Package_Recommendation_Engine_RNN.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Package Recommendation Engine - RNN.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": [
        "qVu3yEACC45f",
        "xcwbeBIKNLVk",
        "91Qswcs-Racr",
        "YLscgV61RK_L",
        "VSG-Oc6kZPuj",
        "4Wg_p5YxjEuM",
        "wzuV40b0Bdo1",
        "fMHGgz8GE4eC",
        "JUE1Wa8ESmh3"
      ],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "[View in Colaboratory](https://colab.research.google.com/github/connected-bsamadi/colab-notebooks/blob/master/Package_Recommendation_Engine_RNN.ipynb)"
      ]
    },
    {
      "metadata": {
        "id": "qVu3yEACC45f",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# Setting up"
      ]
    },
    {
      "metadata": {
        "id": "0XeiWi7llE07",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# Google Cloud authentication\n",
        "from google.colab import auth\n",
        "auth.authenticate_user()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "9qLqtNVSEYIl",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# Installing packages\n",
        "!pip install nltk > /dev/null\n",
        "!pip install rake-nltk > /dev/null\n",
        "!pip install just > /dev/null\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "cs_R3Keakxdj",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "c0a34024-5315-4ca7-af89-f97ae26b3f92"
      },
      "cell_type": "code",
      "source": [
        "# Download a list of stop words for RAKE\n",
        "!wget https://raw.githubusercontent.com/zelandiya/RAKE-tutorial/master/data/stoplists/SmartStoplist.txt > /dev/null"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Redirecting output to ‘wget-log.6’.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "F7IOueubz7XD",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "31884e97-eed7-40f5-8f2b-a3404bc4b03b"
      },
      "cell_type": "code",
      "source": [
        "from google.cloud import bigquery\n",
        "import json\n",
        "import re\n",
        "import pandas\n",
        "from rake_nltk import Rake\n",
        "import sys\n",
        "import just\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "from collections import Counter\n",
        "from keras.models import Sequential\n",
        "from keras.layers import Activation, Dense\n",
        "from keras.layers import LSTM\n",
        "from keras.optimizers import RMSprop"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "metadata": {
        "id": "eUJxhFqJM8Qf",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# Assigning the project name and creating the BigQuery client\n",
        "# https://cloud.google.com/resource-manager/docs/creating-managing-projects\n",
        "project_id = 'ontario-2018'\n",
        "client = bigquery.Client(project=project_id)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "xcwbeBIKNLVk",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# Data Query\n",
        "BigQuery is used to get the data from the github_repos dataset."
      ]
    },
    {
      "metadata": {
        "id": "t81Olv1jMWLf",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# Legacy SQL is used\n",
        "job_config = bigquery.job.QueryJobConfig()\n",
        "job_config.use_legacy_sql=True"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "91Qswcs-Racr",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "## Extracting packages"
      ]
    },
    {
      "metadata": {
        "id": "sFJRNRdPRgit",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "### Get packages from big query\n",
        "query = ('''\n",
        "SELECT sample_repo_name, content\n",
        "FROM [bigquery-public-data:github_repos.sample_contents] contents\n",
        "INNER JOIN [bigquery-public-data:github_repos.sample_files] files\n",
        "  ON contents.id = files.id\n",
        "INNER JOIN [bigquery-public-data:github_repos.sample_repos] repos\n",
        "  ON contents.sample_repo_name = repos.repo_name\n",
        "WHERE files.path = 'package.json' AND repos.watch_count > 1000\n",
        "''')\n",
        "query_job = client.query(query,job_config)\n",
        "package_results = query_job.to_dataframe().values"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "7EQsdb2kNGFG",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "### Transform package_results into package_array\n",
        "package_array = []\n",
        "package_reponame_array = []\n",
        "for package in package_results:\n",
        "    repo_obj = json.loads(package[1])\n",
        "    package_record = {}\n",
        "    package_record[\"reponame\"] = package[0]\n",
        "    package_reponame_array.append(package[0])\n",
        "    if \"name\" in repo_obj:\n",
        "      package_record[\"name\"] = repo_obj[\"name\"]\n",
        "    if \"devDependencies\" in repo_obj:\n",
        "      dev_dependencies = repo_obj[\"devDependencies\"]\n",
        "      if (len(dev_dependencies) > 0):\n",
        "        package_record[\"devDependencies\"] = dev_dependencies\n",
        "    if \"dependencies\" in repo_obj:\n",
        "      dependencies = repo_obj[\"dependencies\"]\n",
        "      if (len(dependencies) > 0):\n",
        "        package_record[\"dependencies\"] = dependencies\n",
        "    if \"keywords\" in repo_obj:\n",
        "      keywords = repo_obj[\"keywords\"]\n",
        "      if (len(keywords) > 0):\n",
        "        package_record[\"keywords\"] = keywords\n",
        "    package_array.append(package_record)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "YLscgV61RK_L",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "## Extracting Readme Files"
      ]
    },
    {
      "metadata": {
        "id": "23hOpMn0EOvJ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "20bddcdd-5a72-4d05-f5c8-ba0bb8aaae76"
      },
      "cell_type": "code",
      "source": [
        "### Get README.md files from big query\n",
        "\n",
        "job_config = bigquery.job.QueryJobConfig()\n",
        "job_config.use_legacy_sql=False\n",
        "query = ('''SELECT files.repo_name, content\n",
        "FROM `bigquery-public-data.github_repos.contents` contents\n",
        "INNER JOIN `bigquery-public-data.github_repos.files` files\n",
        "  ON contents.id = files.id\n",
        "WHERE files.path IN ('readme.md','Readme.md', 'README.md', 'README.MD') AND files.repo_name IN UNNEST(@repolist)\n",
        "''')\n",
        "\n",
        "query_params = [\n",
        "    bigquery.ArrayQueryParameter('repolist', 'STRING', package_reponame_array)\n",
        "]\n",
        "job_config.query_parameters = query_params\n",
        "query_job = client.query(query,job_config)\n",
        "readme_results = query_job.to_dataframe().values\n",
        "print(len(package_reponame_array))\n",
        "print(len(readme_results))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "25\n",
            "22\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "9NKXKxtuLNBL",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "### Transform readme_results into readme_array\n",
        "\n",
        "readme_array = []\n",
        "for readme_interim in readme_results:\n",
        "  readme_record = {}\n",
        "  readme_record[\"reponame\"] = readme_interim[0]\n",
        "  readme_record[\"content\"] = readme_interim[1]\n",
        "  readme_array.append(readme_record)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "VSG-Oc6kZPuj",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# Creating training data"
      ]
    },
    {
      "metadata": {
        "id": "xvOmdIICO-SR",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "### Load the content of SmartStoplist.txt into stop_words\n",
        "\n",
        "stop_words=[]\n",
        "with open(\"SmartStoplist.txt\", 'r') as f:\n",
        "    for i, x in enumerate(f):\n",
        "        if 1 <= i:\n",
        "            stop_words.append(x.rstrip())"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "vetcC0I-R5eh",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "### Extract keywords using Rake\n",
        "\n",
        "rake = Rake(stopwords=stop_words, max_length=1)\n",
        "concatenated_readme = ''\n",
        "for readme in readme_array:\n",
        "  concatenated_readme = concatenated_readme + readme[\"content\"]\n",
        "rake.extract_keywords_from_text(concatenated_readme)\n",
        "ranked_phrases = rake.get_ranked_phrases()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "aoYCyfUBccfq",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 71
        },
        "outputId": "a7264958-13ec-4fa7-8eb7-7d2d309902fd"
      },
      "cell_type": "code",
      "source": [
        "### Remove all non-english characters\n",
        "\n",
        "pattern = re.compile(\"[a-zA-Z0-9]\")\n",
        "input_nodes = list(filter(lambda x: bool(re.match(pattern, x)), ranked_phrases))\n",
        "input_nodes = list(filter(lambda x: len(x)>1, input_nodes))\n",
        "print(input_nodes)\n",
        "len(input_nodes)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "['zos', 'zip', 'zh_cn', 'zfy6ll7efa', 'zenparsing', 'zen', 'youtube', 'youtu', 'yielded', 'yieldables', 'yieldable', 'yield', 'yelouafi', 'yellowgreen', 'yellowapple', 'yellow', 'years', 'yarn', 'yandex', 'xojs', 'xo', 'xmlhttprequest', 'xcodeproj', 'xcode', 'wrong', 'written', 'writing', 'write', 'wrap', 'wpo', 'wow', 'worry', 'world', 'works', 'working', 'workarounds', 'workaround', 'work', 'won', 'withgoogle', 'windowscroller', 'windows', 'windowing', 'window', 'win', 'wikipedia', 'wiki', 'width', 'white', 'whatwg', 'weststrate', 'welcoming', 'weeks', 'webstorm', 'webstore', 'website', 'websemantics', 'webscr', 'webpagetest', 'webpack', 'web', 'wearejh', 'ways', 'watches', 'watch', 'wasn', 'warranty', 'warranties', 'warnings', 'warning', 'warn', 'wait', 'vue', 'vscode', 'voted', 'vortex', 'voorhoede', 'voila', 'visualization', 'visiting', 'visit', 'visionmedia', 'virtualized', 'virtualization', 'viewport', 'viewmodel', 'viewer', 'view', 'videos', 'vi', 'versions', 'version', 'verify', 'verbose', 'vendor', 'ved', 've', 'vars', 'variable', 'vampolo', 'values', 'validator', 'validated', 'val', 'vadimdemedes', 'v5', 'v4', 'v3', 'v2', 'v1', 'v0', 'uvk9zassm8q', 'utm_source', 'utm_medium', 'utm_content', 'utm_campaign', 'utilize', 'usp', 'usingautosizer', 'users', 'username', 'userid', 'user_fetch_requested', 'user', 'useable', 'usd', 'usage', 'urls', 'url', 'uploaded', 'upload', 'upgrade', 'updating', 'updates', 'updated', 'update', 'unzip', 'unprefixed', 'unprefix', 'unpkg', 'unopinionated', 'unobtrusiveness', 'unlink', 'unlike', 'units', 'unit', 'unique', 'uniform', 'unfinishedtodocount', 'unexpected', 'understanding', 'unconditionally', 'uncompressed', 'unavailable', 'unable', 'umd', 'ul', 'ui', 'ug32xx0o3l', 'typically', 'typescriptlang', 'typescript', 'types', 'typeerror', 'typed', 'type', 'ty', 'txt', 'twitter', 'twiliosms', 'twilioipm', 'tweet', 'tweak', 'tutorials', 'turning', 'turned', 'turn', 'tsconfig', 'truthy', 'true', 'trough', 'troubleshooting', 'triggers', 'trigger', 'tricks', 'treo', 'tree', 'treated', 'treasuredata', 'travis', 'transpiles', 'transpile', 'transpilation', 'translation', 'translate', 'transition', 'transformation', 'transform', 'tracked', 'track', 'trace', 'tower', 'touch', 'tos', 'tort', 'top', 'tootallnate', 'tooltips', 'tooltip', 'tools', 'toolkit', 'tooling', 'tool', 'tons', 'ton', 'tokei', 'toggles', 'toggled', 'toggle', 'todoview', 'todos', 'todolistview', 'todolist', 'todo', 'today', 'tmp', 'tldt', 'tldroid', 'tldr_viewer', 'tldr_elixir_client', 'tldr', 'tlcr', 'tl', 'tj', 'titles', 'title', 'tired', 'tiny', 'timing', 'timeseries', 'times', 'timer', 'timeouts', 'timeout', 'timeline', 'time', 'ticking', 'thunks', 'thunk', 'thumb', 'throwsasync', 'thrown', 'thrower', 'throw', 'throttling', 'threaded', 'thread', 'things', 'thing', 'themes', 'theme', 'thearegee', 'tether', 'tests', 'testing', 'test', 'terminology', 'terminal', 'templates', 'template', 'tempfile', 'temp', 'technically', 'technical', 'teams', 'team', 'tealdeer', 'tdlmbo_hzsbp8tmg', 'tc39', 'taylorhakes', 'tasks', 'targetting', 'target', 'tape', 'tap', 'taobao', 'talks', 'talking', 'talk', 'takes', 'takelatest', 'takeevery', 'tags', 'tagged', 'tag', 'tablewithnaturalsort', 'table', 'tabindex', 't7cujdu', 'systemjs', 'system', 'syntax', 'synchronous', 'sync', 'symbol', 'sylvainpolletvillard', 'swiper3', 'swiper2', 'swiper', 'svg', 'surprised', 'surfaced', 'surface', 'suppose', 'supports', 'supported', 'support', 'supplying', 'supply', 'superraytin', 'super', 'summary', 'suitable', 'suggest', 'success', 'subset', 'subsequent', 'subprocesses', 'submitted', 'submit', 'submission', 'sublime', 'sublicense', 'subject', 'subdomain', 'stylus', 'styles', 'stylelint', 'style', 'studio', 'structure', 'strong', 'strive', 'string', 'straightforward', 'straight', 'storing', 'stores', 'stored', 'store', 'storage', 'step', 'stdout', 'statustext', 'statuses', 'status', 'stats', 'statistics', 'statement', 'state', 'stashed', 'startup', 'starter', 'started', 'start', 'stars', 'star', 'standard', 'stand', 'stale', 'stage', 'stable', 'srsudar', 'src', 'squeeze', 'square', 'spying', 'spring', 'spreadsheets', 'spread', 'sponsors', 'sponsored', 'sponsor', 'spin', 'speedcurve', 'specification', 'special', 'spec', 'spawns', 'spacing', 'space', 'sources', 'sourcemaps', 'source', 'sotojuan', 'sorting', 'sorted', 'sortable', 'sonnylazuardi', 'somelibname', 'solve', 'solution', 'solo', 'software', 'social', 'snippets', 'snippetoptions', 'snippet', 'snapshots', 'snapshot', 'snap', 'smart', 'slowed', 'slow', 'slightly', 'slider', 'slashing', 'slack', 'skipping', 'skipped', 'skip', 'size', 'sites', 'site_request_forgery_', 'site', 'single', 'sindresorhus', 'simply', 'simplified', 'simplicity', 'simplest', 'simpler', 'simple', 'similarly', 'similar', 'signing', 'signal', 'sig', 'side', 'shown', 'show', 'shouldn', 'short', 'shoichikaji', 'shirts', 'ships', 'shipped', 'shields', 'sharpened', 'sharing', 'shared', 'share', 'shallowcompare', 'shallow', 'shakyshane', 'sh', 'sg', 'setup', 'settings', 'setting', 'settimeout', 'set', 'session', 'service', 'server', 'serial', 'sending', 'send', 'semver', 'semantically', 'selectortimings', 'selectors', 'selectorcounts', 'selector', 'selected', 'select', 'security', 'section', 'search', 'seal', 'scss', 'scrollsync', 'scripts', 'script', 'screenshots', 'screenshot', 'screencast', 'screen', 'scores', 'scientific', 'science', 'schematics', 'scenario', 'scan', 'scale', 'scalable', 'sboudrias', 'saves', 'saved', 'save', 'satisfied', 'sass', 'samverschueren', 'sagas', 'sagamiddleware', 'saga', 'safeeval', 'safe', 'safari_48x48', 'safari', 'sa', 'rycochet', 'rxjs', 'rusty', 'rust', 'runtime', 'runs', 'running', 'run', 'rules', 'rule', 'ruby', 'ru_ru', 'rs', 'root', 'room', 'rollup', 'robot', 'roadmap', 'rnpm', 'rmsportal', 'rilut', 'rights', 'rgba', 'review', 'reverselist', 'returns', 'returned', 'return', 'results', 'result', 'restriction', 'rest', 'responsibility', 'response', 'respond', 'resources', 'resource', 'resolves', 'resolve', 'resolution', 'resizable', 'res', 'requires', 'requirejs', 'required', 'require', 'requests', 'request', 'req', 'repro', 'repository', 'repos', 'reports', 'reporting', 'reporter', 'reported', 'report', 'repo', 'reply', 'replied', 'replaced', 'repeated', 'repeat', 'rendertimings', 'rendering', 'rendered', 'rendercounts', 'render', 'rems', 'removed', 'remove', 'remix', 'remind', 'remember', 'remains', 'relying', 'relies', 'reliable', 'releases', 'released', 'release', 'relations', 'related', 'reject', 'registry', 'register', 'regions', 'regex', 'regenerator', 'regenerated', 'refguide', 'referral', 'references', 'reference', 'refer', 'reduxsaga', 'redux', 'reduces', 'reducer', 'redis', 'redirection', 'redirect', 'redbubble', 'recursed', 'recordunmountevent', 'recordrenderstopevent', 'recordrenderstartevent', 'recordmountevent', 'recordit', 'recorder', 'recorded', 'recommended', 'recommend', 'recipes', 'recipe', 'receiving', 'received', 'receive', 'rebelmail', 'reassign', 'reason', 'realize', 'real', 'ready', 'reads', 'readme_ru', 'readme_ja', 'readme', 'reading', 'readfile', 'readable', 'read', 'reactive_programming', 'reactive', 'reactions', 'reaction', 'reacted', 'react', 'reached', 'rc', 'raylee', 'raw', 'ratio', 'ranges', 'rails', 'radius', 'rackt', 'quit', 'quickstart', 'quickest', 'quick', 'questions', 'queries', 'python', 'py', 'px', 'pwa', 'putting', 'put', 'pushing', 'push', 'purpose', 'pulls', 'pubsub', 'publish', 'publicly', 'public', 'pt_br', 'psibi', 'prs', 'proxy', 'proxies', 'providing', 'providers', 'providedtitle', 'provided', 'provide', 'props', 'proposals', 'proposal', 'property', 'properties', 'proper', 'proof', 'pronunciation', 'pronounced', 'prompted', 'promises', 'promise', 'projects', 'projectdir', 'project', 'progress', 'programming', 'programmer', 'programmatically', 'profiling', 'profile', 'production', 'product', 'producing', 'produce', 'processors', 'processes', 'processed', 'process', 'problem', 'prints', 'printing', 'printed', 'previous', 'prevented', 'prettier', 'preset', 'presented', 'presentation', 'present', 'prepared', 'prelink', 'prefixes', 'prefixer', 'prefixed', 'prefix', 'preferences', 'prefer', 'predictable', 'precompiling', 'precompiled', 'precedence', 'pre', 'pranavraja', 'practices', 'practice', 'pr', 'pp1', 'pp', 'powered', 'power', 'postlink', 'postinstall', 'postimg', 'postgres', 'postcssjs', 'postcss', 'position', 'port', 'porras', 'pops', 'poppers', 'popperjs', 'popper', 'popovers', 'polyfills', 'polyfill', 'points', 'point', 'png', 'pm', 'plugins', 'plugin', 'pluggable', 'plist', 'pld3jnfklds7wsehsal2cfwg0fex7a6aok', 'playlist', 'playground', 'platforms', 'plans', 'planning', 'plan', 'places', 'placements', 'placeholders', 'placeholder', 'place', 'pixel', 'pivotal', 'pitfalls', 'pipeline', 'pipe', 'pify', 'picture', 'picking', 'picker', 'pick', 'phrases', 'php', 'philosophy', 'pg', 'perl', 'performant', 'performance', 'perform', 'perfect', 'percentiles', 'percentage', 'pepa65', 'people', 'pending', 'peers', 'paypalobjects', 'paypal', 'payload', 'paused', 'paulirish', 'patterns', 'pattern', 'patreon', 'path', 'patch', 'paste', 'past', 'password', 'passing', 'passes', 'passed', 'pass', 'parts', 'part', 'parsing', 'parsejson', 'parse', 'params', 'parameters', 'parallelism', 'parallel', 'panel', 'pairs', 'pain', 'pages', 'pageres', 'page__nav', 'page__main', 'page__head', 'page__footer', 'page', 'padding', 'packagr', 'packages', 'packager', 'packageinstance', 'package', 'owasp', 'overriding', 'overrides', 'override', 'output', 'outdated', 'ots', 'ostera', 'osjsorg', 'os', 'originates', 'origin', 'org', 'order', 'orange', 'options', 'optionally', 'optional', 'option', 'opting', 'optimizer', 'optimized', 'optimization', 'opposite', 'opposed', 'opinionated', 'operations', 'operate', 'opera_48x48', 'opera', 'opensourcenorth', 'opened', 'open', 'opacity', 'onupdate', 'online', 'onepage', 'onenotedev', 'oncreate', 'onclick', 'on_spectrum', 'omit', 'older', 'official', 'offer', 'obvious', 'obtaining', 'observer', 'observables', 'observable', 'objects', 'object', 'obj', 'obama', 'nyvnqkrl69s', 'nyc', 'nutellinoit', 'number', 'null', 'nuget', 'npx', 'npmjs', 'npm', 'novemberborn', 'november', 'notthrowsasync', 'note', 'noryn6gotvo', 'normalized', 'normalize', 'normal', 'nope', 'nontrivial', 'nonthrower', 'noninfringement', 'nolimits4web', 'noise', 'nodejs', 'node_modules', 'node_env', 'node', 'nnmlddkpgoecicoallmimonoboialpap', 'nl', 'nilj0', 'nightlies', 'nice', 'ngtools', 'ng', 'news', 'neutrino', 'network', 'net', 'nestedtimeline', 'nested', 'neighborhood999', 'needed', 'ndmitry', 'nay', 'navigation', 'nav', 'natively', 'native', 'names', 'named', 'mz', 'mysaga', 'mweststrate', 'multigrid', 'multicolumnsorttable', 'mt', 'mskims', 'mscdex', 'mozilla', 'moz', 'moving', 'move', 'months', 'monorepo', 'mongodb', 'modules', 'module', 'modular', 'modifying', 'modify', 'modifiers', 'modifier', 'modified', 'moderate', 'models', 'model', 'mode', 'moddus', 'mocha', 'mobxjs', 'mobx4', 'mobx', 'mobservable', 'mobile', 'mo', 'mixed', 'mit', 'mistake', 'miss', 'mirrors', 'minus', 'minmax', 'minimize', 'minimal', 'minified', 'mini', 'mind', 'min', 'milliseconds', 'mikaeljorhult', 'migrating', 'middlewares', 'middleware', 'middleman', 'microsoftedge', 'microsoft', 'michel', 'mic', 'mflint', 'metrics', 'metric', 'methods', 'method', 'metadata', 'messages', 'message', 'mess', 'merged', 'merge', 'merchantability', 'menu', 'mentioned', 'mendix', 'members', 'melpa', 'meet', 'medium', 'medis', 'media', 'mechanisms', 'mechanism', 'measuring', 'meant', 'means', 'meaning', 'md', 'max', 'mathematical', 'math', 'matching', 'matches', 'matcher', 'matched', 'match', 'master', 'masonry', 'martians', 'margin', 'maplimit', 'map', 'manually', 'manner', 'manekinekko', 'management', 'managed', 'manage', 'man', 'making', 'makes', 'make', 'maintenance', 'maintain', 'main', 'magic', 'made', 'macros', 'macro', 'macos', 'machine', 'mac', 'm4a', 'lukehoban', 'luisrudge', 'luin', 'lts', 'ls', 'lpg', 'lower', 'low', 'love', 'lots', 'lot', 'lord63', 'loops', 'longer', 'long', 'logos', 'logornpm_final4', 'logo_readme', 'logo', 'logging', 'logged', 'log', 'lodin', 'lodash', 'location', 'locate', 'localization', 'loading', 'loader', 'load', 'll', 'listening', 'listeners', 'listen', 'list', 'linux', 'lint', 'links', 'linkingsucceeded', 'linking', 'linked', 'link', 'linear', 'line', 'limited', 'limitations', 'limitation', 'likewise', 'lighthouse', 'lifecycle', 'licenses', 'license', 'library', 'libraries', 'libhive', 'liberapay', 'lib', 'liable', 'liability', 'li', 'lh4u', 'leveraging', 'levels', 'level', 'letting', 'lessons', 'lengthy', 'leighmcculloch', 'leave', 'learned', 'learncode', 'learn', 'leak', 'lead', 'lckifpakia', 'lc', 'layer', 'launched', 'launch', 'latest', 'larger', 'large', 'lang', 'landscape', 'labs', 'labels', 'labeled', 'label', 'l9', 'l81', 'l7', 'kuanyui', 'korean', 'ko_kr', 'knockout', 'kirillseva', 'kinds', 'kindly', 'kind', 'keywords', 'keys', 'keyboard', 'key', 'kevva', 'kdpgvgumpgfqahpjicrk', 'kbd', 'k8dr8bmu7', 'justinribeiro', 'jspm', 'json', 'jsdom', 'jsdelivr', 'jsc', 'js', 'jquery', 'jpg', 'jpeg', 'joytocode', 'jonsuh', 'jonathantneal', 'join', 'joelbarbosa', 'jlhwung', 'jhz3aaqobk', 'jfmengels', 'jest', 'jenius', 'jekyll', 'jeffjewiss', 'javascriptcore', 'javascript', 'jamestalmage', 'ja_jp', 'j30k46l7xw', 'ixzpn44trr', 'ivdgtxspnwlojlhrpkf1nwezfl8', 'iterable', 'items', 'item_name', 'it_it', 'istanbul', 'issuestats', 'issues', 'issue', 'isolation', 'isolated', 'ish', 'iq0', 'ios', 'ioredis', 'io', 'involves', 'invoked', 'inverse', 'introduction', 'introduces', 'introduced', 'introduce', 'intro', 'intricacies', 'internally', 'interfaces', 'interface', 'interesting', 'interested', 'interacting', 'intentions', 'intentional', 'intended', 'integrations', 'integration', 'integrate', 'instructions', 'instanceof', 'instance', 'installing', 'installed', 'installation', 'installable', 'install', 'instability', 'inspired', 'insist', 'insights', 'inside', 'insert', 'ins', 'inquirer', 'input', 'innerhtml', 'inlined', 'injecting', 'initialized', 'ing', 'information', 'info', 'influence', 'infiniteloader', 'infinite', 'infer', 'indirection', 'indexing', 'index', 'increasing', 'including', 'includes', 'included', 'include', 'improved', 'improve', 'improper', 'imported', 'important', 'import', 'implied', 'implements', 'implemented', 'implementation', 'implement', 'imperative_programming', 'imgur', 'img', 'images', 'image', 'imac', 'im', 'ignoreunknownversions', 'ignore', 'iexplore', 'ie9', 'ie11', 'ie10', 'ides', 'identical', 'ideas', 'ideal', 'idea', 'idangero', 'id1071725095', 'id1063631769', 'id', 'iamvdo', 'hybrids', 'human', 'hug', 'hubspot', 'httparchive', 'http', 'html5accessibility', 'html', 'hterkelsen', 'hpe', 'howto', 'howdyai', 'howdy', 'hover', 'hostname', 'host', 'hooks', 'hook', 'home', 'hl', 'hit', 'hill', 'highly', 'highlighting', 'highlighted', 'higher', 'high', 'hidroh', 'hidden', 'hey', 'herokuapp', 'helpful', 'helpers', 'helped', 'height', 'heart', 'hears', 'heard', 'headers', 'header', 'head', 'haven', 'hard', 'happy', 'happened', 'happen', 'handling', 'handles', 'handler', 'handled', 'handle', 'handbook', 'hand', 'hamburgers', 'hamburger', 'hacks', 'hacking', 'h1', 'gzip', 'gulp', 'guise', 'guides', 'guideline', 'guide', 'gui', 'guesstimate', 'guess', 'gucong3000', 'guard', 'guaranteed', 'grunt', 'group', 'grids', 'grid', 'greenkeeper', 'green', 'greater', 'great', 'gray', 'gratipay', 'grateful', 'graphs', 'graph', 'granted', 'grandpa', 'gradients', 'gradient', 'grade', 'gpu', 'gpa', 'gotwarlost', 'googlechrome', 'google', 'godaddy', 'goal', 'gnode', 'glossary', 'globals', 'global_objects', 'global', 'glob', 'glitch', 'giving', 'give', 'gittip', 'gitter', 'githubusercontent', 'github', 'git', 'gif', 'gianasista', 'gh', 'gfx', 'getter', 'getstarted', 'getguesstimate', 'generators', 'generator', 'generated', 'generate', 'generally', 'general', 'gems', 'gemfile', 'gatherers', 'gap', 'gallery', 'ga', 'future', 'furnished', 'functions', 'functional', 'function', 'func', 'fun', 'fully', 'full', 'fs', 'fritz', 'friends', 'friendly', 'friend', 'frequently', 'freesuraj', 'freenode', 'free', 'framework', 'fragment', 'fr_fr', 'foundation', 'found', 'forward', 'formdata', 'form', 'forks', 'forking', 'fork', 'forget', 'foreachof', 'forceupdategrids', 'forceupdategrid', 'forcefully', 'force', 'forbidden_header_name', 'foot', 'foo', 'fonts', 'font', 'follow', 'folder', 'focused', 'focusable', 'focus', 'fn', 'fly', 'flux', 'flowtype', 'flow', 'flexibility', 'flexbugs', 'flexbox', 'flex', 'flavors', 'flat', 'flags', 'flag', 'fkd7hjknpt', 'fixes', 'fixed', 'fix', 'fitness', 'fit', 'firefox_48x48', 'firefox', 'fired', 'finished', 'fine', 'find', 'final', 'filter', 'files', 'filed', 'file', 'fields', 'field', 'ffd7be7bf628', 'ffb400', 'ff69b4', 'fezvrasta', 'fetchuser', 'fetchpolyfill', 'fetch', 'features', 'featured', 'feature', 'fc4e', 'fbos', 'fastlane', 'faster', 'fast', 'faq', 'family', 'familiarize', 'familiar', 'falsy', 'false', 'failwithoutassertions', 'failure', 'fails', 'failing', 'fail', 'fact', 'facebook', 'eyes', 'externalresources', 'external', 'extensions', 'extension', 'extended', 'expressjs', 'express', 'exports', 'exported', 'export', 'exploring', 'explicitly', 'explanations', 'explanation', 'experiments', 'experience', 'expected', 'expanded', 'exits', 'execution', 'executed', 'execute', 'executable', 'excluded', 'exclude', 'excel', 'examples', 'example_audit', 'ex7oladjt', 'evil', 'events', 'event', 'eval', 'establish', 'esnext', 'esm', 'eslint', 'es_es', 'es6', 'es5', 'es2015', 'es', 'errors', 'error', 'errinvalidproj', 'err', 'equivalent', 'epub', 'environment', 'env', 'entry', 'entries', 'enterprise', 'ensure', 'english', 'engine', 'endpoint', 'ending', 'end', 'encourage', 'encountered', 'encounter', 'enables', 'enabled', 'enable', 'en_us', 'en', 'ems', 'emphatic', 'emilebres', 'ember', 'emazzotta', 'elements', 'element', 'elegant', 'electron_48x48', 'electron', 'elastic', 'ejs', 'efficient', 'effects', 'effect', 'ef1589d7063f', 'editing', 'edit', 'edgurgel', 'edge_48x48', 'edge', 'ecmascript', 'ebidel', 'easy', 'easiest', 'easier', 'ease', 'early', 'e8', 'e7v91mu2m5x48ehx', 'e79e', 'e7', 'e6', 'e5', 'dwres', 'duration', 'due', 'dt', 'drop', 'drive', 'draios', 'dragging', 'drag', 'dq', 'downs', 'downloads', 'download', 'dougwilson', 'dont_trip_up_with_vendor_prefixes', 'donation', 'donate', 'don', 'domains', 'domain', 'dom', 'doiuse', 'doesn', 'documentation', 'document', 'docs', 'docker', 'dm', 'dive', 'div', 'distributions', 'distribution', 'distribute', 'dist', 'displaying', 'displayed', 'display', 'dispatcher', 'dispatched', 'dispatch', 'disk', 'discussion', 'discuss', 'discover', 'disabled', 'disable', 'disabilities', 'directory', 'directories', 'directly', 'diogofcunha', 'dig', 'difference', 'diff', 'didn', 'devtools', 'devkit', 'devices', 'device', 'devhub', 'development', 'developers', 'developer', 'developed', 'develop', 'devdeps', 'devdependencies', 'dev', 'determine', 'detection', 'detected', 'detect', 'details', 'detail', 'destroy', 'destination', 'desired', 'designed', 'design', 'description', 'describe', 'derived', 'derive', 'derivations', 'derivation', 'depth', 'deps', 'deploy', 'depends', 'depending', 'dependents', 'dependency', 'dependencies', 'depended', 'depend', 'depdendency', 'demonstrated', 'demand', 'delegation', 'defining', 'defined', 'define', 'defaults', 'default', 'decorators', 'decorator', 'decorate', 'declared', 'declare', 'declarations', 'debugging', 'debug', 'deals', 'dealings', 'deal', 'de', 'dbrgn', 'david', 'datepicker', 'date', 'database', 'data', 'dash', 'damages', 'd460fxti0f', 'cz', 'customizingstyles', 'customize', 'customization', 'customizable', 'custom', 'current', 'currency_code', 'cumbersome', 'cult', 'cssnext', 'cssnano', 'css', 'csrf', 'cs1707', 'crucial', 'cross', 'credits', 'credentials', 'creator', 'creation', 'creatinganinfiniteloadinglist', 'creating', 'createstore', 'created', 'create', 'crawl', 'craving', 'crash', 'cpu', 'cpp', 'covert', 'covered', 'coveralls', 'coverage', 'courses', 'couple', 'counted', 'count', 'cors', 'correct', 'corner', 'coreh', 'core', 'copyright', 'copy', 'copies', 'coordinate', 'cookies', 'cookie', 'cook', 'converted', 'convert', 'convention', 'controls', 'controller', 'control', 'contributors', 'contributions', 'contributing', 'contribute', 'contrast', 'contrary', 'contract', 'continue', 'context', 'contents', 'content', 'constructor', 'constitutes', 'constants', 'const', 'consolidate', 'console', 'consistent', 'considered', 'connection', 'connecting', 'connect', 'conflict', 'configured', 'configure', 'configuration', 'configs', 'config', 'conditions', 'concurrent', 'concurrency', 'concordancejs', 'concordance', 'concise', 'concerned', 'concepts', 'concept', 'computes', 'computer', 'computed', 'compute', 'computations', 'compression', 'comprehensive', 'componentwillupdate', 'componentwillunmount', 'componentwillmount', 'components', 'componentdidupdate', 'component', 'complicated', 'compliant', 'complex', 'completion', 'completing', 'complete', 'compiler', 'compileenhancements', 'compiled', 'compile', 'compatible', 'compatibility', 'compat', 'compass', 'comparison', 'comparing', 'compared', 'compare', 'companies', 'community', 'commune', 'commonjs', 'common', 'committed', 'commitizen', 'commit', 'comment', 'commands', 'commander', 'command', 'comma', 'combined', 'combinations', 'columnsizer', 'column', 'color', 'collections', 'collection', 'collapse', 'coinbase', 'codesandbox', 'codemods', 'codecov', 'code_style', 'code', 'codacy', 'cocoapods', 'cmd', 'cloudup', 'closed', 'clone', 'clients', 'client', 'clicking', 'clicked', 'click', 'cli', 'clean', 'clauderic', 'classfellow', 'classes', 'class', 'claim', 'ciscospark', 'ciscojabber', 'circleci', 'ci', 'chromium', 'chrome_path', 'chrome_48x48', 'chrome', 'chrisallenlane', 'choose', 'choices', 'choice', 'chinese', 'child', 'checkstatus', 'check', 'cheat', 'chat', 'charge', 'character', 'channel', 'changing', 'changelog', 'changed', 'change', 'chance', 'challenges', 'chain', 'cgi', 'cells', 'cellmeasurer', 'cdns', 'cdnjs', 'cdn', 'cc', 'cb', 'causality', 'cats', 'category', 'categorical', 'catched', 'catch', 'cases', 'case', 'cascade', 'care', 'capture', 'capable', 'caolan', 'canonical', 'caniuse', 'cancelled', 'calling', 'called', 'callbacks', 'callback', 'call', 'calibre', 'calculations', 'calculation', 'calculate', 'caching', 'cached', 'bvaughn', 'buy', 'button', 'business', 'built', 'buildtools', 'buildscripts', 'builds', 'building', 'build_webpack', 'build_optimizer', 'build_ng_packagr', 'build_dev', 'build_connector', 'build_angular', 'build', 'bugs', 'bug', 'btn_donatecc_lg', 'btn', 'bs', 'brunch', 'browsersync', 'browserslist', 'browsers', 'browserify', 'browser', 'browse', 'brought', 'brotli', 'broccoli', 'bring', 'brightgreen', 'bridge', 'brevity', 'breaking', 'break', 'branches', 'branch', 'brainmaestro', 'brackets', 'br', 'boyney123', 'box', 'bower', 'bound', 'bots', 'botkit', 'botframework', 'bot', 'boring', 'border', 'bootstrap', 'boom', 'boolean', 'books', 'book', 'boils', 'boilerplates', 'body', 'bn', 'blogs', 'blocksize', 'blocking', 'block', 'blob', 'blipmdconlkpinefehnmjammfjpmpbjk', 'blakeembrey', 'bl', 'bit', 'binaries', 'bin', 'bigquery', 'big', 'bhp', 'beta', 'benefit', 'beneficial', 'belong', 'believed', 'behavior', 'behaves', 'begins', 'beginnertutorial', 'beforeeach', 'beautiful', 'beaconed', 'bd', 'bcoe', 'bc', 'baz', 'basics', 'basically', 'basic', 'bash', 'based', 'bar', 'badges', 'badge_grade', 'badge', 'backport', 'background', 'backers', 'backer', 'backed', 'back', 'babel', 'b8', 'b5', 'b1', 'ay', 'aws40kox90u', 'awesome', 'awaited', 'await', 'avoided', 'avoid', 'avatar', 'avajs', 'ava__js', 'ava', 'autosizer', 'autorun', 'autoprefixer', 'autoprefix', 'automatization', 'automatically', 'authors', 'authoring', 'author', 'authentication', 'audits', 'auditing', 'audit', 'attribute', 'attempts', 'attaching', 'atom', 'atlassian', 'atlaskit', 'asynchronous', 'async', 'assign', 'assets', 'assertions', 'assertion', 'assert', 'art', 'arrowturn', 'arrowkeystepper', 'arrowalt', 'arrow', 'arrays', 'array', 'arising', 'aria', 'arguments', 'argument', 'args', 'aren', 'area', 'archive', 'architecture', 'architect', 'appveyor', 'apps', 'approach', 'applystyles', 'applystyle', 'applyreactstyle', 'applymiddleware', 'apply', 'applies', 'applied', 'application', 'applicable', 'apple', 'apphub', 'append', 'appearance', 'app', 'apmssu3qnf0', 'apis', 'api', 'antd', 'ant', 'answer', 'annotating', 'animations', 'animated', 'angularjs', 'angular_devkit', 'angular', 'android', 'andreasonny83', 'andreascn', 'andersevenrud', 'anatomy', 'analyzing', 'analyze', 'analysis', 'amp', 'amount', 'amd', 'altoghether', 'alternatively', 'alternative', 'alter', 'alrra', 'allowing', 'allowed', 'alipayobjects', 'alias', 'algolia', 'alfred', 'alfmdwaaqbaj', 'aims', 'aimed', 'aim', 'ai', 'ahead', 'aggregate', 'aftereach', 'affecting', 'affect', 'ae', 'advisable', 'advantage', 'advanced', 'adjust', 'adjacent', 'addyosmani', 'addition', 'adding', 'added', 'add', 'actual', 'acts', 'activities', 'actively', 'active', 'actions', 'actioncounts', 'action', 'achieved', 'achieve', 'account', 'accomplished', 'accessing', 'accessible', 'accessibility', 'accessed', 'access_control_cors', 'access', 'accepts', 'ac', 'abortsignal', 'aborting', 'abortcontroller', 'abortable', 'ability', 'aa1bac299dd0', 'a8', 'a7', '9f91', '9df04cb6', '9c', '99', '97f3', '962647f8', '92ce', '92', '9129', '91', '903e', '9000', '900', '90', '8m', '8f', '8de3', '8c', '8b', '89', '88a95d146a75', '88', '84', '83be', '83', '81', '8001', '80', '7c9e7c86e0c6ead141bb0539d33143d0e1f576dd', '732', '6px', '6833f9', '67', '663be15acb3dd2eb0f71b1956ef28c2cd3fdeed0', '640x360', '635902', '6002', '60', '5px', '5ms', '5ed9c7', '59cb', '5983843', '59', '58', '57', '56', '55', '54557dc319b04e92e31cb87427bef194ec1c549c', '54', '5368888', '53', '5224894a7cca', '52', '512', '51', '500', '50', '4x', '4px', '4ef7d128', '4d63', '49', '485645', '484499', '48', '47073b5b58aa6f3fb24f98757be5d3f56218d160', '47', '46', '45000', '45', '44', '43', '42m', '42', '412', '41', '40px', '40m', '40gmail', '40556658', '40', '3dy', '3dxy', '3dx', '3bbase64', '3anonhosted', '3aimage', '3abtn_donate_sm', '3a', '39191', '39', '38', '37311298', '37', '36', '35', '34', '33px', '33', '32', '31', '30330590', '3000x3000', '3000', '300', '30', '2x', '2px', '2m', '2fwebpack', '2fupdate', '2fsvg', '2fschematics', '2fr', '2fpwa', '2fcore', '2fcli', '2fbuild', '2farchitect', '2fangular', '2egif', '2ecom', '2dsync', '2dppx', '2ddonationsbf', '2cphn2zybpzd0itgl2zwxsb18xiibkyxrhlw5hbwu9ikxpdmvsbg8gmsigeg1sbnm9imh0dha6ly93d3cudzmub3jnlziwmdavc3zniib2awv3qm94psiwidagmtagoci', '2bxml', '2bphn0ewxlpi5jbhmtmxtmawxsoinmzmy7ftwvc3r5bgu', '2bphbhdgggy2xhc3m9imnscy0xiibkpsjnnswwqy40miwwldasljyzldasmy4zngmwldeuodqumtksmi43miwxljc0ldmumwgwvjcunthhljq0ljq0ldasmcwwlc42oc4znuw0ljm1ldyunjlinwm0lju4ldasns0unjmsns0zljm1uzkuntgsmcw1ldbattiuodmsnc4xogeunjmunjmsmcwxldesljy1ls42m0eunjqunjqsmcwwldesmi44myw0lje4wk01ldqumthhljyzljyzldasmswxlc42ns0unjnbljy0ljy0ldasmcwxldusnc4xofptmi4xnywwys42my42mywwldesmswunjutljyzqs42nc42ncwwldasmsw3lje3ldqumthaii8', '2bpgrlznm', '2bpc9zdmc', '2bpc9kzwzzpjx0axrszt5zcgvjdhj1btwvdgl0bgu', '2b', '2978551', '29597', '29', '28', '27', '261d', '26', '25', '2464966', '24', '23f662f6', '2301202', '23', '22478294', '22', '214067', '2109932', '21', '20your', '20with', '20weststrate', '20the', '20supercharge', '20state', '20start', '20react', '20quick', '20px', '20mobx', '20in', '20guide', '20client', '20chat', '20apps', '2017', '2000', '20', '1s', '1fr', '1dri6hc2xiptlbkfnh8yczocpxqtoacicdesdyvfg_be', '1c3a711a', '19', '18', '17391516', '17', '16', '15s', '15px', '15', '1462115295', '14333364', '14', '13', '12m', '127', '1234', '12', '11e8', '11e6', '11e5', '11737732', '113399210633478618934', '11', '10s', '1050222', '100px', '100', '10', '0ca1e55e', '098f2f8ed866', '09', '0800', '000']\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "2483"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 13
        }
      ]
    },
    {
      "metadata": {
        "id": "zsXDPcu4axp7",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "### Perform keyword extraction for individual reademe files and\n",
        "### remove keywords that are not in the target input_nodes\n",
        "readme_keywords_array = []\n",
        "for readme_entry in readme_array:\n",
        "  raked_entry = rake.extract_keywords_from_text(readme_entry[\"content\"])\n",
        "  ranked_phrases = rake.get_ranked_phrases()\n",
        "  # Filter keywords\n",
        "  ranked_phrases = list(filter(lambda x: x in input_nodes,ranked_phrases))\n",
        "  # Build keywords object\n",
        "  keywords = {}\n",
        "  keywords[\"reponame\"] = readme_entry[\"reponame\"]\n",
        "  keywords[\"keywords\"] = ranked_phrases\n",
        "  readme_keywords_array.append(keywords)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "4S-R3TpNqUwW",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "f8695a28-e575-49b6-e8a0-986139aab2f9"
      },
      "cell_type": "code",
      "source": [
        "# Creating a dictionary for keywords\n",
        "repos = []\n",
        "keywords_dic={}\n",
        "for i in range(len(readme_keywords_array)):\n",
        "  repos.append(readme_keywords_array[i][\"reponame\"])\n",
        "  keywords_dic[readme_keywords_array[i][\"reponame\"]] = readme_keywords_array[i][\"keywords\"]\n",
        "len(repos)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "22"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 15
        }
      ]
    },
    {
      "metadata": {
        "id": "9tRZPQgdsLIF",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "4c407bda-ed8f-46eb-adfd-9d8c502f96c1"
      },
      "cell_type": "code",
      "source": [
        "# Creating a dictionary for packages\n",
        "packages_dic={}\n",
        "for repo in package_array:\n",
        "  if repo[\"reponame\"] in repos:\n",
        "    packages_dic[repo[\"reponame\"]] = []\n",
        "    if \"devDependencies\" in repo:\n",
        "      packages_dic[repo[\"reponame\"]].append(repo[\"devDependencies\"])\n",
        "    if \"dependencies\" in repo:\n",
        "      packages_dic[repo[\"reponame\"]].append(repo[\"dependencies\"])\n",
        "len(packages_dic.keys())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "22"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 16
        }
      ]
    },
    {
      "metadata": {
        "id": "4Wg_p5YxjEuM",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# Training Data"
      ]
    },
    {
      "metadata": {
        "id": "KoOxSwJwXZ_D",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# list keywords for each repo\n",
        "Questions = [keywords_dic[i] for i in repos]\n",
        "max_input_len = max(map(len, Questions))\n",
        "\n",
        "# list of unique keywords\n",
        "input_tokens = list({keyword for repo_list in Questions for keyword in repo_list})\n",
        "num_input_tokens = len(input_tokens)\n",
        "keyword_to_index = {}\n",
        "for i in range(len(input_tokens)):\n",
        "  keyword_to_index[input_tokens[i]]=i\n",
        "index_to_keyword = {v: k for k, v in keyword_to_index.items()}"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "-6ZkS2YwXxAm",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# list packages for each repo\n",
        "Answers = [list(packages_dic[i][0]) for i in repos]\n",
        "max_output_len = max(map(len, Answers))\n",
        "\n",
        "# list of unique packages\n",
        "output_tokens = list({pkg for repo_pkg in Answers for pkg in repo_pkg})\n",
        "num_output_tokens = len(output_tokens)\n",
        "package_to_index = {}\n",
        "for i in range(len(output_tokens)):\n",
        "  package_to_index[output_tokens[i]]=i\n",
        "index_to_package = {v: k for k, v in package_to_index.items()}"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "bmDcUgyS1oxQ",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "n = len(repos)\n",
        "X = np.zeros((n, max_input_len, num_input_tokens), dtype=np.bool)\n",
        "y = np.zeros((n, num_output_tokens), dtype=np.bool)\n",
        "for num_pair, (question, answer) in enumerate(zip(Questions, Answers)):\n",
        "  for num_token, q_token in enumerate(question):\n",
        "    X[num_pair, num_token, keyword_to_index[q_token]] = 1\n",
        "  for num_token, a_token in enumerate(answer):\n",
        "    y[num_pair, package_to_index[a_token]] = 1"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "wzuV40b0Bdo1",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# Build the model"
      ]
    },
    {
      "metadata": {
        "id": "Tdf-qb0SAWvJ",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "hidden_units=128\n",
        "model = Sequential()\n",
        "input_s = (None, num_input_tokens)\n",
        "model.add(LSTM(hidden_units, input_shape=input_s))\n",
        "model.add(Dense(num_output_tokens))\n",
        "model.add(Activation('softmax'))\n",
        "optimizer = RMSprop(lr=0.01)\n",
        "model.compile(loss='categorical_crossentropy', optimizer=optimizer)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "fMHGgz8GE4eC",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# Training"
      ]
    },
    {
      "metadata": {
        "id": "GL--1oleEzPh",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 3077
        },
        "outputId": "e113122c-bcc1-49d1-c8c9-585b181d9805"
      },
      "cell_type": "code",
      "source": [
        "iterations=20\n",
        "bat_size=256\n",
        "num_epochs=3\n",
        "for iteration in range(iterations):\n",
        "  print()\n",
        "  print('-' * 50)\n",
        "  print('Iteration', iteration)\n",
        "  model.fit(X, y, batch_size=bat_size, epochs=num_epochs)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "--------------------------------------------------\n",
            "Iteration 0\n",
            "Epoch 1/3\n",
            "22/22 [==============================] - 5s 217ms/step - loss: 99.4402\n",
            "Epoch 2/3\n",
            "22/22 [==============================] - 4s 178ms/step - loss: 132.6980\n",
            "Epoch 3/3\n",
            "22/22 [==============================] - 4s 168ms/step - loss: 99.2543\n",
            "\n",
            "--------------------------------------------------\n",
            "Iteration 1\n",
            "Epoch 1/3\n",
            "22/22 [==============================] - 4s 165ms/step - loss: 99.7557\n",
            "Epoch 2/3\n",
            "22/22 [==============================] - 4s 168ms/step - loss: 97.0515\n",
            "Epoch 3/3\n",
            "22/22 [==============================] - 4s 163ms/step - loss: 97.0115\n",
            "\n",
            "--------------------------------------------------\n",
            "Iteration 2\n",
            "Epoch 1/3\n",
            "22/22 [==============================] - 4s 162ms/step - loss: 97.5773\n",
            "Epoch 2/3\n",
            "22/22 [==============================] - 4s 162ms/step - loss: 99.6110\n",
            "Epoch 3/3\n",
            "22/22 [==============================] - 4s 164ms/step - loss: 103.7948\n",
            "\n",
            "--------------------------------------------------\n",
            "Iteration 3\n",
            "Epoch 1/3\n",
            "22/22 [==============================] - 4s 163ms/step - loss: 104.4793\n",
            "Epoch 2/3\n",
            "22/22 [==============================] - 4s 164ms/step - loss: 102.2305\n",
            "Epoch 3/3\n",
            "22/22 [==============================] - 4s 164ms/step - loss: 100.9621\n",
            "\n",
            "--------------------------------------------------\n",
            "Iteration 4\n",
            "Epoch 1/3\n",
            "22/22 [==============================] - 4s 163ms/step - loss: 99.9721\n",
            "Epoch 2/3\n",
            "22/22 [==============================] - 4s 162ms/step - loss: 99.2041\n",
            "Epoch 3/3\n",
            "22/22 [==============================] - 4s 162ms/step - loss: 98.3919\n",
            "\n",
            "--------------------------------------------------\n",
            "Iteration 5\n",
            "Epoch 1/3\n",
            "22/22 [==============================] - 4s 165ms/step - loss: 98.2448\n",
            "Epoch 2/3\n",
            "22/22 [==============================] - 4s 164ms/step - loss: 97.9506\n",
            "Epoch 3/3\n",
            "22/22 [==============================] - 4s 162ms/step - loss: 97.6946\n",
            "\n",
            "--------------------------------------------------\n",
            "Iteration 6\n",
            "Epoch 1/3\n",
            "22/22 [==============================] - 4s 165ms/step - loss: 97.8610\n",
            "Epoch 2/3\n",
            "22/22 [==============================] - 4s 163ms/step - loss: 97.8262\n",
            "Epoch 3/3\n",
            "22/22 [==============================] - 4s 166ms/step - loss: 98.0323\n",
            "\n",
            "--------------------------------------------------\n",
            "Iteration 7\n",
            "Epoch 1/3\n",
            "22/22 [==============================] - 4s 168ms/step - loss: 98.1567\n",
            "Epoch 2/3\n",
            "22/22 [==============================] - 4s 165ms/step - loss: 98.3947\n",
            "Epoch 3/3\n",
            "22/22 [==============================] - 4s 167ms/step - loss: 98.3846\n",
            "\n",
            "--------------------------------------------------\n",
            "Iteration 8\n",
            "Epoch 1/3\n",
            "22/22 [==============================] - 4s 168ms/step - loss: 98.6910\n",
            "Epoch 2/3\n",
            "22/22 [==============================] - 4s 170ms/step - loss: 98.7881\n",
            "Epoch 3/3\n",
            "22/22 [==============================] - 4s 169ms/step - loss: 98.9030\n",
            "\n",
            "--------------------------------------------------\n",
            "Iteration 9\n",
            "Epoch 1/3\n",
            "22/22 [==============================] - 4s 172ms/step - loss: 98.6824\n",
            "Epoch 2/3\n",
            "22/22 [==============================] - 4s 170ms/step - loss: 98.5654\n",
            "Epoch 3/3\n",
            "22/22 [==============================] - 4s 168ms/step - loss: 98.4951\n",
            "\n",
            "--------------------------------------------------\n",
            "Iteration 10\n",
            "Epoch 1/3\n",
            "22/22 [==============================] - 4s 166ms/step - loss: 98.4426\n",
            "Epoch 2/3\n",
            "22/22 [==============================] - 4s 167ms/step - loss: 98.4664\n",
            "Epoch 3/3\n",
            "22/22 [==============================] - 4s 167ms/step - loss: 98.4785\n",
            "\n",
            "--------------------------------------------------\n",
            "Iteration 11\n",
            "Epoch 1/3\n",
            "22/22 [==============================] - 4s 167ms/step - loss: 98.5796\n",
            "Epoch 2/3\n",
            "22/22 [==============================] - 4s 168ms/step - loss: 98.6713\n",
            "Epoch 3/3\n",
            "22/22 [==============================] - 4s 166ms/step - loss: 98.7233\n",
            "\n",
            "--------------------------------------------------\n",
            "Iteration 12\n",
            "Epoch 1/3\n",
            "22/22 [==============================] - 4s 165ms/step - loss: 98.6845\n",
            "Epoch 2/3\n",
            "22/22 [==============================] - 4s 162ms/step - loss: 98.6536\n",
            "Epoch 3/3\n",
            "22/22 [==============================] - 4s 159ms/step - loss: 98.5435\n",
            "\n",
            "--------------------------------------------------\n",
            "Iteration 13\n",
            "Epoch 1/3\n",
            "22/22 [==============================] - 3s 159ms/step - loss: 98.5063\n",
            "Epoch 2/3\n",
            "22/22 [==============================] - 3s 158ms/step - loss: 98.4836\n",
            "Epoch 3/3\n",
            "22/22 [==============================] - 3s 159ms/step - loss: 98.6628\n",
            "\n",
            "--------------------------------------------------\n",
            "Iteration 14\n",
            "Epoch 1/3\n",
            "22/22 [==============================] - 4s 160ms/step - loss: 98.7685\n",
            "Epoch 2/3\n",
            "22/22 [==============================] - 3s 158ms/step - loss: 98.7849\n",
            "Epoch 3/3\n",
            "22/22 [==============================] - 4s 160ms/step - loss: 98.7552\n",
            "\n",
            "--------------------------------------------------\n",
            "Iteration 15\n",
            "Epoch 1/3\n",
            "22/22 [==============================] - 4s 160ms/step - loss: 98.6839\n",
            "Epoch 2/3\n",
            "22/22 [==============================] - 4s 160ms/step - loss: 98.5802\n",
            "Epoch 3/3\n",
            "22/22 [==============================] - 4s 162ms/step - loss: 98.5672\n",
            "\n",
            "--------------------------------------------------\n",
            "Iteration 16\n",
            "Epoch 1/3\n",
            "22/22 [==============================] - 4s 160ms/step - loss: 98.5419\n",
            "Epoch 2/3\n",
            "22/22 [==============================] - 3s 159ms/step - loss: 98.5246\n",
            "Epoch 3/3\n",
            "22/22 [==============================] - 3s 159ms/step - loss: 98.5343\n",
            "\n",
            "--------------------------------------------------\n",
            "Iteration 17\n",
            "Epoch 1/3\n",
            "22/22 [==============================] - 4s 160ms/step - loss: 98.5454\n",
            "Epoch 2/3\n",
            "22/22 [==============================] - 3s 159ms/step - loss: 98.5712\n",
            "Epoch 3/3\n",
            "22/22 [==============================] - 3s 159ms/step - loss: 98.5747\n",
            "\n",
            "--------------------------------------------------\n",
            "Iteration 18\n",
            "Epoch 1/3\n",
            "22/22 [==============================] - 3s 158ms/step - loss: 98.6068\n",
            "Epoch 2/3\n",
            "22/22 [==============================] - 3s 158ms/step - loss: 98.6055\n",
            "Epoch 3/3\n",
            "22/22 [==============================] - 4s 159ms/step - loss: 98.6195\n",
            "\n",
            "--------------------------------------------------\n",
            "Iteration 19\n",
            "Epoch 1/3\n",
            "22/22 [==============================] - 3s 159ms/step - loss: 98.5940\n",
            "Epoch 2/3\n",
            "22/22 [==============================] - 4s 159ms/step - loss: 98.6062\n",
            "Epoch 3/3\n",
            "22/22 [==============================] - 4s 159ms/step - loss: 98.5841\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "roOFONAuLXK6",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "model.save(\"neural_token\")"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "JUE1Wa8ESmh3",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# Prediction"
      ]
    },
    {
      "metadata": {
        "id": "GQqtQNkVSwqi",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "repo=repos[1]\n",
        "keys = keywords_dic[repo]\n",
        "keys_encoded = np.zeros((1, max_input_len, num_input_tokens), dtype=np.bool)\n",
        "for num_token, q in enumerate(keys):\n",
        "    X[1, num_token, keyword_to_index[q]] = 1"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "j5sBKfAmPJ96",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 364
        },
        "outputId": "66c08463-d71e-48db-e966-ebe319b151b4"
      },
      "cell_type": "code",
      "source": [
        "preds = model.predict(keys_encoded, verbose=0)[0]\n",
        "plt.plot(preds)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<matplotlib.lines.Line2D at 0x7f1ad199b128>]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 24
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe0AAAFKCAYAAAAwrQetAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f1ad194ca20>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "MC00ZSr_ZEKU",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 85
        },
        "outputId": "f437511e-1e05-454a-ef45-ce72858f43c2"
      },
      "cell_type": "code",
      "source": [
        "ix=np.argsort(preds)\n",
        "print(index_to_package[ix[-1]])\n",
        "print(index_to_package[ix[-2]])\n",
        "print(index_to_package[ix[-3]])\n",
        "print(index_to_package[ix[-4]])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "gulp\n",
            "babel-loader\n",
            "babel-preset-es2015\n",
            "mocha\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "VDyMlZFoSedr",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 85
        },
        "outputId": "3f1f0c2b-2f38-4ca4-c0da-0f100d645fe5"
      },
      "cell_type": "code",
      "source": [
        "packages_dic[repo]"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[{'browserify': '^10.0.0',\n",
              "  'istanbul-harmony': '0',\n",
              "  'mocha': '^2.0.0',\n",
              "  'mz': '^1.0.2'}]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 26
        }
      ]
    }
  ]
}