ENH Improve error message for sparse multilabel-indicator y in RandomForestClassifier (scikit-learn#15971)

Author: rushabh-v

sklearn/ensemble/_forest.py

@@ -292,11 +292,16 @@ def fit(self, X, y, sample_weight=None):
         -------
         self : object
         """
-        # Validate or convert input data
+        # Validate and convert input data
+        if issparse(y):
+            raise ValueError(
+                "sparse multilabel-indicator for y is not supported."
+                )
         X = check_array(X, accept_sparse="csc", dtype=DTYPE)
-        y = check_array(y, accept_sparse='csc', ensure_2d=False, dtype=None)
+        y = check_array(y, ensure_2d=False, dtype=None)
         if sample_weight is not None:
             sample_weight = _check_sample_weight(sample_weight, X)
+
         if issparse(X):
             # Pre-sort indices to avoid that each individual tree of the
             # ensemble sorts the indices.

sklearn/ensemble/tests/test_forest.py

@@ -1336,6 +1336,15 @@ def test_max_samples_exceptions(name, max_samples, exc_type, exc_msg):
         est.fit(X, y)
 
 
+def test_forest_y_sparse():
+    X = [[1, 2, 3]]
+    y = csr_matrix([4, 5, 6])
+    est = RandomForestClassifier()
+    msg = "sparse multilabel-indicator for y is not supported."
+    with pytest.raises(ValueError, match=msg):
+        est.fit(X, y)
+
+
 @pytest.mark.parametrize(
     'ForestClass', [RandomForestClassifier, RandomForestRegressor]
 )