FIX Work around likely SIMD issue in tree export on 32bit OS (#29327)

lesteve · mr-c · gspr · web-flow · commit a67ebbebc173 · 2024-06-21T18:36:44.000+02:00
Co-authored-by: Michael R. Crusoe &lt;1330696+mr-c@users.noreply.github.com&gt;
Co-authored-by: SGard Spreemann &lt;gspr@nonempty.org&gt;
Co-authored-by: Sergio Pascual &lt;sergiopr@fis.ucm.es&gt;
diff --git a/doc/whats_new/v1.5.rst b/doc/whats_new/v1.5.rst
@@ -56,6 +56,13 @@ Changelog
   grids that have estimators as parameter values.
   :pr:`29179` by :user:`Marco Gorelli<MarcoGorelli>`.
 
+:mod:`sklearn.tree`
+...................
+
+- |Fix| Fix an issue in :func:`tree.export_graphviz` and :func:`tree.plot_tree`
+  that could potentially result in exception or wrong results on 32bit OSes.
+  :pr:`` by :user:`Loïc Estève<lesteve>`.
+
 :mod:`sklearn.utils`
 ....................
 
diff --git a/sklearn/tree/_export.py b/sklearn/tree/_export.py
@@ -259,7 +259,12 @@ def get_fill_color(self, tree, node_id):
             self.colors["rgb"] = _color_brew(tree.n_classes[0])
             if tree.n_outputs != 1:
                 # Find max and min impurities for multi-output
-                self.colors["bounds"] = (np.min(-tree.impurity), np.max(-tree.impurity))
+                # The next line uses -max(impurity) instead of min(-impurity)
+                # and -min(impurity) instead of max(-impurity) on purpose, in
+                # order to avoid what looks like an issue with SIMD on non
+                # memory aligned arrays on 32bit OS. For more details see
+                # https://github.com/scikit-learn/scikit-learn/issues/27506.
+                self.colors["bounds"] = (-np.max(tree.impurity), -np.min(tree.impurity))
             elif tree.n_classes[0] == 1 and len(np.unique(tree.value)) != 1:
                 # Find max and min values in leaf nodes for regression
                 self.colors["bounds"] = (np.min(tree.value), np.max(tree.value))
