{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"<h1>Стэкинг</h1>\n\n**Стекинг** — обучение нескольких разных алгоритмов и передача их результатов на вход последнему, который называют мета-алгоритмом. Мета-алгоритм объединяет результаты работы других алгоритмов и принимает итоговое решение.\n\nКлючевые особенности стекинга:\n\n * использование различных базовых алгоритмов;\n * наличие мета-алгоритма.\n \nФинальное решение мета-алгоритм принимает, опираясь на результаты базовых алгоритмов - метапризнаки.\n\nМетапризнаки — это отдельные ответы каждого базового алгоритма. Они передаются в качестве признаков для мета-алгоритма, на которых он и обучается. Метапризнаки называются так, потому что описывают не сами объекты, а признаки этих объектов.\n\nОбучающую выборку делим на две части. На одной из них обучаем базовые алгоритмы, затем их ответы обучаем на второй части и на тестовой выборке. Только потом на полученных метапризнаках обучают мета-алгоритм и предсказывают итоговое значение для тестовых данных.","metadata":{}},{"cell_type":"markdown","source":"![image.png](attachment:7d79411d-784a-4714-b451-602f2d665525.png)","metadata":{},"attachments":{"7d79411d-784a-4714-b451-602f2d665525.png":{"image/png":"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"}}},{"cell_type":"markdown","source":"Технологию стэкинга опробуем на учебном соревновании Kaggle.com\nhttps://www.kaggle.com/c/ai-academy-dota-2-pro-gamer\n\nЗадача — построить алгоритм, который сможет определить опытность игрока по статистике из одного матча. Даны обучающие (train) данные для построения моделей и тестовые (test) данные для проверки алгоритмов. Решением задачи являются предсказания алгоритмов на тестовых данных.","metadata":{}},{"cell_type":"code","source":"# Импортируем необходимые библиотеки\nfrom sklearn.discriminant_analysis import LinearDiscriminantAnalysis  \nfrom sklearn.neighbors import KNeighborsClassifier  \nfrom sklearn.naive_bayes import GaussianNB\nfrom sklearn.linear_model import LogisticRegression, LinearRegression\nfrom sklearn.ensemble import RandomForestClassifier, StackingClassifier\nfrom sklearn.svm import LinearSVC\nfrom sklearn.svm import LinearSVR\nfrom xgboost import XGBRegressor, XGBRFClassifier\nfrom lightgbm import LGBMClassifier, LGBMRegressor\n\nimport numpy as np \nimport pandas as pd \nfrom sklearn.pipeline import make_pipeline\nfrom sklearn.preprocessing import StandardScaler\nfrom sklearn.preprocessing import OneHotEncoder\nfrom sklearn.model_selection import train_test_split, cross_val_score\nfrom sklearn.metrics import confusion_matrix, roc_auc_score\nimport matplotlib.pyplot as plt","metadata":{"execution":{"iopub.status.busy":"2021-08-18T10:28:18.046177Z","iopub.execute_input":"2021-08-18T10:28:18.046527Z","iopub.status.idle":"2021-08-18T10:28:18.054753Z","shell.execute_reply.started":"2021-08-18T10:28:18.046496Z","shell.execute_reply":"2021-08-18T10:28:18.053398Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"<h1>Анализ и подготовка данных</h1>\nЗагружаем данные и делаем предварительный анализ","metadata":{}},{"cell_type":"code","source":"df_train_full = pd.read_csv('/kaggle/input/ai-academy-dota-2-pro-gamer/dota2_skill_train.csv', index_col='id')\ndisplay(df_train_full)\ndisplay(df_train_full.info())\n\ndf_test_full = pd.read_csv('/kaggle/input/ai-academy-dota-2-pro-gamer/dota2_skill_test.csv', index_col='id')\ndisplay(df_test_full)\ndisplay(df_test_full.info())","metadata":{"execution":{"iopub.status.busy":"2021-08-18T10:28:18.11623Z","iopub.execute_input":"2021-08-18T10:28:18.116718Z","iopub.status.idle":"2021-08-18T10:28:18.942108Z","shell.execute_reply.started":"2021-08-18T10:28:18.11668Z","shell.execute_reply":"2021-08-18T10:28:18.94088Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"В train и test отсутствуют пропущенные значания. Все признаки, кроме первых двух, числовые.\nРассмотрим распределение значаний в категориальных переменных и в целевой переменной.","metadata":{}},{"cell_type":"code","source":"df_train_full['player_team'].value_counts().plot(kind='bar', title='Столбец \"player_team\" в train')\nplt.show()\ndf_train_full['winner_team'].value_counts().plot(kind='bar', title='Столбец \"winner_team\" в train')\nplt.show()\ndf_train_full['skilled'].value_counts().plot(kind='bar', title='Столбец \"skilled\" в train')\nplt.show()\ndf_test_full['player_team'].value_counts().plot(kind='bar', title='Столбец \"player_team\" в test')\nplt.show()\ndf_test_full['winner_team'].value_counts().plot(kind='bar', title='Столбец \"winner_team\" в test')\nNone","metadata":{"execution":{"iopub.status.busy":"2021-08-18T13:14:18.352635Z","iopub.execute_input":"2021-08-18T13:14:18.353041Z","iopub.status.idle":"2021-08-18T13:14:19.292328Z","shell.execute_reply.started":"2021-08-18T13:14:18.353007Z","shell.execute_reply":"2021-08-18T13:14:19.291337Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"В столбце winner_team в train есть странное значение \"other\". Их всего 2, в test эти значения отсутствуют. Удалим эти строки.","metadata":{}},{"cell_type":"code","source":"df_train_full.drop(df_train_full[df_train_full['winner_team']=='other'].index, inplace=True)","metadata":{"execution":{"iopub.status.busy":"2021-08-18T10:28:19.637647Z","iopub.execute_input":"2021-08-18T10:28:19.637956Z","iopub.status.idle":"2021-08-18T10:28:19.674904Z","shell.execute_reply.started":"2021-08-18T10:28:19.637926Z","shell.execute_reply":"2021-08-18T10:28:19.673976Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Закодируем категориальные признаки","metadata":{}},{"cell_type":"code","source":"OH_encoder = OneHotEncoder(handle_unknown='error', sparse=False, drop='if_binary')\nOH_cols_train = pd.DataFrame(OH_encoder.fit_transform(\n                                    df_train_full[['player_team', 'winner_team']]), \n                                    columns=OH_encoder.get_feature_names(input_features=['player_team', 'winner_team']),\n                                    index=df_train_full.index\n                            )\nOH_cols_test = pd.DataFrame(OH_encoder.fit_transform(\n                                    df_test_full[['player_team', 'winner_team']]), \n                                    columns=OH_encoder.get_feature_names(input_features=['player_team', 'winner_team']),\n                                    index=df_test_full.index\n                            )\n\ndisplay(OH_cols_train)\ndisplay(OH_cols_test)","metadata":{"execution":{"iopub.status.busy":"2021-08-18T10:28:19.67684Z","iopub.execute_input":"2021-08-18T10:28:19.67715Z","iopub.status.idle":"2021-08-18T10:28:19.841055Z","shell.execute_reply.started":"2021-08-18T10:28:19.677122Z","shell.execute_reply":"2021-08-18T10:28:19.840031Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Добавим столбцы с закодированными катергориальными признаками к основным датасетам. Незакодированные столбцы удалим","metadata":{}},{"cell_type":"code","source":"df_train = pd.concat([df_train_full, OH_cols_train], axis=1)\ndf_train.drop(columns=['player_team', 'winner_team'], inplace=True)\n\ndf_test = pd.concat([df_test_full, OH_cols_test], axis=1)\ndf_test.drop(columns=['player_team', 'winner_team'], inplace=True)\n\ndisplay(df_train)\ndisplay(df_test)","metadata":{"execution":{"iopub.status.busy":"2021-08-18T10:28:19.842466Z","iopub.execute_input":"2021-08-18T10:28:19.842868Z","iopub.status.idle":"2021-08-18T10:28:19.964298Z","shell.execute_reply.started":"2021-08-18T10:28:19.842826Z","shell.execute_reply":"2021-08-18T10:28:19.963413Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Создаем обучающие, тестовые и валидационные датасеты","metadata":{}},{"cell_type":"code","source":"df_train_y = df_train.pop('skilled')\ndf_train_X = df_train\n\n# Разделите данные на трейн и тест\nX_df, X_test, y_df, y_test = train_test_split(df_train_X, df_train_y, random_state=0, test_size=0.2)\n# Разделите трейн выборку на 2 части для блендинга\nX_train, X_valid, y_train, y_valid = train_test_split(X_df, y_df, random_state=0, test_size=0.4)\n\n# Выборка для обучения базовых моделей без стэкинга (base-line) - 80% исходного датасета\nprint(X_df.shape, y_df.shape)\n# Выборка для валидации базовых моделей без стэкинга (base-line) и матаалгоритма - \n# - 20% исходного датасета\nprint(X_test.shape, y_test.shape)\n# Выборка для обучения базовых моделей для стэкигна\nprint(X_train.shape, y_train.shape)\n# Выборка для обучения метаалгоритма\nprint(X_valid.shape, y_valid.shape)","metadata":{"execution":{"iopub.status.busy":"2021-08-18T10:28:19.965521Z","iopub.execute_input":"2021-08-18T10:28:19.965836Z","iopub.status.idle":"2021-08-18T10:28:20.064623Z","shell.execute_reply.started":"2021-08-18T10:28:19.965805Z","shell.execute_reply":"2021-08-18T10:28:20.063392Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"все датасеты имеют одинаковое количество столбцов и подходят для обучения и валидации","metadata":{}},{"cell_type":"markdown","source":"<h1>Обучение базовых моделей для base-line</h1>\nОбучим базовые модели на большой train выборке для задания нижней границы точности стэкинга (стэкинг должен быть точнее любого базового алгоритма на большой train выборке, в противном случае стэкинг не имеет смысла)","metadata":{}},{"cell_type":"code","source":"rf = RandomForestClassifier(random_state=0)\nrf.fit(X_df, y_df)\nrf_preds = rf.predict(X_test)\nscores = roc_auc_score(y_test, rf_preds)\n\nscores_df = pd.DataFrame([scores], index=['Random Forest Classifier'], columns=['ROC AUC scores'])\nscores_df","metadata":{"execution":{"iopub.status.busy":"2021-08-18T10:28:20.066118Z","iopub.execute_input":"2021-08-18T10:28:20.066428Z","iopub.status.idle":"2021-08-18T10:28:54.665708Z","shell.execute_reply.started":"2021-08-18T10:28:20.0664Z","shell.execute_reply":"2021-08-18T10:28:54.664644Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"lda = LinearDiscriminantAnalysis()\nlda.fit(X_df, y_df)\nlda_preds = lda.predict(X_test)\nscores = roc_auc_score(y_test, lda_preds)\n\nscores_df = scores_df.append(pd.DataFrame([scores], index=['Linear Discriminant Analysis'], columns=['ROC AUC scores']))\nscores_df","metadata":{"execution":{"iopub.status.busy":"2021-08-18T10:28:54.66827Z","iopub.execute_input":"2021-08-18T10:28:54.668588Z","iopub.status.idle":"2021-08-18T10:28:55.381192Z","shell.execute_reply.started":"2021-08-18T10:28:54.668558Z","shell.execute_reply":"2021-08-18T10:28:55.380102Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"gnb = GaussianNB()\ngnb.fit(X_df, y_df)\ngnb_preds = gnb.predict(X_test)\nscores = roc_auc_score(y_test, gnb_preds)\n\nscores_df = scores_df.append(pd.DataFrame([scores], index=['GaussianNB'], columns=['ROC AUC scores']))\nscores_df","metadata":{"execution":{"iopub.status.busy":"2021-08-18T10:28:55.383718Z","iopub.execute_input":"2021-08-18T10:28:55.384362Z","iopub.status.idle":"2021-08-18T10:28:55.535981Z","shell.execute_reply.started":"2021-08-18T10:28:55.384314Z","shell.execute_reply":"2021-08-18T10:28:55.535155Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"clf = make_pipeline(StandardScaler(),\n                    LinearSVC(random_state=0, tol=1e-5))\nclf.fit(X_df, y_df)\nclf_preds = clf.predict(X_test)\nscores = roc_auc_score(y_test, clf_preds)\n\nscores_df = scores_df.append(pd.DataFrame([scores], index=['Linear Support Vector Machines with scaling'], columns=['ROC AUC scores']))\nscores_df","metadata":{"execution":{"iopub.status.busy":"2021-08-18T10:28:55.537289Z","iopub.execute_input":"2021-08-18T10:28:55.537674Z","iopub.status.idle":"2021-08-18T10:29:45.246515Z","shell.execute_reply.started":"2021-08-18T10:28:55.537645Z","shell.execute_reply":"2021-08-18T10:29:45.245495Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"линейные модели показывают лучшую точность после масштабирования данных","metadata":{}},{"cell_type":"code","source":"lrp = make_pipeline(StandardScaler(),\n                    LogisticRegression(solver='newton-cg', random_state=0))\nlrp.fit(X_df, y_df)\nlrp_preds = lrp.predict(X_test)\nscores = roc_auc_score(y_test, lrp_preds)\n\nscores_df = scores_df.append(pd.DataFrame([scores], index=['Logistic Regression with scaling'], columns=['ROC AUC scores']))\nscores_df","metadata":{"execution":{"iopub.status.busy":"2021-08-18T10:29:45.248029Z","iopub.execute_input":"2021-08-18T10:29:45.248621Z","iopub.status.idle":"2021-08-18T10:29:47.93672Z","shell.execute_reply.started":"2021-08-18T10:29:45.248576Z","shell.execute_reply":"2021-08-18T10:29:47.935563Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"логистическая регрессия показала лучшие результаты с алгоритмом 'newton-cg'","metadata":{}},{"cell_type":"code","source":"kncp = make_pipeline(StandardScaler(),\n                    KNeighborsClassifier())\n\nkncp.fit(X_df, y_df)\nkncp_preds = kncp.predict(X_test)\nscores = roc_auc_score(y_test, kncp_preds)\n\nscores_df = scores_df.append(pd.DataFrame([scores], index=['KNeighborsClassifier with scaling'], columns=['ROC AUC scores']))\nscores_df","metadata":{"execution":{"iopub.status.busy":"2021-08-18T10:29:47.94215Z","iopub.execute_input":"2021-08-18T10:29:47.944857Z","iopub.status.idle":"2021-08-18T10:35:15.3612Z","shell.execute_reply.started":"2021-08-18T10:29:47.944786Z","shell.execute_reply":"2021-08-18T10:35:15.360025Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"xgbr = XGBRegressor(n_estimators=1000, learning_rate=0.05, n_jobs=4)\nxgbr.fit(X_df, y_df, \n             early_stopping_rounds=5, \n             eval_set=[(X_test, y_test)], \n             verbose=False)\nxgbr_preds = xgbr.predict(X_test)\nscores = roc_auc_score(y_test, xgbr_preds)\n\nscores_df = scores_df.append(pd.DataFrame([scores], index=['XGBRegressor'], columns=['ROC AUC scores']))\nscores_df","metadata":{"execution":{"iopub.status.busy":"2021-08-18T10:35:15.362686Z","iopub.execute_input":"2021-08-18T10:35:15.362987Z","iopub.status.idle":"2021-08-18T10:36:16.830685Z","shell.execute_reply.started":"2021-08-18T10:35:15.362951Z","shell.execute_reply":"2021-08-18T10:36:16.829696Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"xgbc = XGBRFClassifier(n_estimators=1000, learning_rate=0.05, n_jobs=4)\nxgbc.fit(X_df, y_df, \n             early_stopping_rounds=5, \n             eval_set=[(X_test, y_test)], \n             verbose=False)\nxgbc_preds = xgbc.predict(X_test)\nscores = roc_auc_score(y_test, xgbc_preds)\n\nscores_df = scores_df.append(pd.DataFrame([scores], index=['XGBRFClassifier'], columns=['ROC AUC scores']))\nscores_df","metadata":{"execution":{"iopub.status.busy":"2021-08-18T10:36:16.832289Z","iopub.execute_input":"2021-08-18T10:36:16.83291Z","iopub.status.idle":"2021-08-18T10:38:06.151321Z","shell.execute_reply.started":"2021-08-18T10:36:16.832865Z","shell.execute_reply":"2021-08-18T10:38:06.150279Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"lgbmr = LGBMRegressor(n_estimators=1000, learning_rate=0.05, n_jobs=4)\nlgbmr.fit(X_df, y_df, \n             early_stopping_rounds=5, \n             eval_set=[(X_test, y_test)], \n             verbose=False)\nlgbmr_preds = lgbmr.predict(X_test)\nscores = roc_auc_score(y_test, lgbmr_preds)\n\nscores_df = scores_df.append(pd.DataFrame([scores], index=['LGBMRegressor'], columns=['ROC AUC scores']))\nscores_df","metadata":{"execution":{"iopub.status.busy":"2021-08-18T10:38:06.152786Z","iopub.execute_input":"2021-08-18T10:38:06.15339Z","iopub.status.idle":"2021-08-18T10:38:12.032906Z","shell.execute_reply.started":"2021-08-18T10:38:06.153343Z","shell.execute_reply":"2021-08-18T10:38:12.032123Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"lgbmc = LGBMClassifier(n_estimators=1000, learning_rate=0.05, n_jobs=4)\nlgbmc.fit(X_df, y_df, \n             early_stopping_rounds=5, \n             eval_set=[(X_test, y_test)], \n             verbose=False)\nlgbmc_preds = lgbmc.predict(X_test)\nscores = roc_auc_score(y_test, lgbmc_preds)\n\nscores_df = scores_df.append(pd.DataFrame([scores], index=['LGBMClassifier'], columns=['ROC AUC scores']))\nscores_df","metadata":{"execution":{"iopub.status.busy":"2021-08-18T10:38:12.03416Z","iopub.execute_input":"2021-08-18T10:38:12.034611Z","iopub.status.idle":"2021-08-18T10:38:18.857671Z","shell.execute_reply.started":"2021-08-18T10:38:12.034563Z","shell.execute_reply":"2021-08-18T10:38:18.856992Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Визуализируем результат работы базовых алгоритмов","metadata":{}},{"cell_type":"code","source":"scores_df.plot(kind='bar', legend=False, title='ROC AUC scores', grid=False, ylim=[0.5, 0.9])\nplt.xticks(rotation=90)\nNone","metadata":{"execution":{"iopub.status.busy":"2021-08-18T10:38:18.858882Z","iopub.execute_input":"2021-08-18T10:38:18.859323Z","iopub.status.idle":"2021-08-18T10:38:19.099254Z","shell.execute_reply.started":"2021-08-18T10:38:18.859279Z","shell.execute_reply":"2021-08-18T10:38:19.098574Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"<h1>Реализация стэкинга в библиоткеке scikit-learn</h1>\nВ библиотеке scikit-learn есть реализация стэкинг для задач классификации (класс StackingClassifier). Предсказания базовых моделей используются для финального предсказания конечным классификатором. В отличие от классической реализации стэкинга в scikit-learn базовые модели обучаются на полной обучающей выборке, предсказания базовых моделей используются для обучения мета-алгоритма через кросс-валидацию, чтобы не допустить переобучения (<a href='https://scikit-learn.org/stable/modules/ensemble.html#stacking'>scikit-learn User Guide</a>).","metadata":{}},{"cell_type":"code","source":"estimators = [\n    ('rf', RandomForestClassifier(random_state=0)),\n    ('lda', make_pipeline(StandardScaler(), LinearDiscriminantAnalysis())),\n    ('gnb', GaussianNB()),\n    ('clf', make_pipeline(StandardScaler(),LinearSVC(random_state=0, tol=1e-5))),\n    ('lrp', make_pipeline(StandardScaler(), LogisticRegression(solver='newton-cg', random_state=0))),\n    ('kncp', make_pipeline(StandardScaler(), KNeighborsClassifier()))\n]\n\nstc = StackingClassifier(\n    estimators=estimators, final_estimator=LGBMClassifier(n_estimators=1000, learning_rate=0.05, n_jobs=4))\nstc.fit(X_df, y_df)\nstc_preds = stc.predict(X_test)\nscores = roc_auc_score(y_test, stc_preds)\n\nscores_df = scores_df.append(pd.DataFrame([scores], index=['Stacking Classifier'], columns=['ROC AUC scores']))\nscores_df","metadata":{"execution":{"iopub.status.busy":"2021-08-18T10:38:19.100542Z","iopub.execute_input":"2021-08-18T10:38:19.10114Z","iopub.status.idle":"2021-08-18T11:08:09.200875Z","shell.execute_reply.started":"2021-08-18T10:38:19.101094Z","shell.execute_reply":"2021-08-18T11:08:09.199914Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"estimators = [\n    ('rf', RandomForestClassifier(random_state=0)),\n    ('lda', make_pipeline(StandardScaler(), LinearDiscriminantAnalysis())),\n    ('clf', make_pipeline(StandardScaler(),LinearSVC(random_state=0, tol=1e-5))),\n    ('xgbc', XGBRFClassifier(n_estimators=1000, learning_rate=0.05)),\n    ('lrp', make_pipeline(StandardScaler(), LogisticRegression(solver='newton-cg', random_state=0)))\n]\n\nstc = StackingClassifier(\n    estimators=estimators, final_estimator=LGBMClassifier(n_estimators=1000, learning_rate=0.05, n_jobs=4))\nstc.fit(X_df, y_df)\nstc_preds = stc.predict(X_test)\nscores = roc_auc_score(y_test, stc_preds)\n\nscores_df = scores_df.append(pd.DataFrame([scores], index=['Stacking Classifier short'], columns=['ROC AUC scores']))\nscores_df","metadata":{"execution":{"iopub.status.busy":"2021-08-18T13:36:36.312904Z","iopub.execute_input":"2021-08-18T13:36:36.313353Z","iopub.status.idle":"2021-08-18T13:53:35.085193Z","shell.execute_reply.started":"2021-08-18T13:36:36.313317Z","shell.execute_reply":"2021-08-18T13:53:35.084192Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Реализация стэкинга в библиотеке scikit-learn не смогла продемонстрировать точность лучше, чем мета-алгоритм (LGBMClassifier) на большой выборке, поэтому ее использование (для рассматриваемой задачи) не имеет смысла. Необходимо проверить классическую реализацию стэкинга.\n\nКроме того scikit-learn накладывает ограничение на используемы базовые алгоритмы - можно искользовать только модели-классификаторы.\n\nСтоит отметить, что удаление из набора базовых алгоритмов моделей с низкой точностью привело к снижению общей точности стэкинга. Таким образом, увеличение количества данных пусть и за счет менее точных моделей приводит к увеличению общей точности стэкинга.","metadata":{}},{"cell_type":"markdown","source":"<h1>Классическая реалзиация стэкинга </h1> \nОбучим базовые алгоритмы на первой части тренировочной выборки. Создадим столбцы мета-признаков для обучения мета-алгоритма","metadata":{}},{"cell_type":"code","source":"estimators = [\n    ('rf', RandomForestClassifier(random_state=0)),\n    ('lda', make_pipeline(StandardScaler(), LinearDiscriminantAnalysis())),\n    ('clf', make_pipeline(StandardScaler(),LinearSVR(random_state=0, tol=1e-5))),\n    ('lrp', make_pipeline(StandardScaler(), LogisticRegression(solver='newton-cg', random_state=0))),\n    ('lr', make_pipeline(StandardScaler(), LinearRegression())),\n    ('gnb', GaussianNB()),\n    ('kncp', make_pipeline(StandardScaler(), KNeighborsClassifier())),\n    ('lgbmc', LGBMClassifier(n_estimators=1000, learning_rate=0.05)),\n    ('xgbc', XGBRFClassifier(n_estimators=1000, learning_rate=0.05)),\n    ('lgbmr', LGBMRegressor(n_estimators=1000, learning_rate=0.05))\n]\n\nboosters = ['xgbr', 'xgbc', 'lgbmr', 'lgbmc']\nbase_models_scores = pd.DataFrame()\nbase_models_cols = pd.DataFrame()\n\nfor model in estimators:\n    if model[0] in boosters:\n        model[1].fit(X_train, y_train, \n             early_stopping_rounds=5, \n             eval_set=[(X_valid, y_valid)], \n             verbose=False)\n    else: model[1].fit(X_train, y_train)\n\n    if (model[0] in boosters) or (model[0] == 'clf') or (model[0] == 'lr'):\n        base_models_cols[model[0]] = model[1].predict(X_valid)\n    else:\n        base_models_cols[model[0]] = model[1].predict_proba(X_valid)[:, 1]\n\n    scores = roc_auc_score(y_valid, base_models_cols[model[0]])\n    base_models_scores = base_models_scores.append(pd.DataFrame([scores], index=[model[0]], columns=['ROC AUC scores']))\n\ndisplay(base_models_cols)\ndisplay(base_models_scores)","metadata":{"execution":{"iopub.status.busy":"2021-08-18T12:48:45.943676Z","iopub.execute_input":"2021-08-18T12:48:45.944035Z","iopub.status.idle":"2021-08-18T12:55:15.014766Z","shell.execute_reply.started":"2021-08-18T12:48:45.944005Z","shell.execute_reply":"2021-08-18T12:55:15.013722Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"После сокращения объема обучающей выборки точность базовых алгоритмов бустинга значительно снзилась. При этом точность не бустинговых алгоритмов возрасла. Возможно, это свидетельствует о переобучении не бустинговых алгоритмов на большой выборке и недообучении бустинговых алгоритмов в связи с маленьким объемом обучающей выборки.\n\nСоздадим датафрейм для обучения мета-алгоритма","metadata":{}},{"cell_type":"code","source":"X_valid_meta = pd.DataFrame()\nbase_models_cols.index = X_valid.index\nX_valid_meta = pd.concat([X_valid, base_models_cols], axis=1)\ndisplay(X_valid_meta)","metadata":{"execution":{"iopub.status.busy":"2021-08-18T13:00:27.718736Z","iopub.execute_input":"2021-08-18T13:00:27.719131Z","iopub.status.idle":"2021-08-18T13:00:27.768808Z","shell.execute_reply.started":"2021-08-18T13:00:27.719096Z","shell.execute_reply":"2021-08-18T13:00:27.767828Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Добавим мета-признаки в датасет для оценки точности мета-алгоритма","metadata":{}},{"cell_type":"code","source":"base_models_cols = pd.DataFrame()\n\nfor model in estimators:\n    if (model[0] in boosters) or (model[0] == 'clf') or (model[0] == 'lr'):\n        base_models_cols[model[0]] = model[1].predict(X_test)\n    else:\n        base_models_cols[model[0]] = model[1].predict_proba(X_test)[:, 1]\n\nbase_models_cols.index = X_test.index\nX_test_meta = pd.concat([X_test, base_models_cols], axis=1)\ndisplay(X_test_meta)","metadata":{"execution":{"iopub.status.busy":"2021-08-18T13:01:15.586753Z","iopub.execute_input":"2021-08-18T13:01:15.587292Z","iopub.status.idle":"2021-08-18T13:03:58.667318Z","shell.execute_reply.started":"2021-08-18T13:01:15.587258Z","shell.execute_reply":"2021-08-18T13:03:58.665999Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Обучим мета-алгоритм на второй части обучающей выборки и оценим его точность","metadata":{}},{"cell_type":"code","source":"xgbr.fit(X_valid_meta, y_valid, \n             early_stopping_rounds=5, \n             eval_set=[(X_test_meta, y_test)], \n             verbose=False)\nxgbr_preds = xgbr.predict(X_test_meta)\nscores = roc_auc_score(y_test, xgbr_preds)\n\nscores_df = scores_df.append(pd.DataFrame([scores], index=['META-XGBRegressor'], columns=['ROC AUC scores']))\nscores_df","metadata":{"execution":{"iopub.status.busy":"2021-08-18T13:04:17.500482Z","iopub.execute_input":"2021-08-18T13:04:17.500913Z","iopub.status.idle":"2021-08-18T13:04:32.315532Z","shell.execute_reply.started":"2021-08-18T13:04:17.500879Z","shell.execute_reply":"2021-08-18T13:04:32.31433Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"<h2>Вывод:</h2>\n\nДаже добавление 10 базовых моделей в стэкинг не позволило превзойти самую точную модель на большой обучающей выборке. Возможно, тонкая настройка параметров моделей позволила бы увеличить точность стэкинга, однако такое исследование не является целью настоящего доклада.\n\nВозможно, для стэкинга не хватило данных. Обучение базовых алгоритмов проходило на большой обучающей выборке (64К строк), обучение моделей для стэкинга и мета-алгоритма производилось на меньших обучающих выборках (38К и 26К соответственно), пусть и с дополнительными мета-признаками.\n\nВ любом случае, стэкинг, возможно, повзолит лишь немного увеличить точность существующих моделей на финальном этапе разработки и не может являться основным методом разрабатываемой модели.","metadata":{}}]}