Merge pull request #10 from PrimozGodec/fix-calibration

Temporarily fix explainer to work with calibration modelom

orangecontrib/explain/explainer.py

@@ -102,9 +102,7 @@ def _explain_trees(
     for i in range(0, len(data_sample), batch_size):
         progress_callback(i / len(data_sample))
         batch = data_sample.X[i : i + batch_size]
-        shap_values.append(
-            explainer.shap_values(batch, check_additivity=False)
-        )
+        shap_values.append(explainer.shap_values(batch, check_additivity=False))
 
     shap_values = _join_shap_values(shap_values)
     base_value = explainer.expected_value
@@ -152,7 +150,9 @@ def _explain_other_models(
     for i, row in enumerate(data_sample.X):
         progress_callback(i / len(data_sample))
         shap_values.append(
-            explainer.shap_values(row, nsamples=100, silent=True, l1_reg="num_features(90)")
+            explainer.shap_values(
+                row, nsamples=100, silent=True, l1_reg="num_features(90)"
+            )
         )
     return (
         _join_shap_values(shap_values),
@@ -205,8 +205,24 @@ def compute_shap_values(
             progress_callback = dummy_callback
         progress_callback(0, "Computing explanation ...")
 
-        data_transformed = model.data_to_model_domain(data)
-        reference_data_transformed = model.data_to_model_domain(reference_data)
+        #### workaround for bug with calibration
+        #### remove when fixed
+        from Orange.classification import (
+            ThresholdClassifier,
+            CalibratedClassifier,
+        )
+
+        trans_model = model
+        while isinstance(
+            trans_model, (ThresholdClassifier, CalibratedClassifier)
+        ):
+            trans_model = trans_model.base_model
+        #### end of workaround for bug with calibration
+
+        data_transformed = trans_model.data_to_model_domain(data)
+        reference_data_transformed = trans_model.data_to_model_domain(
+            reference_data
+        )
 
         shap_values, sample_mask, base_value = _explain_trees(
             model,
@@ -422,7 +438,8 @@ def explain_predictions(
         Domain(data.domain.attributes, None, data.domain.metas)
     )
     predictions = model(
-        classless_data, model.Probs if model.domain.class_var.is_discrete else model.Value
+        classless_data,
+        model.Probs if model.domain.class_var.is_discrete else model.Value,
     )
     # for regression - predictions array is 1d transform it shape N x 1
     if predictions.ndim == 1:

orangecontrib/explain/tests/test_explainer.py

@@ -1,18 +1,20 @@
+import inspect
 import unittest
 
 import numpy as np
+import pkg_resources
 
 from Orange.classification import (
     LogisticRegressionLearner,
     RandomForestLearner,
     SGDClassificationLearner,
     SVMLearner,
     TreeLearner,
+    ThresholdLearner,
 )
 from Orange.data import Table, Domain
 from Orange.regression import LinearRegressionLearner
-from Orange.tests.test_classification import LearnerAccessibility
-from Orange.tests import test_regression
+from Orange.tests import test_regression, test_classification
 from Orange.widgets.data import owcolor
 from orangecontrib.explain.explainer import (
     compute_colors,
@@ -159,12 +161,17 @@ def test_class_not_predicted(self):
         # missing class has all shap values 0
         self.assertTrue(not np.any(shap_values[2].sum()))
 
-    @unittest.skip("Enable when learners fixed")
     def test_all_classifiers(self):
         """ Test explanation for all classifiers """
-        for learner in LearnerAccessibility.all_learners(None):
+        for learner in test_classification.all_learners():
             with self.subTest(learner.name):
-                model = learner(self.iris)
+                if learner == ThresholdLearner:
+                    # ThresholdLearner require binary class
+                    continue
+                kwargs = {}
+                if "base_learner" in inspect.signature(learner).parameters:
+                    kwargs = {"base_learner": LogisticRegressionLearner()}
+                model = learner(**kwargs)(self.iris)
                 shap_values, _, _, _ = compute_shap_values(
                     model, self.iris, self.iris
                 )
@@ -176,7 +183,7 @@ def test_all_classifiers(self):
 
     @unittest.skipIf(
         not hasattr(test_regression, "all_learners"),
-        "all_learners not available in Orange < 3.26"
+        "all_learners not available in Orange < 3.26",
     )
     def test_all_regressors(self):
         """ Test explanation for all regressors """
@@ -569,6 +576,16 @@ def test_no_class(self):
         self.assertTupleEqual(self.iris.X.shape, shap_values[0].shape)
         self.assertTupleEqual((len(self.iris),), sample_mask.shape)
 
+    def test_remove_calibration_workaround(self):
+        """
+        When this test start to fail remove the workaround in
+        explainer.py-207:220 if allready fixed - revert the pullrequest
+        that adds those lines.
+        """
+        self.assertGreater(
+            "3.29.0", pkg_resources.get_distribution("orange3").version
+        )
+
 
 if __name__ == "__main__":
     unittest.main()