XGBoost-使用XGBRegressor选择功能

Question

I am trying to perform features selection (for regression tasks) by XGBRegressor() .

More precisely, I would like to know:

• If there is something like the method feature_importances_ , utilized with XGBClassifier , which I could use for regression.
• If the XGBoost's method plot_importance() is reliable when it is used with XGBRegressor()

Answer 1

最后，我通过以下方式解决了这个问题：

model.booster().get_score(importance_type='weight')

Answer 2

Here is my solution (Xnames refers to the feature names):

def xgb_feature_importance(model_xgb, fnames=None):
    b = model_xgb.booster()
    fs = b.get_fscore()
    all_features = [fs.get(f, 0.) for f in b.feature_names]
    all_features = np.array(all_features, dtype=np.float32)
    all_features_imp = all_features / all_features.sum()
    if fnames is not None:
        return pd.DataFrame({'X':fnames, 'IMP': all_features_imp})
    else:
        return all_features_imp