天池大数据比赛-智能制造质量预测.zip

import pandas as pd from sklearn import preprocessing import numpy as np import xgboost as xgb from sklearn.model_selection import train_test_split import math from sklearn.metrics import mean_squared_error from sklearn.linear_model import LinearRegression from sklearn.ensemble import RandomForestRegressor from sklearn.linear_model import Ridge from sklearn.linear_model import Lasso from mlxtend.regressor import StackingRegressor # 800 rows x 5954 columns train_data = pd.read_csv('Code/MetaData/train.csv') # 300 rows x 5953 columns test_data = pd.read_csv('Code/MetaData/test_A.csv') test_data_id = pd.DataFrame(pd.read_csv('Code/MetaData/test_A.csv')['ID']) # 800 rows x 5727 columns train_data = train_data.dropna(how='all', axis=1) # 300 rows x 5953 columns test_data = test_data.dropna(how='all', axis=1) train_data = train_data.fillna(train_data.median()) test_data = test_data.fillna(test_data.median()) # 220x150 220x151 date_tool_column = ['220x150', '220x151'] for index, row in train_data.iterrows(): for column in train_data.columns: if row[column] >= 20170000000000: date_tool_column.append(column) break date_tool_feature = [x for x in train_data.columns if x not in date_tool_column] # 800 rows x 5638 columns train_date_null = train_data[date_tool_feature] zero_var = [] var_val = pd.DataFrame(train_date_null.var()) for index, row in var_val.iterrows(): if row[0] == 0: zero_var.append(index) # 800 rows x 4963 columns zero_var_feature = [x for x in train_date_null if x not in zero_var] train_zero_var = train_date_null[zero_var_feature] feature = [x for x in train_zero_var.columns if x not in ['Value']] train_test_columns = [] for column in train_zero_var[feature].columns: train_test_columns.append(column) test_feature = [x for x in test_data if x in train_test_columns] test_data = test_data[test_feature].reset_index() params = {'booster': 'dart', 'objective': 'reg:linear', 'eval_metric': 'rmse', 'max_depth': 8, 'lambda': 100, 'subsample': 0.75, 'colsample_bytree': 0.65, 'eta': 0.02, 'sample_type': 'uniform', 'normalize': 'tree', 'rate_drop': 0.15, # 0.1 'skip_drop': 0.85, # 0.9 'seed': 1024, 'silent': 1, } X_train, X_test, y_train, y_test = train_test_split(train_zero_var[feature], train_zero_var['Value'], test_size=0.2, random_state=0) xgbtrain = xgb.DMatrix(X_train[feature], y_train) watchlist = [(xgbtrain, 'train'), (xgbtrain, 'evaluate')] model = xgb.train(params, xgbtrain, num_boost_round=2889, early_stopping_rounds=50, evals=watchlist) rfreg = RandomForestRegressor(random_state=1, max_depth=15) ridge_reg = Ridge(normalize=True) lasso_reg = Lasso() linear_reg = LinearRegression(normalize=True) stacking_reg = StackingRegressor(regressors=[rfreg, ridge_reg, lasso_reg], meta_regressor=linear_reg) feature = [x for x in train_zero_var.columns if x not in ['Value']] # X_train, X_test, y_train, y_test = train_test_split(train_zero_var[feature], train_zero_var['Value'], test_size=0.2, # random_state=0) stacking_reg.fit(X_train, y_train) stacking_test = pd.DataFrame(stacking_reg.predict(X_test)) stacking_test.columns = ['stacking_pred'] y_test = pd.DataFrame(y_test) y_test.columns = ['Value'] mean_squared_error(stacking_test['stacking_pred'], y_test['Value']) train_zero_var = train_zero_var.reset_index() # predict for Random Forest rf_pred = pd.DataFrame() for idx in range(0, 5): train = train_zero_var[train_zero_var['index'] % 5 != idx] test = train_zero_var[train_zero_var['index'] % 5 == idx] stacking_feature = [x for x in train.columns if x not in ['index', 'Value']] rfreg.fit(train[stacking_feature], train['Value']) y_pred = pd.DataFrame(test['index']) y_pred['Random Forest'] = rfreg.predict(test[stacking_feature]) rf_pred = rf_pred.append(y_pred) # rf_pred = rf_pred.dropna() df = pd.merge(train_zero_var, rf_pred, how='left', on='index') # test_data predict for Random Forest rf_feature = [x for x in train_zero_var.columns if x not in ['index', 'Value']] rfreg.fit(train_zero_var[rf_feature], train_zero_var['Value']) rf_test_pred = pd.DataFrame(test_data['index']) rf_test_pred['Random Forest'] = rfreg.predict(test_data[rf_feature]) df_test = pd.merge(test_data, rf_test_pred, how='left', on='index') # predict for GBDT from sklearn.ensemble import GradientBoostingRegressor gbdt_reg = GradientBoostingRegressor(max_depth=7) gbdt_pred = pd.DataFrame() for idx in range(0, 5): train = train_zero_var[train_zero_var['index'] % 5 != idx] test = train_zero_var[train_zero_var['index'] % 5 == idx] stacking_feature = [x for x in train.columns if x not in ['index', 'Value']] gbdt_reg.fit(train[stacking_feature], train['Value']) y_pred = pd.DataFrame(test['index']) y_pred['GBDT'] = gbdt_reg.predict(test[stacking_feature]) gbdt_pred = gbdt_pred.append(y_pred) # rf_pred = rf_pred.dropna() df = pd.merge(df, gbdt_pred, how='left', on='index') # predict for Lasso lasso_pred = pd.DataFrame() for idx in range(0, 5): train = train_zero_var[train_zero_var['index'] % 5 != idx] test = train_zero_var[train_zero_var['index'] % 5 == idx] stacking_feature = [x for x in train.columns if x not in ['index', 'Value']] lasso_reg.fit(train[stacking_feature], train['Value']) y_pred = pd.DataFrame(test['index']) y_pred['Lasso'] = lasso_reg.predict(test[stacking_feature]) lasso_pred = lasso_pred.append(y_pred) # lasso_pred = lasso_pred.dropna() df = pd.merge(df, lasso_pred, how='left', on='index') # test_data predict for Lasso lasso_feature = [x for x in train_zero_var.columns if x not in ['index', 'Value']] lasso_reg.fit(train_zero_var[lasso_feature], train_zero_var['Value']) lasso_test_pred = pd.DataFrame(test_data['index']) lasso_test_pred['Lasso'] = lasso_reg.predict(test_data[lasso_feature]) df_test = pd.merge(df_test, lasso_test_pred, how='left', on='index') # predict for Ridge ridge_pred = pd.DataFrame() for idx in range(0, 5): train = train_zero_var[train_zero_var['index'] % 5 != idx] test = train_zero_var[train_zero_var['index'] % 5 == idx] stacking_feature = [x for x in train.columns if x not in ['index', 'Value']] ridge_reg.fit(train[stacking_feature], train['Value']) y_pred = pd.DataFrame(test['index']) y_pred['Ridge'] = ridge_reg.predict(test[stacking_feature]) ridge_pred = ridge_pred.append(y_pred) # lasso_pred = lasso_pred.dropna() df = pd.merge(df, ridge_pred, how='left', on='index') # predict for Dart dart_pred = pd.DataFrame() for idx in range(0, 5): train = train_zero_var[train_zero_var['index'] % 5 != idx] test = train_zero_var[train_zero_var['index'] % 5 == idx] stacking_feature = [x for x in train.columns if x not in ['index', 'Value']] xgbtrain = xgb.DMatrix(train[stacking_feature], train['Value']) watchlist = [(xgbtrain, 'train'), (xgbtrain, 'evaluate')] dart_model = xgb.train(params, xgbtrain, num_boost_round=2889, early_stopping_rounds=50, evals=watchlist) y_pred = pd.DataFrame(test['index']) y_pred['xgb_dart'] = dart_model.predict(xgb.DMatrix(test[stacking_feature])) dart_pred = dart_pred.append(y_pred) # dart_pred = dart_pred.dropna() df = pd.merge(df, dart_pred, how='left', on='index') # test_data predict for Dart dart_feature = [x for x in train_zero_var.columns if x not in ['index', 'Value']] xgbtrain = xgb.DMatrix(train_zero_var[dart_feature], train_zero_var['Value']) watchlist = [(xgbtrain, 'train'), (xgbtrain, 'evaluate')] dart_reg = xgb.train(params, xgbtrain, num_boost_round=2889, early_stopping_rounds=50, evals=watchlist) dart_test_pred = pd.DataFrame(test_data['index']) dart_test_pred['xgb_dart'] = dart_reg.predict(xgb.DMatrix(test_data[dart_feature])) df_test = pd.merge(df_test, dart_test_pred, how='