[Bug]: Regression and classification learner checks for breaking changes in scikit-learn

[PhilippBach]

Describe the bug

Support for is_regressor and is_classifier is deprecated with scikit-learn 1.6.0, see Release notes and scikit-learn/scikit-learn#30122

Minimum reproducible code snippet

https://docs.doubleml.org/stable/examples/py_double_ml_pension.html

see failing job https://github.com/DoubleML/doubleml-docs/actions/runs/12273024534/job/34242980869

Expected Result

sucessful execution

Actual Result

 AttributeError                            Traceback (most recent call last)
Cell In[21], line 9
      4 boost_class = XGBClassifier(use_label_encoder=False, n_jobs=1,
      5                             objective = "binary:logistic", eval_metric = "logloss",
      6                             eta=0.1, n_estimators=34)
      8 np.random.seed(123)
----> 9 dml_plr_boost = dml.DoubleMLPLR(data_dml_base,
     10                                 ml_l = boost,
     11                                 ml_m = boost_class,
     12                                 n_folds = 3)
     13 dml_plr_boost.fit(store_predictions=True)
     14 boost_summary = dml_plr_boost.summary

File ~/work/doubleml-docs/doubleml-docs/doubleml-for-py/doubleml/plm/plr.py:111, in DoubleMLPLR.__init__(self, obj_dml_data, ml_l, ml_m, ml_g, n_folds, n_rep, score, draw_sample_splitting)
    108 valid_scores = ['IV-type', 'partialling out']
    109 _check_score(self.score, valid_scores, allow_callable=True)
--> 111 _ = self._check_learner(ml_l, 'ml_l', regressor=True, classifier=False)
    112 ml_m_is_classifier = self._check_learner(ml_m, 'ml_m', regressor=True, classifier=True)
    113 self._learner = {'ml_l': ml_l, 'ml_m': ml_m}

File ~/work/doubleml-docs/doubleml-docs/doubleml-for-py/doubleml/double_ml.py:906, in DoubleML._check_learner(learner, learner_name, regressor, classifier)
    904 else:
    905     assert regressor  # classifier, regressor or both must be True
--> 906     if not is_regressor(learner):
    907         warnings.warn(warn_msg_prefix + f'{str(learner)} is (probably) no regressor.')
    908     learner_is_classifier = False

File /opt/hostedtoolcache/Python/3.12.7/x64/lib/python3.12/site-packages/sklearn/base.py:1277, in is_regressor(estimator)
   1270     warnings.warn(
   1271         f"passing a class to {print(inspect.stack()[0][3])} is deprecated and "
   1272         "will be removed in 1.8. Use an instance of the class instead.",
   1273         FutureWarning,
   1274     )
   1275     return getattr(estimator, "_estimator_type", None) == "regressor"
-> 1277 return get_tags(estimator).estimator_type == "regressor"

File /opt/hostedtoolcache/Python/3.12.7/x64/lib/python3.12/site-packages/sklearn/utils/_tags.py:405, in get_tags(estimator)
    403 for klass in reversed(type(estimator).mro()):
    404     if "__sklearn_tags__" in vars(klass):
--> 405         sklearn_tags_provider[klass] = klass.__sklearn_tags__(estimator)  # type: ignore[attr-defined]
    406         class_order.append(klass)
    407     elif "_more_tags" in vars(klass):

File /opt/hostedtoolcache/Python/3.12.7/x64/lib/python3.12/site-packages/sklearn/base.py:613, in RegressorMixin.__sklearn_tags__(self)
    612 def __sklearn_tags__(self):
--> 613     tags = super().__sklearn_tags__()
    614     tags.estimator_type = "regressor"
    615     tags.regressor_tags = RegressorTags()

AttributeError: 'super' object has no attribute '__sklearn_tags__'

Versions

See https://github.com/DoubleML/doubleml-docs/actions/runs/12273024534/job/34242980869

[PhilippBach]

If you are experiencing this issue a temporary fix would be to use scikit-learn <=1.5.2

[OliverSchacht]

Hi @PhilippBach,
#279 fixes this as suggested for the next release. In the release after, we can check if it is fixed in xgboost and might remove the restriction again.