Release 1.4.2 (#28774)

Co-authored-by: Guillaume Lemaitre <guillaume@probabl.ai>
Co-authored-by: Loïc Estève <loic.esteve@ymail.com>

doc/templates/index.html

@@ -169,6 +169,8 @@ <h4 class="sk-landing-call-header">News</h4>
         <li><strong>On-going development:</strong>
         <a href="whats_new/v1.5.html#version-1-5-0">scikit-learn 1.5 (Changelog)</a>
         </li>
+        <li><strong>April 2024.</strong> scikit-learn 1.4.2 is available for download (<a href="whats_new/v1.4.html#version-1-4-2">Changelog</a>).
+        </li>
         <li><strong>February 2024.</strong> scikit-learn 1.4.1.post1 is available for download (<a href="whats_new/v1.4.html#version-1-4-1-post1">Changelog</a>).
         </li>
         <li><strong>January 2024.</strong> scikit-learn 1.4.0 is available for download (<a href="whats_new/v1.4.html#version-1-4-0">Changelog</a>).

doc/whats_new/v1.4.rst

@@ -13,6 +13,15 @@ For a short description of the main highlights of the release, please refer to
 
 .. include:: changelog_legend.inc
 
+.. _changes_1_4_2:
+
+Version 1.4.2
+=============
+
+**April 2024**
+
+This release only includes support for numpy 2.
+
 .. _changes_1_4_1:
 
 Version 1.4.1.post1

pyproject.toml

@@ -4,7 +4,7 @@ requires = [
     "setuptools",
     "wheel",
     "Cython>=3.0.8",
-    "numpy>=1.25",
+    "numpy==2.0.0rc1",
     "scipy>=1.6.0",
 ]
 

setup.py

@@ -614,12 +614,6 @@ def setup_package():
         },
     )
 
-    # Overwrite the dependencies to not allow for NumPy >= 2.0
-    metadata["install_requires"] = [
-        f"{dep},<2.0" if dep.startswith("numpy") else dep
-        for dep in metadata["install_requires"]
-    ]
-
     commands = [arg for arg in sys.argv[1:] if not arg.startswith("-")]
     if not all(
         command in ("egg_info", "dist_info", "clean", "check") for command in commands

sklearn/__init__.py

@@ -42,7 +42,7 @@
 # Dev branch marker is: 'X.Y.dev' or 'X.Y.devN' where N is an integer.
 # 'X.Y.dev0' is the canonical version of 'X.Y.dev'
 #
-__version__ = "1.4.1.post1"
+__version__ = "1.4.2"
 
 
 # On OSX, we can get a runtime error due to multiple OpenMP libraries loaded

sklearn/preprocessing/_label.py

@@ -827,7 +827,7 @@ def fit_transform(self, y):
         class_mapping[:] = tmp
         self.classes_, inverse = np.unique(class_mapping, return_inverse=True)
         # ensure yt.indices keeps its current dtype
-        yt.indices = np.array(inverse[yt.indices], dtype=yt.indices.dtype, copy=False)
+        yt.indices = np.asarray(inverse[yt.indices], dtype=yt.indices.dtype)
 
         if not self.sparse_output:
             yt = yt.toarray()

sklearn/preprocessing/tests/test_data.py

@@ -208,7 +208,13 @@ def test_standard_scaler_dtype(add_sample_weight, sparse_container):
     else:
         sample_weight = None
     with_mean = True
-    for dtype in [np.float16, np.float32, np.float64]:
+    if sparse_container is not None:
+        # scipy sparse containers do not support float16, see
+        # https://github.com/scipy/scipy/issues/7408 for more details.
+        supported_dtype = [np.float64, np.float32]
+    else:
+        supported_dtype = [np.float64, np.float32, np.float16]
+    for dtype in supported_dtype:
         X = rng.randn(n_samples, n_features).astype(dtype)
         if sparse_container is not None:
             X = sparse_container(X)

sklearn/utils/fixes.py

@@ -256,14 +256,17 @@ def _sparse_nan_min_max(X, axis):
     from scipy.integrate import trapz as trapezoid  # type: ignore  # noqa
 
 
-# TODO: Remove when Pandas > 2.2 is the minimum supported version
+# TODO: Adapt when Pandas > 2.2 is the minimum supported version
 def pd_fillna(pd, frame):
     pd_version = parse_version(pd.__version__).base_version
     if parse_version(pd_version) < parse_version("2.2"):
         frame = frame.fillna(value=np.nan)
     else:
+        infer_objects_kwargs = (
+            {} if parse_version(pd_version) >= parse_version("3") else {"copy": False}
+        )
         with pd.option_context("future.no_silent_downcasting", True):
-            frame = frame.fillna(value=np.nan).infer_objects(copy=False)
+            frame = frame.fillna(value=np.nan).infer_objects(**infer_objects_kwargs)
     return frame
 
 

sklearn/utils/tests/test_validation.py

@@ -1632,7 +1632,6 @@ def test_check_pandas_sparse_invalid(ntype1, ntype2):
     "ntype1, ntype2, expected_subtype",
     [
         ("double", "longdouble", np.floating),
-        ("float16", "half", np.floating),
         ("single", "float32", np.floating),
         ("double", "float64", np.floating),
         ("int8", "byte", np.integer),