fix bug in shapexplainer with native multioutput support and explaini… (

#2428)

* fix bug in shapexplainer with native multioutput support and explaining only selected target components

* update changelog

CHANGELOG.md

@@ -13,6 +13,7 @@ but cannot always guarantee backwards compatibility. Changes that may **break co
 **Fixed**
 - Fixed a bug when using `historical_forecasts()` with a pre-trained `RegressionModel` that has no target lags `lags=None` but uses static covariates. [#2426](https://github.com/unit8co/darts/pull/2426) by [Dennis Bader](https://github.com/dennisbader).
 - Fixed a bug with `xgboost>=2.1.0`, where multi output regression was not properly handled. [#2426](https://github.com/unit8co/darts/pull/2426) by [Dennis Bader](https://github.com/dennisbader).
+- Fixed a bug when using `ShapExplainer.explain()` with some selected `target_components` and a regression model that natively supports multi output regression: The target components were not properly mapped. [#2428](https://github.com/unit8co/darts/pull/2428) by [Dennis Bader](https://github.com/dennisbader).
 
 **Dependencies**
 

darts/explainability/shap_explainer.py

@@ -658,7 +658,7 @@ def shap_explanations(
             shap_explanation_tmp = self.explainers(foreground_X)
             for h in horizons:
                 tmp_n = {}
-                for t_idx, t in enumerate(target_components):
+                for t_idx, t in enumerate(self.target_components):
                     if t not in target_components:
                         continue
                     if not self.single_output:

darts/tests/explainability/test_shap_explainer.py

@@ -205,9 +205,13 @@ def test_creation(self):
 
         # Good type of explainers
         shap_explain = ShapExplainer(m)
-        assert isinstance(
-            shap_explain.explainers.explainers[0][0], shap.explainers.Tree
-        )
+        if isinstance(m, XGBModel):
+            # since xgboost > 2.1.0, model supports native multi output regression
+            assert isinstance(shap_explain.explainers.explainers, shap.explainers.Tree)
+        else:
+            assert isinstance(
+                shap_explain.explainers.explainers[0][0], shap.explainers.Tree
+            )
 
         # Linear model - also not a MultiOutputRegressor
         m = LinearRegressionModel(
@@ -266,9 +270,12 @@ def test_creation(self):
             future_covariates=self.fut_cov_ts,
         )
         shap_explain = ShapExplainer(m)
-        assert isinstance(
-            shap_explain.explainers.explainers[0][0], shap.explainers.Tree
-        )
+        if isinstance(m, XGBModel):
+            assert isinstance(shap_explain.explainers.explainers, shap.explainers.Tree)
+        else:
+            assert isinstance(
+                shap_explain.explainers.explainers[0][0], shap.explainers.Tree
+            )
 
         # Bad choice of shap explainer
         with pytest.raises(ValueError):
@@ -709,13 +716,26 @@ def test_shap_explanation_object_validity(self):
             shap.Explanation,
         )
 
-    def test_shap_selected_components(self):
-        model_cls = LightGBMModel if lgbm_available else XGBModel
+    @pytest.mark.parametrize(
+        "config",
+        [
+            (XGBModel, {}),
+            (
+                LightGBMModel if lgbm_available else XGBModel,
+                {"likelihood": "quantile", "quantiles": [0.5]},
+            ),
+        ],
+    )
+    def test_shap_selected_components(self, config):
+        """Test selected components with and without Darts' MultiOutputRegressor"""
+        model_cls, model_kwargs = config
+        # model_cls = XGBModel
         model = model_cls(
             lags=4,
             lags_past_covariates=2,
             lags_future_covariates=[1],
             output_chunk_length=1,
+            **model_kwargs,
         )
         model.fit(
             series=self.target_ts,