nycdsa_p4_ridge.py

# Kernel Ridge GridSearch
from sklearn.kernel_ridge import KernelRidge

kridge_grid = KernelRidge()

parameter_grid = {'alpha': [0.0001,0.001,0.01,0.1],
                 'degree': [1,2,3,4],
                 'kernel': ['polynomial']
                 #'n_estimators': [200,210,240,250],
                 #'min_child_weight': [1,2,3,4]
                 }

cross_validation = StratifiedKFold(np.array(label_df['SalePrice']), n_folds=10)

grid_search_kridge = GridSearchCV(kridge_grid,
                           param_grid = parameter_grid,
                           scoring = 'mean_squared_error',
                           cv = cross_validation)

grid_search_kridge.fit(train_df_munged, label_df)

print('Best score (Kernel Ridge): {}'.format(grid_search_kridge.best_score_))
print('Best parameters: {}'.format(grid_search_kridge.best_params_))

y_train_pred_kridge = grid_search_kridge.best_estimator_.predict(train_df_munged)
print("Kernel Ridge score on training set: ", rmse(label_df,y_train_pred_kridge))
y_test_pred_kridge = grid_search_kridge.best_estimator_.predict(test_df_munged)