FIX Raise error when missing-values encountered in scikit-tree trees (#…

…264)

* Implement smoke test

---------

Signed-off-by: Adam Li <adam2392@gmail.com>

.github/workflows/main.yml

@@ -61,6 +61,7 @@ jobs:
         run: |
           brew install ccache
           brew install gcc
+          brew install gettext
 
       - name: show-gcc
         run: |

build_requirements.txt

@@ -1,9 +1,9 @@
-meson
-meson-python
-cython>=3.0.8
+meson>=1.4.0
+meson-python>=0.16.0
+cython>=3.0.10
 ninja
 numpy
-scikit-learn>=1.4.1
+scikit-learn>=1.4.2
 click
 rich-click
 doit

doc/whats_new/v0.8.rst

@@ -13,6 +13,12 @@ Version 0.8
 Changelog
 ---------
 
+- |Fix| Previously missing-values in ``X`` input array for sktree estimators
+    did not raise an error, and silently ran, assuming the missing-values were
+    encoded as infinity value. This is now fixed, and the estimators will raise an
+    ValueError if missing-values are encountered in ``X`` input array.
+    By `Adam Li`_ (:pr:`#264`)
+
 Code and Documentation Contributors
 -----------------------------------
 

pyproject.toml

@@ -1,15 +1,15 @@
 [build-system]
 build-backend = "mesonpy"
 requires = [
-  "meson-python>=0.15.0",
+  "meson-python>=0.16.0",
   'ninja',
   # `wheel` is needed for non-isolated builds, given that `meson-python`
   # doesn't list it as a runtime requirement (at least in 0.10.0)
   # See https://github.com/FFY00/meson-python/blob/main/pyproject.toml#L4
   "wheel",
   "setuptools<=65.5",
   "packaging",
-  "Cython>=3.0.8",
+  "Cython>=3.0.10",
   "scikit-learn>=1.4.1",
   "scipy>=1.5.0",
   "numpy>=1.25; python_version>='3.9'"

sktree/tree/_neighbors.py

@@ -64,3 +64,7 @@ def compute_similarity_matrix(self, X):
             The similarity matrix among the samples.
         """
         return compute_forest_similarity_matrix(self, X)
+
+    def _more_tags(self):
+        # XXX: no scikit-tree estimators support NaNs as of now
+        return {"allow_nan": False}

sktree/tree/tests/test_all_trees.py

@@ -3,7 +3,7 @@
 import pytest
 from numpy.testing import assert_almost_equal, assert_array_equal
 from sklearn.base import is_classifier
-from sklearn.datasets import make_blobs
+from sklearn.datasets import load_iris, make_blobs
 from sklearn.tree._tree import TREE_LEAF
 
 from sktree.tree import (
@@ -162,3 +162,24 @@ def test_similarity_matrix(tree):
 
     assert np.allclose(sim_mat, sim_mat.T)
     assert np.all((sim_mat.diagonal() == 1))
+
+
+@pytest.mark.parametrize("tree", ALL_TREES)
+def test_missing_values(tree):
+    """Smoke test to ensure that correct error is raised when missing values are present.
+
+    xref: https://github.com/neurodata/scikit-tree/issues/263
+    """
+    rng = np.random.default_rng(123)
+
+    iris_X, iris_y = load_iris(return_X_y=True, as_frame=True)
+
+    # Make the feature matrix 25% sparse
+    iris_X = iris_X.mask(rng.standard_normal(iris_X.shape) < 0.25)
+
+    classifier = tree()
+    with pytest.raises(ValueError, match="Input X contains NaN"):
+        if tree.__name__.startswith("Unsupervised"):
+            classifier.fit(iris_X)
+        else:
+            classifier.fit(iris_X, iris_y)