Is n_estimators in BaggingRegressor() the number of trees or data subsets?

Question

I'm learning about using the BaggingRegressor() from scikit learn (https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.BaggingRegressor.html)

Its n_estimators parameter is defined as:

n_estimators: int, default=10
The number of base estimators in the ensemble.

Am I right in thinking this changes the number of tree-based models running in total?

For example, I am using xgboost, If I set

BaggingRegressor(n_estimators=100) 

is it changing the n_estimators in xgb to be:

• xgboost.XGBRegressor(n_estimators=100) or does it do
• xgboost.XGBRegressor()*100?

Also what within BaggingRegressor() decides the number of random subsets to be created if it's not the number of base estimators?

Answer 1

It is the number of base estimators (not necessarily tree-based). So, if you give XGBoost as the base estimator, which I think is a bit complex to be a base estimator, it'll use 100 XGBoost models. XGBoost itself has n_estimators hyper-parameter but BaggingRegressor does not change it. It treats your estimator object as black-box.