Extract Members of Bagging Regressor Ensemble

Question

I have used the BaggingRegressor class in order to build the best model with the following parameters:

from sklearn.tree import DecisionTreeRegressor
from sklearn.ensemble import BaggingRegressor
Reg_ensemble=BaggingRegressor(base_estimator=DecisionTreeRegressor(max_depth=3),n_estimators=10,random_state=0).fit(feature,target)

with above setting, it will create 10 trees. I want to extract and access to each member of ensemble regression(each tree)separately and then fit a test sample on each of them. Is it possible to access each model?

Answer 1

The estimators_ attribute of the fitted model provides a list with the ensemble estimators; here is an example with dummy data and n_estimators=3 for brevity:

from sklearn.tree import DecisionTreeRegressor
from sklearn.ensemble import BaggingRegressor
from sklearn.datasets import make_regression

X, y = make_regression(n_samples=100, n_features=4,
                    n_informative=2, n_targets=1,
                    random_state=0, shuffle=False)
regr = BaggingRegressor(base_estimator=DecisionTreeRegressor(max_depth=3),
                        n_estimators=3, random_state=0)
regr.fit(X, y)

regr.estimators_
# result:
[DecisionTreeRegressor(ccp_alpha=0.0, criterion='mse', max_depth=3,
                       max_features=None, max_leaf_nodes=None,
                       min_impurity_decrease=0.0, min_impurity_split=None,
                       min_samples_leaf=1, min_samples_split=2,
                       min_weight_fraction_leaf=0.0, presort='deprecated',
                       random_state=2087557356, splitter='best'),
 DecisionTreeRegressor(ccp_alpha=0.0, criterion='mse', max_depth=3,
                       max_features=None, max_leaf_nodes=None,
                       min_impurity_decrease=0.0, min_impurity_split=None,
                       min_samples_leaf=1, min_samples_split=2,
                       min_weight_fraction_leaf=0.0, presort='deprecated',
                       random_state=132990059, splitter='best'),
 DecisionTreeRegressor(ccp_alpha=0.0, criterion='mse', max_depth=3,
                       max_features=None, max_leaf_nodes=None,
                       min_impurity_decrease=0.0, min_impurity_split=None,
                       min_samples_leaf=1, min_samples_split=2,
                       min_weight_fraction_leaf=0.0, presort='deprecated',
                       random_state=1109697837, splitter='best')]

After you have fitted the BaggingRegressor (the base estimators do not exist before fitting), you can access the base estimators for fitting with data Xs, ys simply as:

for model in regr.estimators_:
    model.fit(Xs, Ys)