{"nbformat_minor": 0, "cells": [{"source": "# scikit-learn Cookbook\n\nThis cookbook contains recipes for some common applications of machine learning. You'll need a working knowledge of [pandas](http://pandas.pydata.org/), [matplotlib](http://matplotlib.org/), [numpy](http://www.numpy.org/), and, of course, [scikit-learn](http://scikit-learn.org/stable/) to benefit from it.", "cell_type": "markdown", "metadata": {}}, {"execution_count": 1, "cell_type": "code", "source": "# <help:cookbook_setup>\n%matplotlib inline", "outputs": [], "metadata": {"collapsed": false, "trusted": true}}, {"source": "##Training with k-Fold Cross-Validation\nThis recipe repeatedly trains a [logistic regression](http://en.wikipedia.org/wiki/Logistic_regression) classifier over different subsets (folds) of sample data. It attempts to match the percentage of each class in every fold to its percentage in the overall dataset ([stratification](http://en.wikipedia.org/wiki/Stratified_sampling)). It evaluates each model against a test set and collects the confusion matrices for each test fold into a `pandas.Panel`.\n\nThis recipe defaults to using the [Iris data set](http://en.wikipedia.org/wiki/Iris_flower_data_set). To use your own data, set `X` to your instance feature vectors, `y` to the instance classes as a factor, and `labels` to the instance classes as human readable names.", "cell_type": "markdown", "metadata": {}}, {"execution_count": 2, "cell_type": "code", "source": "# <help:scikit_cross_validation>\nimport warnings\nwarnings.filterwarnings('ignore') #notebook outputs warnings, let's ignore them\nimport pandas\nimport sklearn\nimport sklearn.datasets\nimport sklearn.metrics as metrics \nfrom sklearn.linear_model import LogisticRegression\nfrom sklearn.cross_validation import StratifiedKFold\n\n# load the iris dataset\ndataset = sklearn.datasets.load_iris()\n\n# define feature vectors (X) and target (y)\nX = dataset.data \ny = dataset.target \nlabels = dataset.target_names \nlabels ", "outputs": [{"execution_count": 2, "output_type": "execute_result", "data": {"text/plain": "array(['setosa', 'versicolor', 'virginica'], \n dtype='|S10')"}, "metadata": {}}], "metadata": {"collapsed": false, "trusted": true}}, {"execution_count": 3, "cell_type": "code", "source": "# <help:scikit_cross_validation>\n# use log reg classifier\nclf = LogisticRegression()\n\ncms = {}\nscores = []\ncv = StratifiedKFold(y, n_folds=10)\nfor i, (train, test) in enumerate(cv):\n # train then immediately predict the test set\n y_pred = clf.fit(X[train], y[train]).predict(X[test])\n # compute the confusion matrix on each fold, convert it to a DataFrame and stash it for later compute\n cms[i] = pandas.DataFrame(metrics.confusion_matrix(y[test], y_pred), columns=labels, index=labels)\n # stash the overall accuracy on the test set for the fold too\n scores.append(metrics.accuracy_score(y[test], y_pred))\n\n# Panel of all test set confusion matrices\npl = pandas.Panel(cms)\ncm = pl.sum(axis=0) #Sum the confusion matrices to get one view of how well the classifiers perform\ncm", "outputs": [{"execution_count": 3, "output_type": "execute_result", "data": {"text/plain": " setosa versicolor virginica\nsetosa 50 0 0\nversicolor 0 45 5\nvirginica 0 1 49", "text/html": "<div style=\"max-height:1000px;max-width:1500px;overflow:auto;\">\n<table border=\"1\" class=\"dataframe\">\n <thead>\n <tr style=\"text-align: right;\">\n <th></th>\n <th>setosa</th>\n <th>versicolor</th>\n <th>virginica</th>\n </tr>\n </thead>\n <tbody>\n <tr>\n <th>setosa</th>\n <td> 50</td>\n <td> 0</td>\n <td> 0</td>\n </tr>\n <tr>\n <th>versicolor</th>\n <td> 0</td>\n <td> 45</td>\n <td> 5</td>\n </tr>\n <tr>\n <th>virginica</th>\n <td> 0</td>\n <td> 1</td>\n <td> 49</td>\n </tr>\n </tbody>\n</table>\n</div>"}, "metadata": {}}], "metadata": {"collapsed": false, "trusted": true}}, {"execution_count": 4, "cell_type": "code", "source": "# <help:scikit_cross_validation>\n# accuracy predicting the test set for each fold\nscores", "outputs": [{"execution_count": 4, "output_type": "execute_result", "data": {"text/plain": "[0.93333333333333335,\n 0.93333333333333335,\n 0.8666666666666667,\n 1.0,\n 1.0,\n 0.93333333333333335,\n 0.93333333333333335,\n 1.0,\n 1.0,\n 1.0]"}, "metadata": {}}], "metadata": {"collapsed": false, "trusted": true}}, {"source": "## Principal Component Analysis Plots\nThis recipe performs a [PCA](http://en.wikipedia.org/wiki/Principal_component_analysis) and plots the data against the first two principal components in a scatter plot. It then prints the [eigenvalues and eigenvectors of the covariance matrix](http://www.quora.com/What-is-an-eigenvector-of-a-covariance-matrix) and finally prints the precentage of total variance explained by each component. \n\nThis recipe defaults to using the [Iris data set](http://en.wikipedia.org/wiki/Iris_flower_data_set). To use your own data, set `X` to your instance feature vectors, `y` to the instance classes as a factor, and `labels` to human-readable names of the classes.", "cell_type": "markdown", "metadata": {}}, {"execution_count": 5, "cell_type": "code", "source": "# <help:scikit_pca>\nimport warnings\nwarnings.filterwarnings('ignore') #notebook outputs warnings, let's ignore them\nfrom __future__ import division\nimport math\nimport pandas as pd\nimport numpy as np\nimport matplotlib.pyplot as plt\nimport sklearn.datasets\nimport sklearn.metrics as metrics\nfrom sklearn.decomposition import PCA\nfrom sklearn.preprocessing import StandardScaler\n\n# load the iris dataset\ndataset = sklearn.datasets.load_iris()\n# define feature vectors (X) and target (y)\nX = dataset.data \ny = dataset.target \nlabels = dataset.target_names ", "outputs": [], "metadata": {"collapsed": false, "trusted": true}}, {"execution_count": 6, "cell_type": "code", "source": "# <help:scikit_pca>\n# define the number of components to compute, recommend n_components < y_features\npca = PCA(n_components=2) \nX_pca = pca.fit_transform(X)\n\n# plot the first two principal components\nfig, ax = plt.subplots()\nplt.scatter(X_pca[:,0], X_pca[:,1])\nplt.grid()\nplt.title('PCA of the dataset')\nax.set_xlabel('Component #1') \nax.set_ylabel('Component #2')\nplt.show()", "outputs": [{"output_type": "display_data", "data": {"image/png": "iVBORw0KGgoAAAANSUhEUgAAAYkAAAEZCAYAAABiu9n+AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ncMoIdtMS+BwAAAAASUVORK5CYII=\n", "text/plain": "<matplotlib.figure.Figure at 0x7f5ed3f98110>"}, "metadata": {}}], "metadata": {"collapsed": false, "trusted": true}}, {"execution_count": 7, "cell_type": "code", "source": "# <help:scikit_pca>\n# eigendecomposition on the covariance matrix\ncov_mat = np.cov(X_pca.T)\neig_vals, eig_vecs = np.linalg.eig(cov_mat)\nprint('Eigenvectors \\n%s' %eig_vecs)\nprint('\\nEigenvalues \\n%s' %eig_vals)", "outputs": [{"output_type": "stream", "name": "stdout", "text": "Eigenvectors \n[[ 1.00000000e+00 2.27318015e-17]\n [ 0.00000000e+00 1.00000000e+00]]\n\nEigenvalues \n[ 4.22484077 0.24224357]\n"}], "metadata": {"collapsed": false, "trusted": true}}, {"execution_count": 8, "cell_type": "code", "source": "# <help:scikit_pca>\n# prints the percentage of overall variance explained by each component\nprint(pca.explained_variance_ratio_)", "outputs": [{"output_type": "stream", "name": "stdout", "text": "[ 0.92461621 0.05301557]\n"}], "metadata": {"collapsed": false, "trusted": true}}, {"source": "## K-Means Clustering Plots\n\nThis recipe performs a [K-means clustering](http://en.wikipedia.org/wiki/K-means_clustering) `k=1..n` times. It prints and plots the the within-clusters sum of squares error for each `k` (i.e., inertia) as an indicator of what value of `k` might be appropriate for the given dataset.\n\nThis recipe defaults to using the [Iris data set](http://en.wikipedia.org/wiki/Iris_flower_data_set). To use your own data, set `X` to your instance feature vectors, `y` to the instance classes as a factor, and `labels` to human-readable names of the classes. To change the number of clusters, modify `k`.", "cell_type": "markdown", "metadata": {}}, {"execution_count": 9, "cell_type": "code", "source": "# <help:scikit_k_means_cluster>\nimport warnings\nwarnings.filterwarnings('ignore') #notebook outputs warnings, let's ignore them\nfrom time import time\nimport numpy as np\nimport matplotlib.pyplot as plt\nimport sklearn.datasets\nfrom sklearn.cluster import KMeans\n\n# load datasets and assign data and features\ndataset = sklearn.datasets.load_iris()\n# define feature vectors (X) and target (y)\nX = dataset.data\ny = dataset.target\n\n# set the number of clusters, must be >=1\nn = 6\ninertia = [np.NaN]\n\n# perform k-means clustering over i=0...k\nfor k in range(1,n):\n k_means_ = KMeans(n_clusters=k)\n k_means_.fit(X)\n print('k = %d, inertia= %f' % (k, k_means_.inertia_ ))\n inertia.append(k_means_.inertia_) \n \n# plot the SSE of the clusters for each value of i\nax = plt.subplot(111)\nax.plot(inertia, '-o')\nplt.xticks(range(n))\nplt.title(\"Inertia\")\nax.set_ylabel('Inertia')\nax.set_xlabel('# Clusters')\nplt.show() ", "outputs": [{"output_type": "stream", "name": "stdout", "text": "k = 1, inertia= 680.824400\nk = 2, inertia= 152.368706\nk = 3, inertia= 78.940841\nk = 4, inertia= 57.317873\nk = 5, inertia= 46.535582\n"}, {"output_type": "display_data", "data": {"image/png": "iVBORw0KGgoAAAANSUhEUgAAAYQAAAEZCAYAAACXRVJOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nXLcAAAAASUVORK5CYII=\n", "text/plain": "<matplotlib.figure.Figure at 0x7f5ed3edcc10>"}, "metadata": {}}], "metadata": {"collapsed": false, "trusted": true}}, {"source": "## SVM Classifier Hyperparameter Tuning with Grid Search\n\nThis recipe performs a [grid search](http://en.wikipedia.org/wiki/Hyperparameter_optimization) for the best settings for a [support vector machine,](http://en.wikipedia.org/wiki/Support_vector_machine) predicting the class of each flower in the dataset. It splits the dataset into training and test instances once.\n\nThis recipe defaults to using the [Iris data set](http://en.wikipedia.org/wiki/Iris_flower_data_set). To use your own data, set `X` to your instance feature vectors, `y` to the instance classes as a factor, and `labels` to human-readable names of the classes. Modify `parameters` to change the grid search space or the `scoring='accuracy'` value to optimize a different metric for the classifier (e.g., precision, recall).", "cell_type": "markdown", "metadata": {}}, {"execution_count": 10, "cell_type": "code", "source": "#<help_scikit_grid_search>\nimport numpy as np\nimport matplotlib.pyplot as plt\nimport sklearn.datasets\nimport sklearn.metrics as metrics\nfrom sklearn.svm import SVC\nfrom sklearn.grid_search import GridSearchCV\nfrom sklearn.metrics import classification_report\nfrom sklearn.cross_validation import train_test_split\nfrom sklearn.preprocessing import label_binarize\n\n# load datasets and features\ndataset = sklearn.datasets.load_iris()\n# define feature vectors (X) and target (y)\nX = dataset.data\ny = dataset.target\nlabels = dataset.target_names\n\n# separate datasets into training and test datasets once, no folding\nX_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)", "outputs": [], "metadata": {"collapsed": false, "trusted": true}}, {"execution_count": 11, "cell_type": "code", "source": "#<help_scikit_grid_search>\n#define the parameter dictionary with the kernels of SVCs\nparameters = [\n {'kernel': ['rbf'], 'gamma': [1e-3, 1e-4, 1e-2], 'C': [1, 10, 100, 1000]},\n {'kernel': ['linear'], 'C': [1, 10, 100, 1000]},\n {'kernel': ['poly'], 'degree': [1, 3, 5], 'C': [1, 10, 100, 1000]}\n]\n\n# find the best parameters to optimize accuracy\nsvc_clf = SVC(C=1, probability= True)\nclf = GridSearchCV(svc_clf, parameters, cv=5, scoring='accuracy') #5 folds\nclf.fit(X_train, y_train) #train the model \nprint(\"Best parameters found from SVM's:\")\nprint clf.best_params_ \nprint(\"Best score found from SVM's:\") \nprint clf.best_score_", "outputs": [{"output_type": "stream", "name": "stdout", "text": "Best parameters found from SVM's:\n{'kernel': 'linear', 'C': 1}\nBest score found from SVM's:\n0.983333333333\n"}], "metadata": {"collapsed": false, "trusted": true}}, {"source": "## Plot ROC Curves\nThis recipe plots the [reciever operating characteristic (ROC) curve](http://en.wikipedia.org/wiki/Receiver_operating_characteristic) for a [SVM classifier](http://en.wikipedia.org/wiki/Support_vector_machine) trained over the given dataset.\n\nThis recipe defaults to using the [Iris data set](http://en.wikipedia.org/wiki/Iris_flower_data_set) which has three classes. The recipe uses a [one-vs-the-rest strategy](http://scikit-learn.org/stable/modules/multiclass.html#one-vs-the-rest) to create the [binary classifications](http://en.wikipedia.org/wiki/Binary_classification) appropriate for ROC plotting. To use your own data, set `X` to your instance feature vectors, `y` to the instance classes as a factor, and `labels` to human-readable names of the classes.\n\nNote that the recipe adds noise to the iris features to make the ROC plots more realistic. Otherwise, the classification is nearly perfect and the plot hard to study. **Remove the noise generator if you use your own data!**", "cell_type": "markdown", "metadata": {}}, {"execution_count": 12, "cell_type": "code", "source": "# <help:scikit_roc>\nimport warnings\nwarnings.filterwarnings('ignore') #notebook outputs warnings, let's ignore them\nimport numpy as np\nimport matplotlib.pyplot as plt\nimport sklearn.datasets\nimport sklearn.metrics as metrics\nfrom sklearn.svm import SVC\nfrom sklearn.multiclass import OneVsRestClassifier\nfrom sklearn.cross_validation import train_test_split\nfrom sklearn.preprocessing import label_binarize\n\n# load iris, set and data\ndataset = sklearn.datasets.load_iris()\nX = dataset.data\n# binarize the output for binary classification\ny = label_binarize(dataset.target, classes=[0, 1, 2])\nlabels = dataset.target_names", "outputs": [], "metadata": {"collapsed": false, "trusted": true}}, {"execution_count": 13, "cell_type": "code", "source": "# <help:scikit_roc>\n# add noise to the features so the plot is less ideal\n# REMOVE ME if you use your own dataset!\nrandom_state = np.random.RandomState(0)\nn_samples, n_features = X.shape\nX = np.c_[X, random_state.randn(n_samples, 200 * n_features)]", "outputs": [], "metadata": {"collapsed": false, "trusted": true}}, {"execution_count": 14, "cell_type": "code", "source": "# <help:scikit_roc>\n# split data for cross-validation\nX_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3)\n\n# classify instances into more than two classes, one vs rest\n# add param to create probabilities to determine Y or N as the classification\nclf = OneVsRestClassifier(SVC(kernel='linear', probability=True))\n\n# fit estiamators and return the distance of each sample from the decision boundary\ny_score = clf.fit(X_train, y_train).decision_function(X_test)", "outputs": [], "metadata": {"collapsed": false, "trusted": true}}, {"execution_count": 15, "cell_type": "code", "source": "# <help:scikit_roc>\n# plot the ROC curve, best for it to be in top left corner\nplt.figure(figsize=(10,5))\nplt.plot([0, 1], [0, 1], 'k--') # add a straight line representing a random model \nfor i, label in enumerate(labels):\n # false positive and true positive rate for each class\n fpr, tpr, _ = metrics.roc_curve(y_test[:, i], y_score[:, i])\n # area under the curve (auc) for each class\n roc_auc = metrics.auc(fpr, tpr)\n plt.plot(fpr, tpr, label='ROC curve of {0} (area = {1:0.2f})'.format(label, roc_auc))\nplt.xlim([0.0, 1.0])\nplt.ylim([0.0, 1.05])\nplt.title('Receiver Operating Characteristic for Iris data set')\nplt.xlabel('False Positive Rate') # 1- specificity\nplt.ylabel('True Positive Rate') # sensitivity\nplt.legend(loc=\"lower right\")\nplt.show()", "outputs": [{"output_type": "display_data", "data": {"image/png": "iVBORw0KGgoAAAANSUhEUgAAAmUAAAFRCAYAAAA1jNoBAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n", "text/plain": "<matplotlib.figure.Figure at 0x7f5ed29b6e90>"}, "metadata": {}}], "metadata": {"collapsed": false, "trusted": true}}, {"source": "## Build a Transformation and Classification Pipeline \n\nThis recipe builds a [transformation and training pipeline](http://scikit-learn.org/stable/modules/pipeline.html) for a model that can classify a snippet of text as belonging to one of 20 [USENET](http://en.wikipedia.org/wiki/Usenet) [newgroups](http://en.wikipedia.org/wiki/Usenet_newsgroup). It then prints the [precision, recall, and F1-score](http://en.wikipedia.org/wiki/Precision_and_recall) for predictions over a held-out test set as well as the confusion matrix.\n\nThis recipe defaults to using the [20 USENET newsgroup](http://kdd.ics.uci.edu/databases/20newsgroups/20newsgroups.html) dataset. To use your own data, set `X` to your instance feature vectors, `y` to the instance classes as a factor, and `labels` to human-readable names of the classes. Then modify the pipeline components to perform appropriate transformations for your data.\n\n<div class=\"alert alert-block alert-warning\" style=\"margin-top: 20px\">**Warning:** Running this recipe with the sample data may consume a significant amount of memory.</div>", "cell_type": "markdown", "metadata": {}}, {"execution_count": 16, "cell_type": "code", "source": "# <help:scikit_pipeline>\nimport pandas\nimport sklearn.metrics as metrics\nfrom sklearn.feature_extraction.text import CountVectorizer, TfidfTransformer\nfrom sklearn.feature_extraction.text import HashingVectorizer\nfrom sklearn.linear_model import Perceptron\nfrom sklearn.naive_bayes import MultinomialNB\nfrom sklearn.linear_model import SGDClassifier\nfrom sklearn.pipeline import Pipeline\nfrom sklearn.cross_validation import train_test_split\nfrom sklearn.datasets import fetch_20newsgroups\n\n# download the newsgroup dataset\ndataset = fetch_20newsgroups('all')\n\n# define feature vectors (X) and target (y) \nX = dataset.data\ny = dataset.target\nlabels = dataset.target_names\nlabels", "outputs": [{"execution_count": 16, "output_type": "execute_result", "data": {"text/plain": "['alt.atheism',\n 'comp.graphics',\n 'comp.os.ms-windows.misc',\n 'comp.sys.ibm.pc.hardware',\n 'comp.sys.mac.hardware',\n 'comp.windows.x',\n 'misc.forsale',\n 'rec.autos',\n 'rec.motorcycles',\n 'rec.sport.baseball',\n 'rec.sport.hockey',\n 'sci.crypt',\n 'sci.electronics',\n 'sci.med',\n 'sci.space',\n 'soc.religion.christian',\n 'talk.politics.guns',\n 'talk.politics.mideast',\n 'talk.politics.misc',\n 'talk.religion.misc']"}, "metadata": {}}], "metadata": {"collapsed": false, "trusted": true}}, {"execution_count": 17, "cell_type": "code", "source": "# <help:scikit_pipeline>\n# split data holding out 30% for testing the classifier\nX_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=0)\n\n# pipelines concatenate functions serially, output of 1 becomes input of 2\nclf = Pipeline([\n ('vect', HashingVectorizer(analyzer='word', ngram_range=(1,3))), # count frequency of words, using hashing trick\n ('tfidf', TfidfTransformer()), # transform counts to tf-idf values,\n ('clf', SGDClassifier(loss='hinge', penalty='l2', alpha=1e-3, n_iter=5))\n])", "outputs": [], "metadata": {"collapsed": false, "trusted": true}}, {"execution_count": 18, "cell_type": "code", "source": "# <help:scikit_pipeline>\n# train the model and predict the test set\ny_pred = clf.fit(X_train, y_train).predict(X_test)\n\n# standard information retrieval metrics\nprint metrics.classification_report(y_test, y_pred, target_names=labels)", "outputs": [{"output_type": "stream", "name": "stdout", "text": " precision recall f1-score support\n\n alt.atheism 0.86 0.95 0.90 137\n comp.graphics 0.89 0.84 0.86 178\n comp.os.ms-windows.misc 0.85 0.90 0.88 168\ncomp.sys.ibm.pc.hardware 0.85 0.78 0.81 180\n comp.sys.mac.hardware 0.95 0.83 0.89 174\n comp.windows.x 0.88 0.91 0.90 193\n misc.forsale 0.78 0.92 0.84 157\n rec.autos 0.93 0.89 0.91 188\n rec.motorcycles 0.96 0.95 0.95 185\n rec.sport.baseball 0.94 0.95 0.95 183\n rec.sport.hockey 0.92 0.98 0.95 180\n sci.crypt 0.95 0.97 0.96 181\n sci.electronics 0.93 0.80 0.86 180\n sci.med 0.95 0.94 0.95 159\n sci.space 0.91 0.98 0.95 185\n soc.religion.christian 0.88 0.92 0.90 176\n talk.politics.guns 0.93 0.98 0.95 176\n talk.politics.mideast 0.92 0.98 0.95 180\n talk.politics.misc 0.97 0.91 0.94 127\n talk.religion.misc 0.99 0.69 0.81 108\n\n avg / total 0.91 0.91 0.91 3395\n\n"}], "metadata": {"collapsed": false, "trusted": true}}, {"execution_count": 19, "cell_type": "code", "source": "# <help:scikit_pipeline>\n# show the confusion matrix in a labeled dataframe for ease of viewing\nindex_labels = ['{} {}'.format(i, l) for i, l in enumerate(labels)]\npandas.DataFrame(metrics.confusion_matrix(y_test,y_pred), index=index_labels)", "outputs": [{"execution_count": 19, "output_type": "execute_result", "data": {"text/plain": " 0 1 2 3 4 5 6 7 8 9 \\\n0 alt.atheism 130 0 0 0 0 0 0 0 0 0 \n1 comp.graphics 1 150 5 4 1 7 3 0 0 0 \n2 comp.os.ms-windows.misc 0 1 152 3 0 6 2 0 0 0 \n3 comp.sys.ibm.pc.hardware 0 3 10 140 4 3 7 1 0 1 \n4 comp.sys.mac.hardware 1 2 2 4 145 4 4 0 1 3 \n5 comp.windows.x 0 5 5 2 0 176 1 1 0 0 \n6 misc.forsale 0 2 0 3 0 0 144 2 0 1 \n7 rec.autos 1 1 0 0 1 1 5 167 3 3 \n8 rec.motorcycles 2 0 0 0 0 0 4 2 175 0 \n9 rec.sport.baseball 0 1 0 0 0 0 3 1 1 174 \n10 rec.sport.hockey 0 1 0 0 0 0 1 0 0 1 \n11 sci.crypt 0 1 1 0 0 1 0 0 0 0 \n12 sci.electronics 2 1 3 8 1 2 5 3 0 1 \n13 sci.med 0 0 0 0 0 0 1 1 0 0 \n14 sci.space 0 0 0 0 0 0 0 0 0 0 \n15 soc.religion.christian 2 1 0 1 0 0 4 1 0 0 \n16 talk.politics.guns 1 0 1 0 0 0 0 0 0 1 \n17 talk.politics.mideast 0 0 0 0 0 0 0 1 0 0 \n18 talk.politics.misc 1 0 0 0 0 0 0 0 2 0 \n19 talk.religion.misc 11 0 0 0 0 0 0 0 0 0 \n\n 10 11 12 13 14 15 16 17 18 19 \n0 alt.atheism 0 0 0 0 0 3 1 2 0 1 \n1 comp.graphics 2 1 1 1 1 0 1 0 0 0 \n2 comp.os.ms-windows.misc 1 0 1 0 1 1 0 0 0 0 \n3 comp.sys.ibm.pc.hardware 2 3 2 0 3 1 0 0 0 0 \n4 comp.sys.mac.hardware 0 1 2 2 1 0 0 2 0 0 \n5 comp.windows.x 1 1 0 0 1 0 0 0 0 0 \n6 misc.forsale 2 1 1 0 0 1 0 0 0 0 \n7 rec.autos 1 0 2 0 1 0 1 0 1 0 \n8 rec.motorcycles 1 0 0 0 1 0 0 0 0 0 \n9 rec.sport.baseball 2 0 0 1 0 0 0 0 0 0 \n10 rec.sport.hockey 177 0 0 0 0 0 0 0 0 0 \n11 sci.crypt 0 176 1 0 0 0 0 1 0 0 \n12 sci.electronics 1 2 144 1 5 0 0 0 1 0 \n13 sci.med 0 1 0 150 2 1 2 0 1 0 \n14 sci.space 1 0 1 0 182 0 0 1 0 0 \n15 soc.religion.christian 0 0 0 2 0 162 0 3 0 0 \n16 talk.politics.guns 0 0 0 0 1 0 172 0 0 0 \n17 talk.politics.mideast 1 0 0 0 1 0 0 177 0 0 \n18 talk.politics.misc 0 0 0 0 0 1 4 4 115 0 \n19 talk.religion.misc 0 0 0 1 0 15 4 3 0 74 ", "text/html": "<div style=\"max-height:1000px;max-width:1500px;overflow:auto;\">\n<table border=\"1\" class=\"dataframe\">\n <thead>\n <tr style=\"text-align: right;\">\n <th></th>\n <th>0</th>\n <th>1</th>\n <th>2</th>\n <th>3</th>\n <th>4</th>\n <th>5</th>\n <th>6</th>\n <th>7</th>\n <th>8</th>\n <th>9</th>\n <th>10</th>\n <th>11</th>\n <th>12</th>\n <th>13</th>\n <th>14</th>\n <th>15</th>\n <th>16</th>\n <th>17</th>\n <th>18</th>\n <th>19</th>\n </tr>\n </thead>\n <tbody>\n <tr>\n <th>0 alt.atheism</th>\n <td> 130</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 3</td>\n <td> 1</td>\n <td> 2</td>\n <td> 0</td>\n <td> 1</td>\n </tr>\n <tr>\n <th>1 comp.graphics</th>\n <td> 1</td>\n <td> 150</td>\n <td> 5</td>\n <td> 4</td>\n <td> 1</td>\n <td> 7</td>\n <td> 3</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 2</td>\n <td> 1</td>\n <td> 1</td>\n <td> 1</td>\n <td> 1</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n </tr>\n <tr>\n <th>2 comp.os.ms-windows.misc</th>\n <td> 0</td>\n <td> 1</td>\n <td> 152</td>\n <td> 3</td>\n <td> 0</td>\n <td> 6</td>\n <td> 2</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 1</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n </tr>\n <tr>\n <th>3 comp.sys.ibm.pc.hardware</th>\n <td> 0</td>\n <td> 3</td>\n <td> 10</td>\n <td> 140</td>\n <td> 4</td>\n <td> 3</td>\n <td> 7</td>\n <td> 1</td>\n <td> 0</td>\n <td> 1</td>\n <td> 2</td>\n <td> 3</td>\n <td> 2</td>\n <td> 0</td>\n <td> 3</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n </tr>\n <tr>\n <th>4 comp.sys.mac.hardware</th>\n <td> 1</td>\n <td> 2</td>\n <td> 2</td>\n <td> 4</td>\n <td> 145</td>\n <td> 4</td>\n <td> 4</td>\n <td> 0</td>\n <td> 1</td>\n <td> 3</td>\n <td> 0</td>\n <td> 1</td>\n <td> 2</td>\n <td> 2</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 2</td>\n <td> 0</td>\n <td> 0</td>\n </tr>\n <tr>\n <th>5 comp.windows.x</th>\n <td> 0</td>\n <td> 5</td>\n <td> 5</td>\n <td> 2</td>\n <td> 0</td>\n <td> 176</td>\n <td> 1</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n </tr>\n <tr>\n <th>6 misc.forsale</th>\n <td> 0</td>\n <td> 2</td>\n <td> 0</td>\n <td> 3</td>\n <td> 0</td>\n <td> 0</td>\n <td> 144</td>\n <td> 2</td>\n <td> 0</td>\n <td> 1</td>\n <td> 2</td>\n <td> 1</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n </tr>\n <tr>\n <th>7 rec.autos</th>\n <td> 1</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 1</td>\n <td> 5</td>\n <td> 167</td>\n <td> 3</td>\n <td> 3</td>\n <td> 1</td>\n <td> 0</td>\n <td> 2</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n </tr>\n <tr>\n <th>8 rec.motorcycles</th>\n <td> 2</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 4</td>\n <td> 2</td>\n <td> 175</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n </tr>\n <tr>\n <th>9 rec.sport.baseball</th>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 3</td>\n <td> 1</td>\n <td> 1</td>\n <td> 174</td>\n <td> 2</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n </tr>\n <tr>\n <th>10 rec.sport.hockey</th>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 177</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n </tr>\n <tr>\n <th>11 sci.crypt</th>\n <td> 0</td>\n <td> 1</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 176</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n </tr>\n <tr>\n <th>12 sci.electronics</th>\n <td> 2</td>\n <td> 1</td>\n <td> 3</td>\n <td> 8</td>\n <td> 1</td>\n <td> 2</td>\n <td> 5</td>\n <td> 3</td>\n <td> 0</td>\n <td> 1</td>\n <td> 1</td>\n <td> 2</td>\n <td> 144</td>\n <td> 1</td>\n <td> 5</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n </tr>\n <tr>\n <th>13 sci.med</th>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 150</td>\n <td> 2</td>\n <td> 1</td>\n <td> 2</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n </tr>\n <tr>\n <th>14 sci.space</th>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 182</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n </tr>\n <tr>\n <th>15 soc.religion.christian</th>\n <td> 2</td>\n <td> 1</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 4</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 2</td>\n <td> 0</td>\n <td> 162</td>\n <td> 0</td>\n <td> 3</td>\n <td> 0</td>\n <td> 0</td>\n </tr>\n <tr>\n <th>16 talk.politics.guns</th>\n <td> 1</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 172</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n </tr>\n <tr>\n <th>17 talk.politics.mideast</th>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 177</td>\n <td> 0</td>\n <td> 0</td>\n </tr>\n <tr>\n <th>18 talk.politics.misc</th>\n <td> 1</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 2</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 4</td>\n <td> 4</td>\n <td> 115</td>\n <td> 0</td>\n </tr>\n <tr>\n <th>19 talk.religion.misc</th>\n <td> 11</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 0</td>\n <td> 1</td>\n <td> 0</td>\n <td> 15</td>\n <td> 4</td>\n <td> 3</td>\n <td> 0</td>\n <td> 74</td>\n </tr>\n </tbody>\n</table>\n</div>"}, "metadata": {}}], "metadata": {"collapsed": false, "trusted": true}}, {"source": "<div class=\"alert\">\n<div class=\"row\">\n <div class=\"col-sm-1\"><img src=\"https://knowledgeanyhow.org/static/images/favicon_32x32.png\" style=\"margin-top: -6px\"/></div>\n <div class=\"col-sm-11\">This notebook was created using <a href=\"https://knowledgeanyhow.org\">IBM Knowledge Anyhow Workbench</a>. To learn more, visit us at <a href=\"https://knowledgeanyhow.org\">https://knowledgeanyhow.org</a>.</div>\n </div>\n</div>", "cell_type": "markdown", "metadata": {}}], "nbformat": 4, "metadata": {"kernelspec": {"display_name": "Python 2", "name": "python2", "language": "python"}, "language_info": {"mimetype": "text/x-python", "nbconvert_exporter": "python", "version": "2.7.6", "name": "python", "file_extension": ".py", "pygments_lexer": "ipython2", "codemirror_mode": {"version": 2, "name": "ipython"}}}}