nycdsa_p4_selection.py

# Extra Trees Regressor

et_regr = ExtraTreesRegressor()
et_regr.fit(train_df_munged, label_df)

# Run prediction on training set to get a rough idea of how well it does.
y_pred = et_regr.predict(train_df_munged)
y_test = label_df
print("Extra Trees Regressor score on training set: ", rmse(y_test, y_pred))

# Run prediction on the Kaggle test set.
y_test_pred_et = et_regr.predict(test_df_munged)

# Fit model using each importance as a threshold
thresholds = sort(et_regr.feature_importances_)
#thresholds = sort([0.1,0.2])
for thresh in thresholds:
    # select features using threshold
    selection = SelectFromModel(et_regr, threshold=thresh, prefit=True)
    select_X_train = selection.transform(train_df_munged)
    # train model
    selection_model = ExtraTreesRegressor()
    selection_model.fit(select_X_train, y_test)
    # eval model
    select_X_test = selection.transform(train_df_munged)
    
    y_pred = selection_model.predict(select_X_test)
    print("Thresh=%.3f, n=%d, RMSE= %.10f" % (thresh, select_X_train.shape[1], rmse(y_test, y_pred)))
    
selection = SelectFromModel(et_regr, threshold=0.01, prefit=True)
select_X_train = selection.transform(train_df_munged)
# train model
selection_model = ExtraTreesRegressor()
selection_model.fit(select_X_train, y_test)
# eval model
select_X_test = selection.transform(test_df_munged)

y_test_pred_et_selec = selection_model.predict(select_X_test)