bbc-tf-idf-vectorizer.ipynb

{"cells":[{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"cell_type":"code","source":"# This Python 3 environment comes with many helpful analytics libraries installed\n# It is defined by the kaggle/python docker image: https://github.com/kaggle/docker-python\n# For example, here's several helpful packages to load in \n\nimport numpy as np # linear algebra\nimport pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\nfrom sklearn.feature_extraction.text import TfidfVectorizer\nimport matplotlib\n\n# Input data files are available in the \"../input/\" directory.\n# For example, running this (by clicking run or pressing Shift+Enter) will list the files in the input directory\n\nimport os\ndata = pd.read_csv(\"../input/ai-academy-intermediate-class-competition-1/BBC News Train.csv\")\ndata = data[[\"Text\", \"Category\"]]\ndata\n# Any results you write to the current directory are saved as output.","execution_count":null,"outputs":[]},{"metadata":{"_cell_guid":"79c7e3d0-c299-4dcb-8224-4455121ee9b0","_uuid":"d629ff2d2480ee46fbb7e2d37f6b5fab8052498a","trusted":true},"cell_type":"code","source":"vectorizer = TfidfVectorizer(sublinear_tf=True, min_df=5, norm='l2', encoding='latin-1', ngram_range=(1, 2), stop_words='english')\nX = vectorizer.fit_transform(data[\"Text\"])\nprint(len(vectorizer.get_feature_names()))\nprint(X.shape)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"9ec3c56b4adf5666d95eadd9d3285173a3dbf05d"},"cell_type":"code","source":"data[\"category_id\"]=data[\"Category\"].factorize()[0]","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"fbcc58e981769b0197cbcc7284b28eee572e1da2"},"cell_type":"code","source":"data = data[[\"Text\", \"category_id\", \"Category\"]]\ndata\ncategory_id_data = data[['Category', 'category_id']].drop_duplicates().sort_values('category_id')\ncategory_to_id = dict(category_id_data.values)\nid_to_category = dict(category_id_data[['category_id', 'Category']].values)\ncategory_id_data\nid_to_category","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"02a98e4445eab97c33cfa6fa4f17133353c8180b"},"cell_type":"code","source":"data.sample(5, random_state=0)\ndata.groupby('Category').category_id.count()\ndata.groupby('Category').category_id.count().plot.bar(ylim=0)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"0243d143d0710900811505282759a334a8b6b1f3"},"cell_type":"code","source":"tfidf = TfidfVectorizer(sublinear_tf=True, min_df=5, norm='l2', encoding='latin-1', ngram_range=(1, 2), stop_words='english')\nfeatures = tfidf.fit_transform(data.Text).toarray()\nprint(features)\nlabels = data.category_id\nprint(labels)\nfeatures.shape","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"d739d24292d74221a1d4cc8a4abebb7225bce2fd"},"cell_type":"code","source":"from sklearn.feature_selection import chi2\n\nN = 5\nfor category, category_id in sorted(category_to_id.items()):\n  features_chi2 = chi2(features, labels == category_id)\n  indices = np.argsort(features_chi2[0])\n  feature_names = np.array(tfidf.get_feature_names())[indices]\n  unigrams = [v for v in feature_names if len(v.split(' ')) == 1]\n  bigrams = [v for v in feature_names if len(v.split(' ')) == 2]\n  print(\"# '{}':\".format(category))\n  print(\"  . Most correlated unigrams:\\n       . {}\".format('\\n       . '.join(unigrams[-N:])))\n  print(\"  . Most correlated bigrams:\\n       . {}\".format('\\n       . '.join(bigrams[-N:])))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"3c1146ccb0537fc2989c2625dbbd07616c3e70b4"},"cell_type":"code","source":"from sklearn.manifold import TSNE\nimport matplotlib as plt\nSAMPLE_SIZE = int(len(features) * 0.3)\nnp.random.seed(0)\nindices = np.random.choice(range(len(features)), size=SAMPLE_SIZE, replace=False)\nprojected_features = TSNE(n_components=2, random_state=0).fit_transform(features[indices])\ncolors = ['pink', 'green', 'midnightblue', 'orange', 'darkgrey']\nfor category, category_id in sorted(category_to_id.items()):\n    points = projected_features[(labels[indices] == category_id).values]\n    plt.pyplot.scatter(points[:, 0], points[:, 1], s=30, c=colors[category_id], label=category)\nplt.pyplot.title(\"tf-idf feature vector for each article, projected on 2 dimensions.\",\n          fontdict=dict(fontsize=15))\nplt.pyplot.legend()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"f5e41224f5abb4dc2f85907af6346651f5320838"},"cell_type":"code","source":"from sklearn.linear_model import LogisticRegression\nfrom sklearn.ensemble import RandomForestClassifier\nfrom sklearn.naive_bayes import MultinomialNB\n\nfrom sklearn.model_selection import cross_val_score\n\n\nmodels = [\n    RandomForestClassifier(n_estimators=500, max_depth=4),\n    MultinomialNB(),\n    LogisticRegression(random_state=4),\n]\nCV = 5\ncv_df = pd.DataFrame(index=range(CV * len(models)))\nprint(cv_df)\nentries = []\nfor model in models:\n  model_name = model.__class__.__name__\n  accuracies = cross_val_score(model, features, labels, scoring='accuracy', cv=CV)\n  print(accuracies)\n  for fold_idx, accuracy in enumerate(accuracies):\n    entries.append((model_name, fold_idx, accuracy))\nprint(entries)\ncv_df = pd.DataFrame(entries, columns=['model_name', 'fold_idx', 'accuracy'])\nprint(cv_df)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"897a619ee0e8712b804cbc29e2f424ef8c2b0b5f"},"cell_type":"code","source":"from sklearn.model_selection import train_test_split\nfrom sklearn.metrics import accuracy_score\nX_train, X_test, y_train, y_test, indices_train, indices_test = train_test_split(features, labels, data.index, test_size=0.20, random_state=0)\nfor model in models:\n    model.fit(X_train, y_train)\n    y_pred_proba = model.predict_proba(X_test)\n    y_pred = model.predict(X_test)\n    accuracy = accuracy_score(y_test, y_pred)\n    print(accuracy)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"b1c0a67a80463603b67880b97ecf9bbd47ea367a"},"cell_type":"code","source":"model = models[2]\nmodel.fit(features, labels)\nmodel.coef_","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"331a4edeba86ab4b4a48611b27c9287de25186c9"},"cell_type":"code","source":"test_data = pd.read_csv(\"../input/bbc-test-3/BBC News Test.csv\")\ntest_data","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"1a3f902da5ff0df24bbcb412121d1c4f270bbc41"},"cell_type":"code","source":"test_data.Text.tolist()\ntest_features = tfidf.transform(test_data.Text.tolist())\nY_pred = model.predict(test_features)\nY_pred\nsubmission = []\nfor pred in Y_pred:\n    submission.append(id_to_category[pred])\nsubmission","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"fe47cae10529036b6beb51e04bebf72a64c88e6a"},"cell_type":"code","source":"submission = pd.DataFrame({\n    \"ArticleId\": test_data[\"ArticleId\"],\n    \"Category\": submission\n})\nsubmission","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"893b9181bf8ec3497dde80c28511737d99deaa78"},"cell_type":"code","source":"submission.to_csv('submission.csv', index=False)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"64d05207e3ae1383dc654ca5bc88d46e168bece5"},"cell_type":"code","source":"ls","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_uuid":"f02ea0a31941b6fa5f12dd6e0fc3d7d834020c24"},"cell_type":"code","source":"","execution_count":null,"outputs":[]}],"metadata":{"kernelspec":{"display_name":"Python 3","language":"python","name":"python3"},"language_info":{"name":"python","version":"3.6.6","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"}},"nbformat":4,"nbformat_minor":1}