AdaBoost: deprecation of "base_estimator" does not handle "base_estimator=None" setting properly #26241
Description
Describe the bug
Scikit-learn 1.2 deprecated AdaBoostClassifier 's base_estimator in favour of estimator (see #23819). Because there are also validators in place, old code that explicitly defined base_estimator=None stopped working.
A solution that fixes the deprecation is to add a possible None to a list allowed values in _parameter_constraints; I will do that in a PR.
Steps/Code to Reproduce
from sklearn.ensemble import AdaBoostClassifier
clf = AdaBoostClassifier(base_estimator=None)
clf.fit([[1]], [0])
Expected Results
No error is thrown.
Actual Results
Traceback (most recent call last):
File "<string>", line 1, in <module>
File "/Users/marko/opt/miniconda3/envs/orange310/lib/python3.10/site-packages/sklearn/ensemble/_weight_boosting.py", line 124, in fit
self._validate_params()
File "/Users/marko/opt/miniconda3/envs/orange310/lib/python3.10/site-packages/sklearn/base.py", line 600, in _validate_params
validate_parameter_constraints(
File "/Users/marko/opt/miniconda3/envs/orange310/lib/python3.10/site-packages/sklearn/utils/_param_validation.py", line 97, in validate_parameter_constraints
raise InvalidParameterError(
sklearn.utils._param_validation.InvalidParameterError: The 'base_estimator' parameter of AdaBoostClassifier must be an object implementing 'fit' and 'predict' or a str among {'deprecated'}. Got None instead.
Versions
sklearn: 1.2.2; others are not important