" + 0.208*\"album\" + 0.180*\"rural\" + -0.156*\"administrative\" + -0.155*\"east\" + -0.154*\"lies\"')\n",
      "(4, '-0.260*\"album\" + -0.252*\"gmina\" + -0.252*\"poland\" + -0.250*\"voivodeship\" + -0.171*\"administrative\" + -0.161*\"lies\" + -0.152*\"song\" + -0.136*\"chart\" + -0.135*\"approximately\" + -0.129*\"village\"')\n",
      "\n",
      "lda:\n",
      "(11, '0.066*\"de\" + 0.034*\"art\" + 0.030*\"french\" + 0.028*\"la\" + 0.022*\"france\" + 0.019*\"paris\" + 0.017*\"le\" + 0.016*\"museum\" + 0.013*\"van\" + 0.013*\"saint\"')\n",
      "(45, '0.033*\"new\" + 0.027*\"states\" + 0.025*\"united\" + 0.023*\"york\" + 0.023*\"american\" + 0.023*\"county\" + 0.021*\"state\" + 0.017*\"city\" + 0.014*\"california\" + 0.012*\"washington\"')\n",
      "(40, '0.028*\"radio\" + 0.025*\"show\" + 0.021*\"tv\" + 0.020*\"television\" + 0.016*\"news\" + 0.015*\"station\" + 0.014*\"channel\" + 0.012*\"fm\" + 0.012*\"network\" + 0.011*\"media\"')\n",
      "(28, '0.064*\"university\" + 0.018*\"research\" + 0.015*\"college\" + 0.014*\"institute\" + 0.013*\"science\" + 0.011*\"professor\" + 0.010*\"has\" + 0.010*\"international\" + 0.009*\"national\" + 0.009*\"society\"')\n",
      "(20, '0.179*\"he\" + 0.123*\"his\" + 0.015*\"born\" + 0.014*\"after\" + 0.013*\"him\" + 0.011*\"who\" + 0.011*\"career\" + 0.010*\"had\" + 0.010*\"later\" + 0.009*\"where\"')\n",
      "\n",
      "sklearn_nmf:\n",
      "(0, '0.067*\"gmina\" + 0.067*\"poland\" + 0.066*\"voivodeship\" + 0.047*\"administrative\" + 0.043*\"lies\" + 0.038*\"approximately\" + 0.037*\"east\" + 0.032*\"west\" + 0.031*\"county\" + 0.03*\"regional\"')\n",
      "(1, '0.098*\"district\" + 0.077*\"romanized\" + 0.075*\"iran\" + 0.071*\"rural\" + 0.062*\"census\" + 0.061*\"province\" + 0.054*\"families\" + 0.048*\"population\" + 0.043*\"county\" + 0.031*\"village\"')\n",
      "(2, '0.095*\"apelor\" + 0.09*\"bucurești\" + 0.047*\"cadastrul\" + 0.047*\"hidrologie\" + 0.047*\"meteorologie\" + 0.047*\"institutul\" + 0.047*\"române\" + 0.047*\"româniei\" + 0.047*\"rîurile\" + 0.046*\"river\"')\n",
      "(3, '0.097*\"commune\" + 0.05*\"department\" + 0.045*\"communes\" + 0.03*\"insee\" + 0.029*\"france\" + 0.018*\"population\" + 0.015*\"saint\" + 0.014*\"region\" + 0.01*\"town\" + 0.01*\"french\"')\n",
      "(4, '0.148*\"township\" + 0.05*\"county\" + 0.018*\"townships\" + 0.018*\"unincorporated\" + 0.015*\"community\" + 0.011*\"indiana\" + 0.01*\"census\" + 0.01*\"creek\" + 0.009*\"pennsylvania\" + 0.009*\"illinois\"')\n"
     ]
    }
   ],
   "source": [
    "def compare_topics(tm_metrics):\n",
    "    for _, row in tm_metrics.iterrows():\n",
    "        print('\\n{}:'.format(row.model))\n",
    "        print(\"\\n\".join(str(topic) for topic in row.topics))\n",
    "        \n",
    "compare_topics(tm_metrics)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It seems all four models successfully learned useful topics from the Wikipedia corpus."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 5. And now for something completely different: Face decomposition from images"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The NMF algorithm in Gensim is optimized for extremely large (sparse) text corpora, but it will also work on vectors from other domains!\n",
    "\n",
    "Let's compare our model to other factorization algorithms on dense image vectors and check out the results.\n",
    "\n",
    "To do that we'll patch sklearn's [Faces Dataset Decomposition](https://scikit-learn.org/stable/auto_examples/decomposition/plot_faces_decomposition.html)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Sklearn wrapper\n",
    "Let's create an Scikit-learn wrapper in order to run Gensim NMF on images."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "lines_to_next_cell": 2
   },
   "outputs": [],
   "source": [
    "import logging\n",
    "import time\n",
    "\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "from numpy.random import RandomState\n",
    "from sklearn import decomposition\n",
    "from sklearn.cluster import MiniBatchKMeans\n",
    "from sklearn.datasets import fetch_olivetti_faces\n",
    "from sklearn.decomposition.nmf import NMF as SklearnNmf\n",
    "from sklearn.linear_model import LogisticRegressionCV\n",
    "from sklearn.metrics import f1_score\n",
    "from sklearn.model_selection import ParameterGrid\n",
    "\n",
    "import gensim.downloader\n",
    "from gensim import matutils\n",
    "from gensim.corpora import Dictionary\n",
    "from gensim.models import CoherenceModel, LdaModel, LdaMulticore\n",
    "from gensim.models.nmf import Nmf as GensimNmf\n",
    "from gensim.parsing.preprocessing import preprocess_string\n",
    "\n",
    "logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s', level=logging.INFO)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "lines_to_next_cell": 2
   },
   "outputs": [],
   "source": [
    "from sklearn.base import BaseEstimator, TransformerMixin\n",
    "import scipy.sparse as sparse\n",
    "\n",
    "\n",
    "class NmfWrapper(BaseEstimator, TransformerMixin):\n",
    "    def __init__(self, bow_matrix, **kwargs):\n",
    "        self.corpus = sparse.csc.csc_matrix(bow_matrix)\n",
    "        self.nmf = GensimNmf(**kwargs)\n",
    "\n",
    "    def fit(self, X):\n",
    "        self.nmf.update(self.corpus)\n",
    "\n",
    "    @property\n",
    "    def components_(self):\n",
    "        return self.nmf.get_topics()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Modified face decomposition notebook"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Adapted from the excellent [Scikit-learn tutorial](https://github.com/scikit-learn/scikit-learn/blob/master/examples/decomposition/plot_faces_decomposition.py) (BSD license):"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Turn off the logger due to large number of info messages during training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "gensim.models.nmf.logger.propagate = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "============================\n",
      "Faces dataset decompositions\n",
      "============================\n",
      "\n",
      "This example applies to :ref:`olivetti_faces` different unsupervised\n",
      "matrix decomposition (dimension reduction) methods from the module\n",
      ":py:mod:`sklearn.decomposition` (see the documentation chapter\n",
      ":ref:`decompositions`) .\n",
      "\n",
      "\n",
      "Dataset consists of 400 faces\n",
      "Extracting the top 6 Eigenfaces - PCA using randomized SVD...\n",
      "done in 0.025s\n",
      "Extracting the top 6 Non-negative components - NMF (Sklearn)...\n",
      "done in 0.171s\n",
      "Extracting the top 6 Non-negative components - NMF (Gensim)...\n",
      "done in 0.582s\n",
      "Extracting the top 6 Independent components - FastICA...\n",
      "done in 0.097s\n",
      "Extracting the top 6 Sparse comp. - MiniBatchSparsePCA...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/anotherbugmaster/.local/lib/python3.5/site-packages/sklearn/decomposition/sparse_pca.py:405: DeprecationWarning: normalize_components=False is a backward-compatible setting that implements a non-standard definition of sparse PCA. This compatibility mode will be removed in 0.22.\n",
      "  DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "done in 0.694s\n",
      "Extracting the top 6 MiniBatchDictionaryLearning...\n",
      "done in 0.479s\n",
      "Extracting the top 6 Cluster centers - MiniBatchKMeans...\n",
      "done in 0.087s\n",
      "Extracting the top 6 Factor Analysis components - FA...\n",
      "done in 0.048s\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/anotherbugmaster/.local/lib/python3.5/site-packages/sklearn/decomposition/factor_analysis.py:238: ConvergenceWarning: FactorAnalysis did not converge. You might want to increase the number of iterations.\n",
      "  ConvergenceWarning)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x325.44 with 6 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x325.44 with 6 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x325.44 with 6 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x325.44 with 6 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x325.44 with 6 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x325.44 with 6 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x325.44 with 6 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x325.44 with 6 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 144x162.72 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x325.44 with 6 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "\"\"\"\n",
    "============================\n",
    "Faces dataset decompositions\n",
    "============================\n",
    "\n",
    "This example applies to :ref:`olivetti_faces` different unsupervised\n",
    "matrix decomposition (dimension reduction) methods from the module\n",
    ":py:mod:`sklearn.decomposition` (see the documentation chapter\n",
    ":ref:`decompositions`) .\n",
    "\n",
    "\"\"\"\n",
    "print(__doc__)\n",
    "\n",
    "# Authors: Vlad Niculae, Alexandre Gramfort\n",
    "# License: BSD 3 claus\n",
    "\n",
    "n_row, n_col = 2, 3\n",
    "n_components = n_row * n_col\n",
    "image_shape = (64, 64)\n",
    "rng = RandomState(0)\n",
    "\n",
    "# #############################################################################\n",
    "# Load faces data\n",
    "dataset = fetch_olivetti_faces(shuffle=True, random_state=rng)\n",
    "faces = dataset.data\n",
    "\n",
    "n_samples, n_features = faces.shape\n",
    "\n",
    "# global centering\n",
    "faces_centered = faces - faces.mean(axis=0)\n",
    "\n",
    "# local centering\n",
    "faces_centered -= faces_centered.mean(axis=1).reshape(n_samples, -1)\n",
    "\n",
    "print(\"Dataset consists of %d faces\" % n_samples)\n",
    "\n",
    "\n",
    "def plot_gallery(title, images, n_col=n_col, n_row=n_row):\n",
    "    plt.figure(figsize=(2. * n_col, 2.26 * n_row))\n",
    "    plt.suptitle(title, size=16)\n",
    "    for i, comp in enumerate(images):\n",
    "        plt.subplot(n_row, n_col, i + 1)\n",
    "        vmax = max(comp.max(), -comp.min())\n",
    "        plt.imshow(comp.reshape(image_shape), cmap=plt.cm.gray,\n",
    "                   interpolation='nearest',\n",
    "                   vmin=-vmax, vmax=vmax)\n",
    "        plt.xticks(())\n",
    "        plt.yticks(())\n",
    "    plt.subplots_adjust(0.01, 0.05, 0.99, 0.93, 0.04, 0.)\n",
    "\n",
    "\n",
    "# #############################################################################\n",
    "# List of the different estimators, whether to center and transpose the\n",
    "# problem, and whether the transformer uses the clustering API.\n",
    "estimators = [\n",
    "    ('Eigenfaces - PCA using randomized SVD',\n",
    "     decomposition.PCA(n_components=n_components, svd_solver='randomized',\n",
    "                       whiten=True),\n",
    "     True),\n",
    "\n",
    "    ('Non-negative components - NMF (Sklearn)',\n",
    "     decomposition.NMF(n_components=n_components, init='nndsvda', tol=5e-3),\n",
    "     False),\n",
    "\n",
    "    ('Non-negative components - NMF (Gensim)',\n",
    "     NmfWrapper(\n",
    "         bow_matrix=faces.T,\n",
    "         chunksize=3,\n",
    "         eval_every=400,\n",
    "         passes=2,\n",
    "         id2word={idx: idx for idx in range(faces.shape[1])},\n",
    "         num_topics=n_components,\n",
    "         minimum_probability=0,\n",
    "         random_state=42,\n",
    "     ),\n",
    "     False),\n",
    "\n",
    "    ('Independent components - FastICA',\n",
    "     decomposition.FastICA(n_components=n_components, whiten=True),\n",
    "     True),\n",
    "\n",
    "    ('Sparse comp. - MiniBatchSparsePCA',\n",
    "     decomposition.MiniBatchSparsePCA(n_components=n_components, alpha=0.8,\n",
    "                                      n_iter=100, batch_size=3,\n",
    "                                      random_state=rng),\n",
    "     True),\n",
    "\n",
    "    ('MiniBatchDictionaryLearning',\n",
    "     decomposition.MiniBatchDictionaryLearning(n_components=15, alpha=0.1,\n",
    "                                               n_iter=50, batch_size=3,\n",
    "                                               random_state=rng),\n",
    "     True),\n",
    "\n",
    "    ('Cluster centers - MiniBatchKMeans',\n",
    "     MiniBatchKMeans(n_clusters=n_components, tol=1e-3, batch_size=20,\n",
    "                     max_iter=50, random_state=rng),\n",
    "     True),\n",
    "\n",
    "    ('Factor Analysis components - FA',\n",
    "     decomposition.FactorAnalysis(n_components=n_components, max_iter=2),\n",
    "     True),\n",
    "]\n",
    "\n",
    "# #############################################################################\n",
    "# Plot a sample of the input data\n",
    "\n",
    "plot_gallery(\"First centered Olivetti faces\", faces_centered[:n_components])\n",
    "\n",
    "# #############################################################################\n",
    "# Do the estimation and plot it\n",
    "\n",
    "for name, estimator, center in estimators:\n",
    "    print(\"Extracting the top %d %s...\" % (n_components, name))\n",
    "    t0 = time.time()\n",
    "    data = faces\n",
    "    if center:\n",
    "        data = faces_centered\n",
    "    estimator.fit(data)\n",
    "    train_time = (time.time() - t0)\n",
    "    print(\"done in %0.3fs\" % train_time)\n",
    "    if hasattr(estimator, 'cluster_centers_'):\n",
    "        components_ = estimator.cluster_centers_\n",
    "    else:\n",
    "        components_ = estimator.components_\n",
    "\n",
    "    # Plot an image representing the pixelwise variance provided by the\n",
    "    # estimator e.g its noise_variance_ attribute. The Eigenfaces estimator,\n",
    "    # via the PCA decomposition, also provides a scalar noise_variance_\n",
    "    # (the mean of pixelwise variance) that cannot be displayed as an image\n",
    "    # so we skip it.\n",
    "    if (hasattr(estimator, 'noise_variance_') and\n",
    "            estimator.noise_variance_.ndim > 0):  # Skip the Eigenfaces case\n",
    "        plot_gallery(\"Pixelwise variance\",\n",
    "                     estimator.noise_variance_.reshape(1, -1), n_col=1,\n",
    "                     n_row=1)\n",
    "    plot_gallery('%s - Train time %.1fs' % (name, train_time),\n",
    "                 components_[:n_components])\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As you can see, Gensim's NMF implementation is slower than Sklearn's on **dense** vectors, while achieving comparable quality."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Conclusion\n",
    "\n",
    "Gensim NMF is an extremely fast and memory-optimized model. Use it to obtain interpretable topics, as an alternative to SVD / LDA.\n",
    "\n",
    "---\n",
    "\n",
    "The NMF implementation in Gensim was created by [Timofey Yefimov](https://github.com/anotherbugmaster/) as a part of his [RARE Technologies Student Incubator](https://rare-technologies.com/incubator/) graduation project."
   ]
  }
 ],
 "metadata": {
  "jupytext": {
   "text_representation": {
    "extension": ".py",
    "format_name": "percent",
    "format_version": "1.1",
    "jupytext_version": "0.8.3"
   }
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}