FIX Raise TypeError when sparse input is not supported

python/cuml/cuml/internals/array.py

@@ -983,12 +983,12 @@ def from_input(
 
         if is_sparse(X):
             # We don't support coercing sparse arrays to dense via this method.
-            # Raising a NotImplementedError here lets us nicely error
-            # for estimators that don't support sparse arrays without requiring
-            # an additional external check. Otherwise they'd get an opaque error
-            # for code below.
-            raise NotImplementedError(
-                "Sparse inputs are not currently supported for this method"
+            # Raising a TypeError here lets us nicely error for estimators that
+            # don't support sparse arrays without requiring an additional
+            # external check. Using TypeError with "sparse" in the message
+            # satisfies sklearn's check_estimator_sparse_tag check.
+            raise TypeError(
+                "A sparse matrix was passed, but dense data is required. "
             )
         if convert_to_mem_type is not False:
             convert_to_mem_type = MemoryType.from_str(convert_to_mem_type)

python/cuml/tests/test_exceptions.py

@@ -1,5 +1,5 @@
 #
-# SPDX-FileCopyrightText: Copyright (c) 2024-2025, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION.
 # SPDX-License-Identifier: Apache-2.0
 #
 
@@ -24,7 +24,7 @@
     NearestNeighbors,
 )
 
-# Currently only certain estimators raise a NotImplementedError
+# Estimators that raise TypeError when given sparse input (they don't support sparse)
 estimators = {
     "KMeans": lambda: KMeans(n_clusters=2, random_state=0),
     "DBSCAN": lambda: DBSCAN(eps=1.0),
@@ -44,7 +44,7 @@ def test_sparse_not_implemented_exception(estimator_name):
     y_reg = np.array([0.0, 1.0])
     estimator = estimators[estimator_name]()
     # Fit or fit_transform depending on the estimator type
-    with pytest.raises(NotImplementedError):
+    with pytest.raises(TypeError, match="sparse"):
         if isinstance(
             estimator, (KMeans, DBSCAN, TruncatedSVD, NearestNeighbors)
         ):