fix: Cache all possible predictions in cache_predictions (#1098)

fixes #1093 

Make sure that we cache all possible type of predictions when calling
`reporter.cache_predictions`. We here check the number of keys in the
dictionary that is enough.

skore/src/skore/sklearn/_estimator/report.py

@@ -163,9 +163,9 @@ def cache_predictions(self, response_methods="auto", n_jobs=None):
             if response_methods == "auto":
                 response_methods = ("predict",)
                 if hasattr(self._estimator, "predict_proba"):
-                    response_methods = ("predict_proba",)
+                    response_methods += ("predict_proba",)
                 if hasattr(self._estimator, "decision_function"):
-                    response_methods = ("decision_function",)
+                    response_methods += ("decision_function",)
             pos_labels = self._estimator.classes_
         else:
             if response_methods == "auto":
@@ -175,8 +175,8 @@ def cache_predictions(self, response_methods="auto", n_jobs=None):
         data_sources = ("test",)
         Xs = (self._X_test,)
         if self._X_train is not None:
-            data_sources = ("train",)
-            Xs = (self._X_train,)
+            data_sources += ("train",)
+            Xs += (self._X_train,)
 
         parallel = joblib.Parallel(n_jobs=n_jobs, return_as="generator_unordered")
         generator = parallel(
@@ -188,8 +188,8 @@ def cache_predictions(self, response_methods="auto", n_jobs=None):
                 pos_label=pos_label,
                 data_source=data_source,
             )
-            for response_method, pos_label, data_source, X in product(
-                response_methods, pos_labels, data_sources, Xs
+            for response_method, pos_label, (data_source, X) in product(
+                response_methods, pos_labels, zip(data_sources, Xs)
             )
         )
         # trigger the computation

skore/tests/unit/sklearn/test_estimator.py

@@ -289,17 +289,33 @@ def test_estimator_report_repr(binary_classification_data):
 
 
 @pytest.mark.parametrize(
-    "fixture_name", ["binary_classification_data", "regression_data"]
+    "fixture_name, pass_train_data, expected_n_keys",
+    [
+        ("binary_classification_data", True, 6),
+        ("binary_classification_data_svc", True, 6),
+        ("multiclass_classification_data", True, 8),
+        ("regression_data", True, 2),
+        ("binary_classification_data", False, 3),
+        ("binary_classification_data_svc", False, 3),
+        ("multiclass_classification_data", False, 4),
+        ("regression_data", False, 1),
+    ],
 )
-def test_estimator_report_cache_predictions(request, fixture_name):
+def test_estimator_report_cache_predictions(
+    request, fixture_name, pass_train_data, expected_n_keys
+):
     """Check that calling cache_predictions fills the cache."""
     estimator, X_test, y_test = request.getfixturevalue(fixture_name)
-    report = EstimatorReport(
-        estimator, X_train=X_test, y_train=y_test, X_test=X_test, y_test=y_test
-    )
+    if pass_train_data:
+        report = EstimatorReport(
+            estimator, X_train=X_test, y_train=y_test, X_test=X_test, y_test=y_test
+        )
+    else:
+        report = EstimatorReport(estimator, X_test=X_test, y_test=y_test)
 
     assert report._cache == {}
     report.cache_predictions()
+    assert len(report._cache) == expected_n_keys
     assert report._cache != {}
     stored_cache = deepcopy(report._cache)
     report.cache_predictions()