scikit-learn · leweex95 · Dec 8, 2025 · Dec 8, 2025 · Dec 8, 2025 · Dec 8, 2025
diff --git a/doc/whats_new/upcoming_changes/sklearn.inspection/32867.fix.rst b/doc/whats_new/upcoming_changes/sklearn.inspection/32867.fix.rst
@@ -0,0 +1 @@
+- :class:`inspection.DecisionBoundaryDisplay` now displays distinct colors for all classes when using ``plot_method="contour"`` with ``response_method="predict"``, and when using ``plot_method="contourf"`` with ``response_method="predict_proba"`` or ``"decision_function"``. By :user:`Levente Csibi <leweex95>`. :pr:`32867`
diff --git a/sklearn/inspection/_plot/decision_boundary.py b/sklearn/inspection/_plot/decision_boundary.py
@@ -217,6 +217,21 @@ def plot(self, plot_method="contourf", ax=None, xlabel=None, ylabel=None, **kwar
 
         plot_func = getattr(ax, plot_method)
         if self.response.ndim == 2:
+            # For discrete responses (e.g., from predict), ensure all levels are used
+            # to display distinct colors for each class
+            if plot_method in ("contour", "contourf") and np.issubdtype(
+                self.response.dtype, np.integer
+            ):
+                unique_levels = np.unique(self.response)
+                if plot_method == "contourf":
+                    levels = np.concatenate(
+                        [unique_levels - 0.5, [unique_levels.max() + 0.5]]
+                    )
+                else:
+                    levels = unique_levels
+
+                if "levels" not in kwargs:
+                    kwargs["levels"] = levels
             self.surface_ = plot_func(self.xx0, self.xx1, self.response, **kwargs)
         else:  # self.response.ndim == 3
             n_responses = self.response.shape[-1]
@@ -260,6 +275,9 @@ def plot(self, plot_method="contourf", ax=None, xlabel=None, ylabel=None, **kwar
             if plot_method == "contour":
                 # Plot only argmax map for contour
                 class_map = self.response.argmax(axis=2)
+                # Ensure levels match the number of classes for distinct colors
+                if "levels" not in kwargs:
+                    kwargs["levels"] = np.unique(class_map)
                 self.surface_ = plot_func(
                     self.xx0, self.xx1, class_map, colors=colors, **kwargs
                 )

diff --git a/sklearn/inspection/_plot/tests/test_boundary_decision_display.py b/sklearn/inspection/_plot/tests/test_boundary_decision_display.py
@@ -708,3 +708,48 @@ class SubclassOfDisplay(DecisionBoundaryDisplay):
     curve = SubclassOfDisplay.from_estimator(estimator=clf, X=X)
 
     assert isinstance(curve, SubclassOfDisplay)
+
+
+@pytest.mark.parametrize("plot_method", ["contourf", "contour"])
+@pytest.mark.parametrize(
+    "response_method", ["predict", "predict_proba", "decision_function"]
+)
+@pytest.mark.parametrize("y_type", ["int", "str"])
+def test_decision_boundary_display_many_classes(
+    pyplot, plot_method, response_method, y_type
+):
+    """Check that contour plots use all levels for classifiers with many classes.
+
+    Non-regression test for:
+    https://github.com/scikit-learn/scikit-learn/issues/32866
+    """
+    # Create a dataset with 11 classes
+    pts = np.array(
+        [
+            [-1, -1],
+            [-2, -1],
+            [1, 1],
+            [2, 1],
+            [2, 2],
+            [3, 2],
+            [3, 3],
+            [4, 3],
+            [4, 4],
+            [5, 4],
+            [5, 5],
+        ]
+    )
+    if y_type == "int":
+        y = np.arange(11)
+    else:
+        y = [str(i) for i in range(11)]
+    clf = LogisticRegression().fit(pts, y)
+
+    disp = DecisionBoundaryDisplay.from_estimator(
+        clf, pts, response_method=response_method, plot_method=plot_method
+    )
+
+    # Check that the surface has levels for all classes when applicable
+    if hasattr(disp.surface_, "levels"):
+        expected_levels = 12 if plot_method == "contourf" else 11
+        assert len(disp.surface_.levels) == expected_levels
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1 @@
		- :class:`inspection.DecisionBoundaryDisplay` now displays distinct colors for all classes when using ``plot_method="contour"`` with ``response_method="predict"``, and when using ``plot_method="contourf"`` with ``response_method="predict_proba"`` or ``"decision_function"``. By :user:`Levente Csibi <leweex95>`. :pr:`32867`