Merge branch 'scikit-learn:main' into submodulev2

Author: adam2392

build_tools/circle/doc_min_dependencies_environment.yml

@@ -23,7 +23,7 @@ dependencies:
   - memory_profiler
   - compilers
   - sphinx=6.0.0  # min
-  - sphinx-gallery=0.7.0  # min
+  - sphinx-gallery=0.10.1  # min
   - sphinx-copybutton=0.5.2  # min
   - numpydoc=1.2.0  # min
   - sphinx-prompt=1.3.0  # min

build_tools/circle/doc_min_dependencies_linux-64_conda.lock

@@ -1,6 +1,6 @@
 # Generated by conda-lock.
 # platform: linux-64
-# input_hash: bd03405f744d4d5d0c59edd290d1e4245873b8593fcd3bbc3efdf8654a283161
+# input_hash: d465abb23248872aed411e2f2f7293a661f0b783f1a84420ffa5431229814cab
 @EXPLICIT
 https://conda.anaconda.org/conda-forge/linux-64/_libgcc_mutex-0.1-conda_forge.tar.bz2#d7c89558ba9fa0495403155b64376d81
 https://conda.anaconda.org/conda-forge/linux-64/ca-certificates-2023.5.7-hbcca054_0.conda#f5c65075fc34438d5b456c7f3f5ab695
@@ -100,7 +100,7 @@ https://conda.anaconda.org/conda-forge/linux-64/markupsafe-2.1.3-py38h01eb140_0.
 https://conda.anaconda.org/conda-forge/noarch/networkx-3.1-pyhd8ed1ab_0.conda#254f787d5068bc89f578bf63893ce8b4
 https://conda.anaconda.org/conda-forge/noarch/packaging-23.1-pyhd8ed1ab_0.conda#91cda59e66e1e4afe9476f8ef98f5c30
 https://conda.anaconda.org/conda-forge/linux-64/pillow-9.4.0-py38hde6dc18_1.conda#3de5619d3f556f966189e5251a266125
-https://conda.anaconda.org/conda-forge/noarch/pluggy-1.0.0-pyhd8ed1ab_5.tar.bz2#7d301a0d25f424d96175f810935f0da9
+https://conda.anaconda.org/conda-forge/noarch/pluggy-1.2.0-pyhd8ed1ab_0.conda#7263924c642d22e311d9e59b839f1b33
 https://conda.anaconda.org/conda-forge/linux-64/psutil-5.9.5-py38h1de0b5d_0.conda#92e899e7b0ed27c793014d1fa54f9b7b
 https://conda.anaconda.org/conda-forge/noarch/py-1.11.0-pyh6c4a22f_0.tar.bz2#b4613d7e7a493916d867842a6a148054
 https://conda.anaconda.org/conda-forge/noarch/pygments-2.15.1-pyhd8ed1ab_0.conda#d316679235612869eba305aa7d41d9bf
@@ -138,15 +138,15 @@ https://conda.anaconda.org/conda-forge/noarch/memory_profiler-0.61.0-pyhd8ed1ab_
 https://conda.anaconda.org/conda-forge/noarch/partd-1.4.0-pyhd8ed1ab_0.conda#721dab5803ea92ce02ddc4ee50aa0c48
 https://conda.anaconda.org/conda-forge/noarch/pip-23.1.2-pyhd8ed1ab_0.conda#7288da0d36821349cf1126e8670292df
 https://conda.anaconda.org/conda-forge/noarch/plotly-5.14.0-pyhd8ed1ab_0.conda#6a7bcc42ef58dd6cf3da9333ea102433
-https://conda.anaconda.org/conda-forge/noarch/pytest-7.3.2-pyhd8ed1ab_1.conda#f2465696f4396245eca4613f6e924796
+https://conda.anaconda.org/conda-forge/noarch/pytest-7.4.0-pyhd8ed1ab_0.conda#3cfe9b9e958e7238a386933c75d190db
 https://conda.anaconda.org/conda-forge/noarch/python-dateutil-2.8.2-pyhd8ed1ab_0.tar.bz2#dd999d1cc9f79e67dbb855c8924c7984
 https://conda.anaconda.org/conda-forge/noarch/typing-extensions-4.6.3-hd8ed1ab_0.conda#3876f650ed7d0f95d70fa4b647621909
 https://conda.anaconda.org/conda-forge/noarch/urllib3-2.0.3-pyhd8ed1ab_0.conda#ae465d0fbf9f1979cb2d8d4043d885e2
 https://conda.anaconda.org/conda-forge/linux-64/gst-plugins-base-1.14.5-h0935bb2_2.tar.bz2#eb125ee86480e00a4a1ed45a577c3311
 https://conda.anaconda.org/conda-forge/noarch/importlib_metadata-6.7.0-hd8ed1ab_0.conda#27a4cec373ec84d1c1aa02a1e37f8eaf
 https://conda.anaconda.org/conda-forge/linux-64/liblapacke-3.8.0-20_mkl.tar.bz2#8274dc30518af9df1de47f5d9e73165c
 https://conda.anaconda.org/conda-forge/linux-64/numpy-1.17.3-py38h95a1406_0.tar.bz2#bc0cbf611fe2f86eab29b98e51404f5e
-https://conda.anaconda.org/conda-forge/noarch/platformdirs-3.6.0-pyhd8ed1ab_0.conda#741384b21c1b512617f4ee4ea8457c5d
+https://conda.anaconda.org/conda-forge/noarch/platformdirs-3.8.0-pyhd8ed1ab_0.conda#3e4aca765371893ad848397794600632
 https://conda.anaconda.org/conda-forge/noarch/pytest-forked-1.6.0-pyhd8ed1ab_0.conda#a46947638b6e005b63d2d6271da529b0
 https://conda.anaconda.org/conda-forge/noarch/requests-2.31.0-pyhd8ed1ab_0.conda#a30144e4156cdbb236f99ebb49828f8b
 https://conda.anaconda.org/conda-forge/linux-64/blas-2.20-mkl.tar.bz2#e7d09a07f5413e53dca5282b8fa50bed
@@ -167,7 +167,7 @@ https://conda.anaconda.org/conda-forge/linux-64/pyqt-5.12.3-py38ha8c2ead_3.tar.b
 https://conda.anaconda.org/conda-forge/linux-64/scikit-image-0.16.2-py38hb3f55d8_0.tar.bz2#468b398fefac8884cd6e6513af66549b
 https://conda.anaconda.org/conda-forge/noarch/seaborn-base-0.12.2-pyhd8ed1ab_0.conda#cf88f3a1c11536bc3c10c14ad00ccc42
 https://conda.anaconda.org/conda-forge/noarch/sphinx-copybutton-0.5.2-pyhd8ed1ab_0.conda#ac832cc43adc79118cf6e23f1f9b8995
-https://conda.anaconda.org/conda-forge/noarch/sphinx-gallery-0.7.0-py_0.tar.bz2#80bad3f857ecc86a4ab73f3e57addd13
+https://conda.anaconda.org/conda-forge/noarch/sphinx-gallery-0.10.1-pyhd8ed1ab_0.tar.bz2#4918585fe5e5341740f7e63c61743efb
 https://conda.anaconda.org/conda-forge/noarch/sphinx-prompt-1.3.0-py_0.tar.bz2#9363002e2a134a287af4e32ff0f26cdc
 https://conda.anaconda.org/conda-forge/linux-64/matplotlib-3.1.3-py38_0.tar.bz2#1992ab91bbff86ded8d99d1f488d8e8b
 https://conda.anaconda.org/conda-forge/linux-64/statsmodels-0.12.2-py38h5c078b8_0.tar.bz2#33787719ad03d33cffc4e2e3ea82bc9e

doc/conf.py

@@ -72,13 +72,15 @@
     import jupyterlite_sphinx  # noqa: F401
 
     extensions.append("jupyterlite_sphinx")
+    with_jupyterlite = True
 except ImportError:
     # In some cases we don't want to require jupyterlite_sphinx to be installed,
     # e.g. the doc-min-dependencies build
     warnings.warn(
         "jupyterlite_sphinx is not installed, you need to install it "
         "if you want JupyterLite links to appear in each example"
     )
+    with_jupyterlite = False
 
 # Produce `plot::` directives for examples that contain `import matplotlib` or
 # `from matplotlib import`.
@@ -526,13 +528,16 @@ def reset_sklearn_config(gallery_conf, fname):
         "dependencies": "./binder/requirements.txt",
         "use_jupyter_lab": True,
     },
-    "jupyterlite": {"notebook_modification_function": notebook_modification_function},
     # avoid generating too many cross links
     "inspect_global_variables": False,
     "remove_config_comments": True,
     "plot_gallery": "True",
     "reset_modules": ("matplotlib", "seaborn", reset_sklearn_config),
 }
+if with_jupyterlite:
+    sphinx_gallery_conf["jupyterlite"] = {
+        "notebook_modification_function": notebook_modification_function
+    }
 
 
 # The following dictionary contains the information used to create the

doc/developers/advanced_installation.rst

@@ -90,7 +90,7 @@ feature, code or documentation improvement).
 
    .. prompt:: bash $
 
-     pip install --verbose --no-use-pep517 --no-build-isolation --editable .
+     pip install -v --no-use-pep517 --no-build-isolation -e .
 
 #. Check that the installed scikit-learn has a version number ending with
    `.dev0`:
@@ -109,7 +109,9 @@ feature, code or documentation improvement).
     (ending in `.pyx` or `.pxd`). This can happen when you edit them or when you
     use certain git commands such as `git pull`. Use the ``--no-build-isolation`` flag
     to avoid compiling the whole project each time, only the files you have
-    modified.
+    modified. Include the ``--no-use-pep517`` flag because the ``--no-build-isolation``
+    option might not work otherwise (this is due to a bug which will be fixed in the
+    future).
 
 Dependencies
 ------------
@@ -242,7 +244,7 @@ Finally, build scikit-learn from this command prompt:
 
 .. prompt:: bash $
 
-    pip install --verbose --no-use-pep517 --no-build-isolation --editable .
+    pip install -v --no-use-pep517 --no-build-isolation -e .
 
 .. _compiler_macos:
 
@@ -284,7 +286,7 @@ scikit-learn from source:
         joblib threadpoolctl pytest compilers llvm-openmp
     conda activate sklearn-dev
     make clean
-    pip install --verbose --no-use-pep517 --no-build-isolation --editable .
+    pip install -v --no-use-pep517 --no-build-isolation -e .
 
 .. note::
 
@@ -364,7 +366,7 @@ Finally, build scikit-learn in verbose mode (to check for the presence of the
 .. prompt:: bash $
 
     make clean
-    pip install --verbose --no-use-pep517 --no-build-isolation --editable .
+    pip install -v --no-use-pep517 --no-build-isolation -e .
 
 .. _compiler_linux:
 
@@ -424,7 +426,7 @@ in the user folder using conda:
     conda create -n sklearn-dev -c conda-forge python numpy scipy cython \
         joblib threadpoolctl pytest compilers
     conda activate sklearn-dev
-    pip install --verbose --no-use-pep517 --no-build-isolation --editable .
+    pip install -v --no-use-pep517 --no-build-isolation -e .
 
 .. _compiler_freebsd:
 
@@ -453,7 +455,7 @@ Finally, build the package using the standard command:
 
 .. prompt:: bash $
 
-    pip install --verbose --no-use-pep517 --no-build-isolation --editable .
+    pip install -v --no-use-pep517 --no-build-isolation -e .
 
 For the upcoming FreeBSD 12.1 and 11.3 versions, OpenMP will be included in
 the base system and these steps will not be necessary.
@@ -514,7 +516,7 @@ and environment variable as follows before calling the ``pip install`` or
 ``python setup.py build_ext`` commands::
 
     export SKLEARN_BUILD_PARALLEL=3
-    pip install --verbose --no-use-pep517 --no-build-isolation --editable .
+    pip install -v --no-use-pep517 --no-build-isolation -e .
 
 On a machine with 2 CPU cores, it can be beneficial to use a parallelism level
 of 3 to overlap IO bound tasks (reading and writing files on disk) with CPU

doc/whats_new/v1.3.rst

@@ -231,6 +231,11 @@ Changelog
   :user:`Jérémie du Boisberranger <jeremiedbb>`,
   :user:`Guillaume Lemaitre <glemaitre>`.
 
+- |Fix| :class:`cluster.KMeans`, :class:`cluster.MiniBatchKMeans` and
+  :func:`cluster.k_means` now correctly handle the combination of `n_init="auto"`
+  and `init` being an array-like, running one initialization in that case.
+  :pr:`26657` by :user:`Binesh Bannerjee <bnsh>`.
+
 - |API| The `sample_weight` parameter in `predict` for
   :meth:`cluster.KMeans.predict` and :meth:`cluster.MiniBatchKMeans.predict`
   is now deprecated and will be removed in v1.5.
@@ -455,6 +460,11 @@ Changelog
   on linearly separable problems.
   :pr:`25214` by `Tom Dupre la Tour`_.
 
+- |Fix| Fix a crash when calling `fit` on
+  :class:`linear_model.LogisticRegression(solver="newton-cholesky", max_iter=0)`
+  which failed to inspect the state of the model prior to the first parameter update.
+  :pr:`26653` by :user:`Olivier Grisel <ogrisel>`.
+
 - |API| Deprecates `n_iter` in favor of `max_iter` in
   :class:`linear_model.BayesianRidge` and :class:`linear_model.ARDRegression`.
   `n_iter` will be removed in scikit-learn 1.5. This change makes those

doc/whats_new/v1.4.rst

@@ -58,3 +58,13 @@ TODO: update at the time of the release.
   :meth:`base.OutlierMixin.fit_predict` now accept ``**kwargs`` which are
   passed to the ``fit`` method of the the estimator. :pr:`26506` by `Adrin
   Jalali`_.
+
+:mod:`sklearn.decomposition`
+............................
+
+- |Enhancement| An "auto" option was added to the `n_components` parameter of
+  :func:`decomposition.non_negative_factorization`, :class:`decomposition.NMF` and
+  :class:`decomposition.MiniBatchNMF` to automatically infer the number of components from W or H shapes
+  when using a custom initialization. The default value of this parameter will change
+  from `None` to `auto` in version 1.6.
+  :pr:`26634` by :user:`Alexandre Landeau <AlexL>` and :user:`Alexandre Vigny <avigny>`.

sklearn/_min_dependencies.py

@@ -44,7 +44,7 @@
     "pyarrow": ("12.0.0", "tests"),
     "sphinx": ("6.0.0", "docs"),
     "sphinx-copybutton": ("0.5.2", "docs"),
-    "sphinx-gallery": ("0.7.0", "docs"),
+    "sphinx-gallery": ("0.10.1", "docs"),
     "numpydoc": ("1.2.0", "docs, tests"),
     "Pillow": ("7.1.2", "docs"),
     "pooch": ("1.6.0", "docs, examples, tests"),

sklearn/cluster/_kmeans.py

@@ -354,7 +354,8 @@ def k_means(
         n_init consecutive runs in terms of inertia.
 
         When `n_init='auto'`, the number of runs depends on the value of init:
-        10 if using `init='random'`, 1 if using `init='k-means++'`.
+        10 if using `init='random'` or `init` is a callable;
+        1 if using `init='k-means++'` or `init` is an array-like.
 
         .. versionadded:: 1.2
            Added 'auto' option for `n_init`.
@@ -884,10 +885,14 @@ def _check_params_vs_input(self, X, default_n_init=None):
             )
             self._n_init = default_n_init
         if self._n_init == "auto":
-            if self.init == "k-means++":
+            if isinstance(self.init, str) and self.init == "k-means++":
                 self._n_init = 1
-            else:
+            elif isinstance(self.init, str) and self.init == "random":
+                self._n_init = default_n_init
+            elif callable(self.init):
                 self._n_init = default_n_init
+            else:  # array-like
+                self._n_init = 1
 
         if _is_arraylike_not_scalar(self.init) and self._n_init != 1:
             warnings.warn(
@@ -1241,7 +1246,8 @@ class KMeans(_BaseKMeans):
         high-dimensional problems (see :ref:`kmeans_sparse_high_dim`).
 
         When `n_init='auto'`, the number of runs depends on the value of init:
-        10 if using `init='random'`, 1 if using `init='k-means++'`.
+        10 if using `init='random'` or `init` is a callable;
+        1 if using `init='k-means++'` or `init` is an array-like.
 
         .. versionadded:: 1.2
            Added 'auto' option for `n_init`.
@@ -1777,7 +1783,8 @@ class MiniBatchKMeans(_BaseKMeans):
         :ref:`kmeans_sparse_high_dim`).
 
         When `n_init='auto'`, the number of runs depends on the value of init:
-        3 if using `init='random'`, 1 if using `init='k-means++'`.
+        3 if using `init='random'` or `init` is a callable;
+        1 if using `init='k-means++'` or `init` is an array-like.
 
         .. versionadded:: 1.2
            Added 'auto' option for `n_init`.

sklearn/cluster/tests/test_k_means.py

@@ -348,6 +348,37 @@ def test_minibatch_kmeans_partial_fit_init(init):
     _check_fitted_model(km)
 
 
+@pytest.mark.parametrize(
+    "init, expected_n_init",
+    [
+        ("k-means++", 1),
+        ("random", "default"),
+        (
+            lambda X, n_clusters, random_state: random_state.uniform(
+                size=(n_clusters, X.shape[1])
+            ),
+            "default",
+        ),
+        ("array-like", 1),
+    ],
+)
+@pytest.mark.parametrize("Estimator", [KMeans, MiniBatchKMeans])
+def test_kmeans_init_auto_with_initial_centroids(Estimator, init, expected_n_init):
+    """Check that `n_init="auto"` chooses the right number of initializations.
+    Non-regression test for #26657:
+    https://github.com/scikit-learn/scikit-learn/pull/26657
+    """
+    n_sample, n_features, n_clusters = 100, 10, 5
+    X = np.random.randn(n_sample, n_features)
+    if init == "array-like":
+        init = np.random.randn(n_clusters, n_features)
+    if expected_n_init == "default":
+        expected_n_init = 3 if Estimator is MiniBatchKMeans else 10
+
+    kmeans = Estimator(n_clusters=n_clusters, init=init, n_init="auto").fit(X)
+    assert kmeans._n_init == expected_n_init
+
+
 @pytest.mark.parametrize("Estimator", [KMeans, MiniBatchKMeans])
 def test_fortran_aligned_data(Estimator, global_random_seed):
     # Check that KMeans works with fortran-aligned data.