fix for sklearn tags feature

.github/workflows/ci.yml

@@ -21,7 +21,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: [3.8, 3.9, '3.10', 3.11]
+        python-version: ['3.10', 3.11, 3.12, 3.13]
 
     env:
       UCI_DB: "https://archive.ics.uci.edu/ml/machine-learning-databases"

README.md

@@ -76,9 +76,9 @@ Supported Python Configurations:
 
 | OS      | Python version |
 | ------- | -------------- |
-| macOS   | 3.8 – 3.11     |
-| Ubuntu  | 3.8 – 3.11     |
-| Windows | 3.8 – 3.11     |
+| macOS   | 3.10 – 3.13     |
+| Ubuntu  | 3.10 – 3.13     |
+| Windows | 3.10 – 3.13     |
 
 ### (Optional) Create a virtual environment
 

aif360/sklearn/inprocessing/infairness.py

@@ -46,15 +46,16 @@ def __init__(self, *args, criterion, train_split=None, regression='auto',
                          dataset=dataset, **kwargs)
         self.regression = regression
 
-    @property
-    def _estimator_type(self):
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
         if hasattr(self, "regression_"):
-            return 'regressor' if self.regression_ else 'classifier'
+            tags.estimator_type = 'regressor' if self.regression_ else 'classifier'
         elif self.regression != 'auto':
-            return 'regressor' if self.regression else 'classifier'
+            tags.estimator_type = 'regressor' if self.regression else 'classifier'
         else:
             raise NotFittedError("regression is set to 'auto'. Call 'fit' with "
                     "appropriate arguments or set regression manually.")
+        return tags
 
     def get_loss(self, y_pred, y_true, X=None, training=False):
         """Return the loss for this batch.

aif360/sklearn/postprocessing/__init__.py

@@ -2,12 +2,20 @@
 Post-processing algorithms modify predictions to be more fair (predictions in,
 predictions out).
 """
+from dataclasses import dataclass
 from logging import warning
 
 import pandas as pd
 from sklearn.base import BaseEstimator, MetaEstimatorMixin, clone
 from sklearn.model_selection import train_test_split
 from sklearn.utils.metaestimators import available_if
+from sklearn.utils import get_tags, Tags
+
+
+@dataclass
+class PostProcessingTags(Tags):
+    requires_proba: bool = True
+
 
 from aif360.sklearn.postprocessing.calibrated_equalized_odds import CalibratedEqualizedOdds
 from aif360.sklearn.postprocessing.reject_option_classification import RejectOptionClassifier, RejectOptionClassifierCV
@@ -54,9 +62,10 @@ def __init__(self, estimator, postprocessor, *, prefit=False, val_size=0.25,
         self.val_size = val_size
         self.options = options
 
-    @property
-    def _estimator_type(self):
-        return self.postprocessor._estimator_type
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        tags.estimator_type = get_tags(self.postprocessor).estimator_type
+        return tags
 
     @property
     def classes_(self):
@@ -85,8 +94,8 @@ def fit(self, X, y, sample_weight=None, **fit_params):
         self.estimator_ = self.estimator if self.prefit else clone(self.estimator)
 
         try:
-            use_proba = self.postprocessor._get_tags()['requires_proba']
-        except KeyError:
+            use_proba = get_tags(self.postprocessor).requires_proba
+        except AttributeError:
             raise TypeError("`postprocessor` (type: {}) does not have a "
                             "'requires_proba' tag.".format(type(self.estimator)))
         if use_proba and not hasattr(self.estimator, 'predict_proba'):
@@ -145,7 +154,7 @@ def predict(self, X):
         Returns:
             numpy.ndarray: Predicted class label per sample.
         """
-        use_proba = self.postprocessor_._get_tags()['requires_proba']
+        use_proba = get_tags(self.postprocessor_).requires_proba
         y_score = (self.estimator_.predict_proba(X) if use_proba else
                    self.estimator_.predict(X))
         y_score = pd.DataFrame(y_score, index=X.index).squeeze('columns')
@@ -169,7 +178,7 @@ def predict_proba(self, X):
             in the model, where classes are ordered as they are in
             ``self.classes_``.
         """
-        use_proba = self.postprocessor_._get_tags()['requires_proba']
+        use_proba = get_tags(self.postprocessor_).requires_proba
         y_score = (self.estimator_.predict_proba(X) if use_proba else
                    self.estimator_.predict(X))
         y_score = pd.DataFrame(y_score, index=X.index).squeeze('columns')
@@ -193,7 +202,7 @@ def predict_log_proba(self, X):
             the model, where classes are ordered as they are in
             ``self.classes_``.
         """
-        use_proba = self.postprocessor_._get_tags()['requires_proba']
+        use_proba = get_tags(self.postprocessor_).requires_proba
         y_score = (self.estimator_.predict_proba(X) if use_proba else
                    self.estimator_.predict(X))
         y_score = pd.DataFrame(y_score, index=X.index).squeeze('columns')
@@ -216,7 +225,7 @@ def score(self, X, y, sample_weight=None):
         Returns:
             float: Score value.
         """
-        use_proba = self.postprocessor_._get_tags()['requires_proba']
+        use_proba = get_tags(self.postprocessor_).requires_proba
         y_score = (self.estimator_.predict_proba(X) if use_proba else
                    self.estimator_.predict(X))
         y_score = pd.DataFrame(y_score, index=X.index).squeeze('columns')

aif360/sklearn/postprocessing/calibrated_equalized_odds.py

@@ -1,10 +1,13 @@
+from dataclasses import asdict
+
 import numpy as np
 from sklearn.base import BaseEstimator, ClassifierMixin
 from sklearn.utils import check_random_state
 from sklearn.utils.validation import check_is_fitted
 
 from aif360.sklearn.metrics import difference, base_rate
 from aif360.sklearn.metrics import generalized_fnr, generalized_fpr
+from aif360.sklearn.postprocessing import PostProcessingTags
 from aif360.sklearn.utils import check_inputs, check_groups
 
 
@@ -67,8 +70,11 @@ def __init__(self, prot_attr=None, cost_constraint='weighted',
         self.cost_constraint = cost_constraint
         self.random_state = random_state
 
-    def _more_tags(self):
-        return {'requires_proba': True}
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        tags = PostProcessingTags(**asdict(tags))
+        tags.requires_proba = True
+        return tags
 
     def _weighted_cost(self, y_true, probas_pred, pos_label=1,
                        sample_weight=None):

aif360/sklearn/postprocessing/reject_option_classification.py

@@ -1,3 +1,4 @@
+from dataclasses import asdict
 import warnings
 
 import numpy as np
@@ -10,6 +11,7 @@
 from aif360.sklearn.metrics import equal_opportunity_difference
 from aif360.sklearn.metrics import disparate_impact_ratio
 from aif360.sklearn.metrics import make_scorer
+from aif360.sklearn.postprocessing import PostProcessingTags
 from aif360.sklearn.utils import check_groups
 
 
@@ -90,8 +92,11 @@ def __init__(self, prot_attr=None, threshold=0.5, margin=0.1):
         self.threshold = threshold
         self.margin = margin
 
-    def _more_tags(self):
-        return {'requires_proba': True}
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        tags = PostProcessingTags(**asdict(tags))
+        tags.requires_proba = True
+        return tags
 
     def fit(self, X, y, labels=None, pos_label=1, priv_group=1,
             sample_weight=None):
@@ -284,8 +289,11 @@ def __init__(self, prot_attr=None, *, scoring, step=0.05, refit=True, **kwargs):
         super().__init__(RejectOptionClassifier(), {}, scoring=scoring,
                          refit=refit, **kwargs)
 
-    def _more_tags(self):
-        return {'requires_proba': True}
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        tags = PostProcessingTags(**asdict(tags))
+        tags.requires_proba = True
+        return tags
 
     def fit(self, X, y, **fit_params):
         """Run fit with all sets of parameters.

aif360/sklearn/preprocessing/reweighing.py

@@ -1,5 +1,6 @@
 import numpy as np
 from sklearn.base import BaseEstimator, MetaEstimatorMixin, clone
+from sklearn.utils import get_tags
 from sklearn.utils.metaestimators import available_if
 from sklearn.utils.validation import has_fit_parameter
 
@@ -116,9 +117,10 @@ def __init__(self, estimator, reweigher=None):
         self.reweigher = reweigher
         self.estimator = estimator
 
-    @property
-    def _estimator_type(self):
-        return self.estimator._estimator_type
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        tags.estimator_type = get_tags(self.estimator).estimator_type
+        return tags
 
     @property
     def classes_(self):

setup.py

@@ -46,18 +46,18 @@
           "Intended Audience :: Science/Research",
           "License :: OSI Approved :: Apache Software License",
           "Operating System :: OS Independent",
-          "Programming Language :: Python :: 3.8",
-          "Programming Language :: Python :: 3.9",
           "Programming Language :: Python :: 3.10",
           "Programming Language :: Python :: 3.11",
+          "Programming Language :: Python :: 3.12",
+          "Programming Language :: Python :: 3.13",
       ],
       packages=find_packages(include=["aif360*"]),
       python_requires='>=3.8',
       install_requires=[
           'numpy>=1.16',
           'scipy>=1.2.0',
           'pandas>=0.24.0',
-          'scikit-learn>=1.0',
+          'scikit-learn>=1.6',
           'matplotlib',
       ],
       extras_require=extras,