bagofwords-clustering.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## BAG OF WORDS DATABASE\n",
    "### KOS blog entries: orig source: dailykos.com "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Descarga de la colección de textos preprocesada"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import classicdb.fetch as fetch\n",
    "\n",
    "url = \"https://sites.google.com/site/xchgdir/public/kos.tar.gz\"\n",
    "classic_home = os.path.join(fetch.get_data_home(), \"kos_home\")\n",
    "corpus_root = os.path.join(classic_home, \"kos\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "download and extracting file from https://sites.google.com/site/xchgdir/public/kos.tar.gz\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "downloaded and extracted to C:\\Users\\lmamani\\scikit_learn_data\\kos_home\n"
     ]
    }
   ],
   "source": [
    "print(\"download and extracting file from \" + url)\n",
    "fetch.download_and_unzip(url, classic_home, \"kos.tar.gz\")\n",
    "print(\"downloaded and extracted to \" + classic_home)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Parseo y carga en memoria"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "6906 words in vocabulary\n3430 documents\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "\n",
    "vocabfile = os.path.join(corpus_root, \"vocab.kos.txt\")\n",
    "docwordfile = os.path.join(corpus_root, \"docword.kos.txt\")\n",
    "\n",
    "with open(vocabfile, 'r') as f:\n",
    "    vocab = [line.strip() for line in f.readlines()]\n",
    "\n",
    "bagdicts = []\n",
    "with open(docwordfile, 'r') as f:\n",
    "    for line in f.readlines():\n",
    "        w = line.strip().split()\n",
    "        if len(w) == 3:\n",
    "            docid = int(w[0]) - 1\n",
    "            wordid = int(w[1]) - 1\n",
    "            wordcount = int(w[2])\n",
    "            if len(bagdicts) <= docid:\n",
    "                bagdicts.append({})  # add new dictionary\n",
    "            dict = bagdicts[docid]\n",
    "            dict[vocab[wordid]] = wordcount  # add new (vocab[wordid], count) into dictionary\n",
    "            # dict[wordid] = wordcount  # add new (wordid, count) into dictionary\n",
    "\n",
    "print(len(vocab), \"words in vocabulary\")\n",
    "print(len(bagdicts), \"documents\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Conteo de tokens del corpus\n",
    "- Cada fila representa un documento y está en el orden leído del documento\n",
    "- Cada columna representa un término y ha sido reordenado (por el Dict)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cols = frequency by term, row = frequency by file\n[[ 0.  0.  0. ...,  0.  1.  0.]\n [ 0.  0.  0. ...,  0.  0.  0.]\n [ 0.  0.  0. ...,  0.  0.  0.]\n ..., \n [ 0.  0.  0. ...,  0.  0.  0.]\n [ 0.  0.  0. ...,  0.  0.  0.]\n [ 0.  0.  0. ...,  0.  0.  0.]]\n\nfile # 0 has 137 terms\nfile # 1 has 77 terms\nfile # 2 has 157 terms\n...\n\nSome feature terms\n['aarp', 'abandon', 'abandoned', 'abandoning', 'abb', 'abc', 'abcs', 'abdullah', 'ability', 'aboard', 'abortion', 'abortions', 'abraham', 'abrams', 'abroad', 'absence', 'absent', 'absentee', 'absolute', 'absolutely', 'abstain', 'absurd', 'abu', 'abuse', 'abuses', 'academy', 'accent', 'accept', 'acceptable', 'acceptance', 'accepted', 'accepting', 'access', 'accident', 'accidentally', 'accidents', 'accomplish', 'accomplished', 'accomplishment', 'account', 'accountability', 'accountable', 'accounting', 'accounts', 'accuracy', 'accurate', 'accurately', 'accusations', 'accused', 'accusing', 'achieve', 'achieved', 'achievements', 'acknowledge', 'acknowledged', 'aclu', 'act', 'actblue', 'acted', 'acting', 'action', 'actions', 'active', 'actively', 'activism', 'activist', 'activists', 'activities', 'activity', 'actor', 'acts', 'actual', 'adam', 'adams', 'add', 'added', 'adding', 'addition', 'additional', 'address', 'addressed', 'addresses', 'addressing', 'adds', 'adjust', 'adjusted', 'adjustments', 'admin', 'administration', 'administrations', 'administrative', 'administrator', 'admire', 'admission', 'admit', 'admits', 'admitted', 'admitting', 'ado', 'adopt']\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "from sklearn.feature_extraction import DictVectorizer\n",
    "\n",
    "vectorizer = DictVectorizer(sparse=False)\n",
    "count_matrix = vectorizer.fit_transform(bagdicts)\n",
    "print(\"cols = frequency by term, row = frequency by file\")\n",
    "print(count_matrix)\n",
    "print()\n",
    "\n",
    "for f in range(0, 3):\n",
    "    print(\"file #\", f, \"has\", int(np.sum(count_matrix[f])), \"terms\")\n",
    "print(\"...\")\n",
    "print()\n",
    "\n",
    "print(\"Some feature terms\")\n",
    "print(vectorizer.get_feature_names()[0:100])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Cálculo de matriz TFID"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TFID matrix: (3430, 6906)\n[[ 0.          0.          0.         ...,  0.          0.08710494  0.        ]\n [ 0.          0.          0.         ...,  0.          0.          0.        ]\n [ 0.          0.          0.         ...,  0.          0.          0.        ]\n ..., \n [ 0.          0.          0.         ...,  0.          0.          0.        ]\n [ 0.          0.          0.         ...,  0.          0.          0.        ]\n [ 0.          0.          0.         ...,  0.          0.          0.        ]]\n\n"
     ]
    }
   ],
   "source": [
    "from sklearn.feature_extraction.text import TfidfTransformer\n",
    "\n",
    "transformer = TfidfTransformer()\n",
    "tfidf = transformer.fit_transform(count_matrix)\n",
    "\n",
    "print(\"TFID matrix:\", tfidf.shape)\n",
    "print(tfidf.toarray())\n",
    "print()\n",
    "\n",
    "# print(\"Weights of each feature computed by fit\")\n",
    "# print(transformer.idf_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Clusterización y evaluación de la clusterización\n",
    "Clasificación de 2 a 50 clústeres, en cada uno se repite 10 veces la clusterización.\n",
    "- 10, 21, 7, 15 y 9 los #5 clústeres que presentan el mejor s-score\n",
    "- 2, 3 y 4 clústeres presentan variación mínima en el s-score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAtsAAAF/CAYAAACYFQXNAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n404MAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x9ab3b00>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.cluster import KMeans\n",
    "from sklearn import metrics\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "% matplotlib inline\n",
    "\n",
    "x = []\n",
    "y = []\n",
    "yerr = []\n",
    "\n",
    "for n_clusters in range(2, 51, 1):\n",
    "    km = KMeans(\n",
    "        n_clusters=n_clusters,\n",
    "        init='k-means++',  # or 'random'  \n",
    "        n_init=1,  # repetition of the process to choose the best\n",
    "        max_iter=100  # iterations for kmeans\n",
    "    )\n",
    "    \n",
    "    sscore = []\n",
    "    for i in range(1, 10):\n",
    "        km.fit(tfidf)\n",
    "        sscore.append(metrics.silhouette_score(tfidf, km.labels_))\n",
    "    \n",
    "    x.append(n_clusters)\n",
    "    y.append(np.median(sscore))\n",
    "    yerr.append(np.std(sscore))\n",
    "\n",
    "plt.figure(figsize=(12, 6))\n",
    "plt.xlabel('Number of clusters')\n",
    "plt.ylabel('Silhouette score value')\n",
    "plt.errorbar(x, y, yerr=yerr, fmt='-o', ecolor='g', color='g')\n",
    "plt.xlim(xmin=1, xmax=51)\n",
    "plt.ylim(ymin=0, ymax=0.040)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2 \t 0.0190743565792 \t 0.0\n3 \t 0.02324056488 \t 4.16648616608e-06\n4 \t 0.0253310745076 \t 0.000538742678941\n5 \t 0.0263202140122 \t 0.000266222143283\n6 \t 0.026769260356 \t 0.000770307457046\n7 \t 0.0272521022281 \t 0.00148287319871\n8 \t 0.0276697535109 \t 0.00434488611273\n9 \t 0.0271413782831 \t 0.00594472721928\n10 \t 0.026090995065 \t 0.00567452045779\n11 \t 0.024934376114 \t 0.00607792927016\n12 \t 0.026763783359 \t 0.00550448976143\n13 \t 0.0263651671887 \t 0.00567546722879\n14 \t 0.0264476247016 \t 0.00647462361814\n15 \t 0.0257732207956 \t 0.00741273128293\n16 \t 0.025586154174 \t 0.00723777625177\n17 \t 0.0153402085248 \t 0.00586652450744\n18 \t 0.0170175070241 \t 0.00661247505153\n19 \t 0.0149415675895 \t 0.00636827443661\n20 \t 0.0141846043254 \t 0.00559473526154\n21 \t 0.0159078107314 \t 0.00612846024641\n22 \t 0.016926964947 \t 0.00496403958676\n23 \t 0.0155094076308 \t 0.00141726738741\n24 \t 0.0183791131525 \t 0.00551112767074\n25 \t 0.015790708479 \t 0.00561343369359\n26 \t 0.0174976446072 \t 0.00627563955228\n27 \t 0.0162267527856 \t 0.00107189065962\n28 \t 0.0162762077712 \t 0.00562595993422\n29 \t 0.0165521501878 \t 0.00349986540148\n30 \t 0.016054943654 \t 0.00422518220743\n31 \t 0.0173338093732 \t 0.00524250228854\n32 \t 0.0162427256722 \t 0.00567711178993\n33 \t 0.0169111400851 \t 0.00216840585778\n34 \t 0.0179809671563 \t 0.00369800214732\n35 \t 0.0171144319049 \t 0.0014451017946\n36 \t 0.0173305869798 \t 0.00116133342161\n37 \t 0.0177958731611 \t 0.00138244386559\n38 \t 0.0175083023115 \t 0.00568325895553\n39 \t 0.018282621144 \t 0.000587431301276\n40 \t 0.0181215934611 \t 0.000787607322509\n41 \t 0.0185074591439 \t 0.00576550425647\n42 \t 0.0173878646088 \t 0.00125311982731\n43 \t 0.0192225067547 \t 0.00149925674022\n44 \t 0.0187825602238 \t 0.00409313901179\n45 \t 0.019277247095 \t 0.00100703064064\n46 \t 0.0191381943698 \t 0.00131734430517\n47 \t 0.0196570571583 \t 0.00264972184742\n48 \t 0.0189312664059 \t 0.00165403184476\n49 \t 0.0192038588747 \t 0.00126214257126\n50 \t 0.0193937170765 \t 0.00059636859763\n"
     ]
    }
   ],
   "source": [
    "for i, ye in enumerate(yerr):\n",
    "    print(x[i], \"\\t\", y[i], \"\\t\", yerr[i])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Más clusters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAtsAAAF/CAYAAACYFQXNAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x9ab3f60>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.cluster import KMeans\n",
    "from sklearn import metrics\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "% matplotlib inline\n",
    "\n",
    "x = []\n",
    "y = []\n",
    "yerr = []\n",
    "\n",
    "for n_clusters in range(10, 105, 5):\n",
    "    km = KMeans(\n",
    "        n_clusters=n_clusters,\n",
    "        init='k-means++',  # or 'random'  \n",
    "        n_init=1,  # repetition of the process to choose the best\n",
    "        max_iter=100  # iterations for kmeans\n",
    "    )\n",
    "    \n",
    "    sscore = []\n",
    "    for i in range(1, 11):\n",
    "        km.fit(tfidf)\n",
    "        sscore.append(metrics.silhouette_score(tfidf, km.labels_))\n",
    "    \n",
    "    x.append(n_clusters)\n",
    "    y.append(np.median(sscore))\n",
    "    yerr.append(np.std(sscore))\n",
    "\n",
    "plt.figure(figsize=(12, 6))\n",
    "plt.xlabel('Number of clusters')\n",
    "plt.ylabel('Silhouette score value')\n",
    "plt.errorbar(x, y, yerr=yerr, fmt='-o', ecolor='g', color='g')\n",
    "plt.xlim(xmin=5, xmax=105)\n",
    "# plt.ylim(ymin=0, ymax=0.040)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAtsAAAF/CAYAAACYFQXNAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\npIEYbEuSJEkD+RgE54PxITLvuwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x80d8278>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.cluster import KMeans\n",
    "from sklearn import metrics\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "% matplotlib inline\n",
    "\n",
    "x = []\n",
    "y = []\n",
    "yerr = []\n",
    "\n",
    "for n_clusters in range(10, 210, 10):\n",
    "    km = KMeans(\n",
    "        n_clusters=n_clusters,\n",
    "        init='k-means++',  # or 'random'  \n",
    "        n_init=1,  # repetition of the process to choose the best\n",
    "        max_iter=100  # iterations for kmeans\n",
    "    )\n",
    "    \n",
    "    sscore = []\n",
    "    for i in range(1, 11):\n",
    "        km.fit(tfidf)\n",
    "        sscore.append(metrics.silhouette_score(tfidf, km.labels_))\n",
    "    \n",
    "    x.append(n_clusters)\n",
    "    y.append(np.median(sscore))\n",
    "    yerr.append(np.std(sscore))\n",
    "\n",
    "plt.figure(figsize=(12, 6))\n",
    "plt.xlabel('Number of clusters')\n",
    "plt.ylabel('Silhouette score value')\n",
    "plt.errorbar(x, y, yerr=yerr, fmt='-o', ecolor='g', color='g')\n",
    "plt.xlim(xmin=0, xmax=210)\n",
    "# plt.ylim(ymin=0, ymax=0.040)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 8 clústers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Top terms per cluster:\nCluster 0: carson coburn senate specter toomey oklahoma knowles bunning murkowski race\nCluster 1: party house republican democratic elections senate republicans campaign democrats state\nCluster 2: iraq war iraqi military soldiers troops baghdad american killed forces\nCluster 3: dean clark edwards kerry lieberman gephardt primary poll iowa democratic\nCluster 4: november voting account electoral governor vote senate republicans house polls\nCluster 5: bush administration president cheney iraq bushs white war house people\nCluster 6: november account electoral turnout governor duderino parecommend poll polls trouble\nCluster 7: kerry bush poll voters percent general polls campaign president kerrys\n"
     ]
    }
   ],
   "source": [
    "from sklearn.cluster import KMeans\n",
    "\n",
    "n_clusters = 8\n",
    "km = KMeans(\n",
    "    n_clusters=n_clusters,\n",
    "    init='k-means++',  # or 'random'  \n",
    "    n_init=10,  # repetition of the process to choose the best\n",
    "    max_iter=100  # iterations for kmeans\n",
    ")\n",
    "\n",
    "km.fit(tfidf)\n",
    "\n",
    "print(\"Top terms per cluster:\")\n",
    "order_centroids = km.cluster_centers_.argsort()[:, ::-1]\n",
    "for i in range(n_clusters):\n",
    "    print(\"Cluster %d:\" % i, end='')\n",
    "    for ind in order_centroids[i, :10]:\n",
    "        print(' %s' % vectorizer.get_feature_names()[ind], end='')\n",
    "    print()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 7 clústers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Top terms per cluster:\nCluster 0: party senate house republican elections democratic campaign race democrats republicans\nCluster 1: dean clark edwards kerry lieberman gephardt primary poll iowa democratic\nCluster 2: kerry bush poll percent voters polls general results nader polling\nCluster 3: november account electoral turnout governor parecommend duderino poll polls trouble\nCluster 4: bush administration president cheney kerry bushs iraq war campaign white\nCluster 5: november voting account electoral governor vote senate republicans house polls\nCluster 6: iraq war iraqi military bush troops soldiers baghdad american killed\n"
     ]
    }
   ],
   "source": [
    "from sklearn.cluster import KMeans\n",
    "\n",
    "n_clusters = 7\n",
    "km = KMeans(\n",
    "    n_clusters=n_clusters,\n",
    "    init='k-means++',  # or 'random'  \n",
    "    n_init=10,  # repetition of the process to choose the best\n",
    "    max_iter=100  # iterations for kmeans\n",
    ")\n",
    "\n",
    "km.fit(tfidf)\n",
    "\n",
    "print(\"Top terms per cluster:\")\n",
    "order_centroids = km.cluster_centers_.argsort()[:, ::-1]\n",
    "for i in range(n_clusters):\n",
    "    print(\"Cluster %d:\" % i, end='')\n",
    "    for ind in order_centroids[i, :10]:\n",
    "        print(' %s' % vectorizer.get_feature_names()[ind], end='')\n",
    "    print()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 9 clústers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Top terms per cluster:\nCluster 0: million donors money raised campaign candidates fundraising dkos dnc dozen\nCluster 1: november account electoral turnout governor parecommend duderino poll polls trouble\nCluster 2: senate house elections republican race seat district delay party gop\nCluster 3: bush people general party time media republicans campaign president democratic\nCluster 4: kerry bush poll percent voters polls general nader results polling\nCluster 5: dean clark edwards kerry lieberman gephardt primary poll iowa democratic\nCluster 6: november voting account electoral governor vote senate republicans house polls\nCluster 7: iraq bush war administration iraqi military american troops president intelligence\nCluster 8: kerry cheney bush john president campaign kerrys general news debate\n"
     ]
    }
   ],
   "source": [
    "from sklearn.cluster import KMeans\n",
    "\n",
    "n_clusters = 9\n",
    "km = KMeans(\n",
    "    n_clusters=n_clusters,\n",
    "    init='k-means++',  # or 'random'  \n",
    "    n_init=10,  # repetition of the process to choose the best\n",
    "    max_iter=100  # iterations for kmeans\n",
    ")\n",
    "\n",
    "km.fit(tfidf)\n",
    "\n",
    "print(\"Top terms per cluster:\")\n",
    "order_centroids = km.cluster_centers_.argsort()[:, ::-1]\n",
    "for i in range(n_clusters):\n",
    "    print(\"Cluster %d:\" % i, end='')\n",
    "    for ind in order_centroids[i, :10]:\n",
    "        print(' %s' % vectorizer.get_feature_names()[ind], end='')\n",
    "    print()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Dendogram "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "% matplotlib inline\n",
    "\n",
    "from matplotlib import pyplot as plt\n",
    "from sklearn.metrics.pairwise import cosine_similarity\n",
    "from scipy.cluster.hierarchy import ward, dendrogram\n",
    "\n",
    "dist = 1 - cosine_similarity(tfidf)\n",
    "linkage_matrix = ward(dist)\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(8, 30))\n",
    "ax = dendrogram(linkage_matrix, orientation=\"right\")\n",
    "plt.tick_params(\n",
    "    axis='x',  # changes apply to the x-axis\n",
    "    which='both',  # both major and minor ticks are affected\n",
    "    bottom='off',  # ticks along the bottom edge are off\n",
    "    top='off',  # ticks along the top edge are off\n",
    "    labelbottom='off')\n",
    "\n",
    "plt.tight_layout()  #show plot with tight layout"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2.0
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}