Mark missing values as unsupported

samronsin · samronsin · commit 16a88636b3eb · 2023-06-22T14:05:01.000+02:00
diff --git a/sklearn/tree/_classes.py b/sklearn/tree/_classes.py
@@ -183,7 +183,11 @@ def get_n_leaves(self):
         return self.tree_.n_leaves
 
     def _support_missing_values(self, X):
-        return not issparse(X) and self._get_tags()["allow_nan"]
+        return (
+            not issparse(X)
+            and self._get_tags()["allow_nan"]
+            and self.monotonic_cst is None
+        )
 
     def _compute_missing_values_in_feature_mask(self, X):
         """Return boolean mask denoting if there are missing values for each feature.
diff --git a/sklearn/tree/tests/test_monotonic_tree.py b/sklearn/tree/tests/test_monotonic_tree.py
@@ -3,17 +3,17 @@
 import scipy.sparse
 
 from sklearn.datasets import make_classification, make_regression
+from sklearn.ensemble import (
+    ExtraTreesClassifier,
+    ExtraTreesRegressor,
+    RandomForestClassifier,
+    RandomForestRegressor,
+)
 from sklearn.tree import (
-    DecisionTreeRegressor,
     DecisionTreeClassifier,
-    ExtraTreeRegressor,
+    DecisionTreeRegressor,
     ExtraTreeClassifier,
-)
-from sklearn.ensemble import (
-    RandomForestRegressor,
-    RandomForestClassifier,
-    ExtraTreesRegressor,
-    ExtraTreesClassifier,
+    ExtraTreeRegressor,
 )
 
 TREE_CLASSIFIER_CLASSES = [DecisionTreeClassifier, ExtraTreeClassifier]
@@ -31,7 +31,7 @@
 @pytest.mark.parametrize("TreeClassifier", TREE_BASED_CLASSIFIER_CLASSES)
 @pytest.mark.parametrize("depth_first_builder", (True, False))
 @pytest.mark.parametrize("sparse_splitter", (True, False))
-def test_montonic_constraints_classifications(
+def test_monotonic_constraints_classifications(
     TreeClassifier, depth_first_builder, sparse_splitter, global_random_seed
 ):
     n_samples = 1000
@@ -87,7 +87,7 @@ def test_montonic_constraints_classifications(
 @pytest.mark.parametrize("depth_first_builder", (True, False))
 @pytest.mark.parametrize("sparse_splitter", (True, False))
 @pytest.mark.parametrize("criterion", ("absolute_error", "squared_error"))
-def test_montonic_constraints_regressions(
+def test_monotonic_constraints_regressions(
     TreeRegressor, depth_first_builder, sparse_splitter, criterion, global_random_seed
 ):
     n_samples = 1000
@@ -173,6 +173,25 @@ def test_multiple_output_raises(TreeClassifier):
         est.fit(X, y)
 
 
+@pytest.mark.parametrize(
+    "DecisionTreeEstimator", [DecisionTreeClassifier, DecisionTreeRegressor]
+)
+def test_missing_values_raises(DecisionTreeEstimator):
+    X, y = make_classification(
+        n_samples=100, n_features=5, n_classes=2, n_informative=3, random_state=0
+    )
+    X[0, 0] = np.nan
+    monotonic_cst = np.zeros(X.shape[1])
+    monotonic_cst[0] = 1
+    est = DecisionTreeEstimator(
+        max_depth=None, monotonic_cst=monotonic_cst, random_state=0
+    )
+
+    msg = "Input X contains NaN"
+    with pytest.raises(ValueError, match=msg):
+        est.fit(X, y)
+
+
 @pytest.mark.parametrize("TreeClassifier", TREE_BASED_CLASSIFIER_CLASSES)
 def test_bad_monotonic_cst_raises(TreeClassifier):
     X = [[1, 2], [3, 4], [5, 6], [7, 8], [9, 10]]
