{"cells":[{"metadata":{"trusted":true},"cell_type":"code","source":"'''\n!pip install ../input/python-datatable/datatable-0.11.0-cp37-cp37m-manylinux2010_x86_64.whl\nimport datatable as dt\n'''","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"cell_type":"code","source":"import riiideducation\nimport pandas as pd\nimport numpy as np\nfrom sklearn.metrics import roc_auc_score\nenv = riiideducation.make_env()","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"d629ff2d2480ee46fbb7e2d37f6b5fab8052498a","_cell_guid":"79c7e3d0-c299-4dcb-8224-4455121ee9b0","trusted":true},"cell_type":"code","source":"train_csv = pd.read_csv(\"../input/riiid-test-answer-prediction/train.csv\", \n                        usecols=[1, 2, 3, 4, 7, 8, 9],\n                        dtype={'timestamp': 'int64',\n                          'user_id': 'int32',\n                          'content_id': 'int16',\n                          'content_type_id': 'int8',\n                          'answered_correctly':'int8',\n                          'prior_question_elapsed_time': 'float32',\n                          'prior_question_had_explanation': 'boolean'}, nrows=65000000  \n                   )\nquestions_csv = pd.read_csv(\"../input/riiid-test-answer-prediction/questions.csv\")","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_csv = train_csv[train_csv.content_type_id == False]\n\n#arrange by timestamp\ntrain_csv = train_csv.sort_values(['timestamp'], ascending=True).reset_index(drop = True)\ntrain_csv = train_csv.drop(columns=['content_type_id'])\ntrain_csv.head(10)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# how, on average, the users correclty answered at the first question, second question and so on \ncontent_mean_final = train_csv[['content_id','answered_correctly']].groupby(['content_id']).agg(['mean'])\ncontent_mean_final.columns = [\"answered_correctly_content_mean\"]  \n","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# in average how much a student correclty replay and the total number of questions for student \nuser_mean_final = train_csv[['user_id','answered_correctly']].groupby(['user_id']).agg(['mean', 'count'])\nuser_mean_final.columns = [\"answered_correctly_user_mean\", 'count']","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"#saving value to fillna\nelapsed_mean = train_csv.prior_question_elapsed_time.mean()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"'''\nfrom datetime import datetime\ntrain_csv['timestamp'] = pd.to_datetime(train_csv['timestamp'], unit='ms',origin='2017-1-1')\ntrain_csv['month']=(train_csv.timestamp.dt.month)\naveg = train_csv[['user_id','month','prior_question_elapsed_time']].groupby(['user_id','month']).mean()\naveg.columns=['mean']\n'''","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"import gc\ngc.collect()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# for each user, the last 3 intereaction will be the validation set \nvalidation = pd.DataFrame()\nfor i in range(4):\n    last_records = train_csv.drop_duplicates('user_id', keep = 'last')\n    train_csv = train_csv[~train_csv.index.isin(last_records.index)]\n    validation = validation.append(last_records)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_csv = pd.merge(train_csv, user_mean_final , on=['user_id'], how=\"left\")\ntrain_csv= pd.merge(train_csv, content_mean_final, on=['content_id'], how=\"left\")\n\nvalidation = pd.merge(validation, user_mean_final, on=['user_id'], how=\"left\")\nvalidation = pd.merge(validation,content_mean_final , on=['content_id'], how=\"left\")","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"from sklearn.preprocessing import LabelEncoder\n\nlb_make = LabelEncoder()\n\ntrain_csv.prior_question_had_explanation.fillna(False, inplace = True)\nvalidation.prior_question_had_explanation.fillna(False, inplace = True)\n\nvalidation[\"prior_question_had_explanation_enc\"] = lb_make.fit_transform(validation[\"prior_question_had_explanation\"])\ntrain_csv[\"prior_question_had_explanation_enc\"] = lb_make.fit_transform(train_csv[\"prior_question_had_explanation\"])","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"questions_csv = questions_csv.drop(columns = ['bundle_id'])\nquestions_csv = questions_csv.drop(columns = ['correct_answer'])\nquestions_csv = questions_csv.drop(columns = ['tags'])","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_csv= pd.merge(train_csv, questions_csv, left_on = 'content_id', right_on = 'question_id', how = 'left')\nvalidation = pd.merge(validation, questions_csv, left_on = 'content_id', right_on = 'question_id', how = 'left')","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"y_train = train_csv['answered_correctly']\ntrain_csv = train_csv.drop(['answered_correctly'], axis=1)\n\ny_val = validation['answered_correctly']\nvalidation = validation.drop(['answered_correctly'], axis=1)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_csv = train_csv[['answered_correctly_user_mean', 'answered_correctly_content_mean', 'count',\n       'prior_question_elapsed_time','prior_question_had_explanation_enc', 'part']]\nvalidation = validation[['answered_correctly_user_mean', 'answered_correctly_content_mean', 'count',\n       'prior_question_elapsed_time','prior_question_had_explanation_enc', 'part']]","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_csv['prior_question_elapsed_time'].fillna(train_csv['prior_question_elapsed_time'].mean(), inplace = True)\nvalidation['prior_question_elapsed_time'].fillna(validation['prior_question_elapsed_time'].mean(), inplace = True)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"import lightgbm as lgb\n\nparams = {\n    'objective': 'binary',\n    'max_bin': 700,\n    'learning_rate': 0.0175,\n    'num_leaves': 80\n}\n\nlgb_train = lgb.Dataset(train_csv, y_train, categorical_feature = ['part', 'prior_question_had_explanation_enc'])\nlgb_eval = lgb.Dataset(validation, y_val, categorical_feature = ['part', 'prior_question_had_explanation_enc'], reference=lgb_train)\nmodel = lgb.train(\n    params, lgb_train,\n    valid_sets=[lgb_train, lgb_eval],\n    verbose_eval=50, # ogni quanti cicli mostra il valore ottenuto \n    num_boost_round=1000,\n    early_stopping_rounds=10\n)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"y_pred = model.predict(validation)\ny_true = np.array(y_val)\nroc_auc_score(y_true, y_pred)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"iter_test = env.iter_test()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"for (test_df, sample_prediction_df) in iter_test:\n    test_df = pd.merge(test_df, questions_csv, left_on = 'content_id', right_on = 'question_id', how = 'left')\n    test_df = pd.merge(test_df, user_mean_final, on=['user_id'],  how=\"left\")\n    test_df = pd.merge(test_df, content_mean_final, on=['content_id'],  how=\"left\")\n    test_df['answered_correctly_user_mean'].fillna(0.5,  inplace=True)\n    test_df['answered_correctly_content_mean'].fillna(0.5,  inplace=True)\n    #test_df['part'] = test_df.part - 1\n\n    test_df['part'].fillna(int(test_df['part'].mean()), inplace = True)\n    test_df['count'].fillna(0, inplace=True)\n    test_df['prior_question_elapsed_time'].fillna(test_df['prior_question_elapsed_time'].mean(), inplace = True)\n    test_df['prior_question_had_explanation'].fillna(False, inplace=True)\n    test_df[\"prior_question_had_explanation_enc\"] = lb_make.fit_transform(test_df[\"prior_question_had_explanation\"])\n    test_df['answered_correctly'] =  model.predict(test_df[['answered_correctly_user_mean', 'answered_correctly_content_mean','count',\n                                                                  'prior_question_elapsed_time','prior_question_had_explanation_enc', 'part']])\n    env.predict(test_df.loc[test_df['content_type_id'] == 0, ['row_id', 'answered_correctly']])","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"print('finish :)')","execution_count":null,"outputs":[]}],"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":4,"nbformat_minor":4}