sklearn/doc/whats_new/v1.3.rst

1004 lines
44 KiB
ReStructuredText

.. include:: _contributors.rst
.. currentmodule:: sklearn
.. _release_notes_1_3:
===========
Version 1.3
===========
For a short description of the main highlights of the release, please refer to
:ref:`sphx_glr_auto_examples_release_highlights_plot_release_highlights_1_3_0.py`.
.. include:: changelog_legend.inc
.. _changes_1_3_2:
Version 1.3.2
=============
**October 2023**
Changelog
---------
:mod:`sklearn.datasets`
.......................
- |Fix| All dataset fetchers now accept `data_home` as any object that implements
the :class:`os.PathLike` interface, for instance, :class:`pathlib.Path`.
:pr:`27468` by :user:`Yao Xiao <Charlie-XIAO>`.
:mod:`sklearn.decomposition`
............................
- |Fix| Fixes a bug in :class:`decomposition.KernelPCA` by forcing the output of
the internal :class:`preprocessing.KernelCenterer` to be a default array. When the
arpack solver is used, it expects an array with a `dtype` attribute.
:pr:`27583` by :user:`Guillaume Lemaitre <glemaitre>`.
:mod:`sklearn.metrics`
......................
- |Fix| Fixes a bug for metrics using `zero_division=np.nan`
(e.g. :func:`~metrics.precision_score`) within a paralell loop
(e.g. :func:`~model_selection.cross_val_score`) where the singleton for `np.nan`
will be different in the sub-processes.
:pr:`27573` by :user:`Guillaume Lemaitre <glemaitre>`.
:mod:`sklearn.tree`
...................
- |Fix| Do not leak data via non-initialized memory in decision tree pickle files and make
the generation of those files deterministic. :pr:`27580` by :user:`Loïc Estève <lesteve>`.
.. _changes_1_3_1:
Version 1.3.1
=============
**September 2023**
Changed models
--------------
The following estimators and functions, when fit with the same data and
parameters, may produce different models from the previous version. This often
occurs due to changes in the modelling logic (bug fixes or enhancements), or in
random sampling procedures.
- |Fix| Ridge models with `solver='sparse_cg'` may have slightly different
results with scipy>=1.12, because of an underlying change in the scipy solver
(see `scipy#18488 <https://github.com/scipy/scipy/pull/18488>`_ for more
details)
:pr:`26814` by :user:`Loïc Estève <lesteve>`
Changes impacting all modules
-----------------------------
- |Fix| The `set_output` API correctly works with list input. :pr:`27044` by
`Thomas Fan`_.
Changelog
---------
:mod:`sklearn.calibration`
..........................
- |Fix| :class:`calibration.CalibratedClassifierCV` can now handle models that
produce large prediction scores. Before it was numerically unstable.
:pr:`26913` by :user:`Omar Salman <OmarManzoor>`.
:mod:`sklearn.cluster`
......................
- |Fix| :class:`cluster.BisectingKMeans` could crash when predicting on data
with a different scale than the data used to fit the model.
:pr:`27167` by `Olivier Grisel`_.
- |Fix| :class:`cluster.BisectingKMeans` now works with data that has a single feature.
:pr:`27243` by :user:`Jérémie du Boisberranger <jeremiedbb>`.
:mod:`sklearn.cross_decomposition`
..................................
- |Fix| :class:`cross_decomposition.PLSRegression` now automatically ravels the output
of `predict` if fitted with one dimensional `y`.
:pr:`26602` by :user:`Yao Xiao <Charlie-XIAO>`.
:mod:`sklearn.ensemble`
.......................
- |Fix| Fix a bug in :class:`ensemble.AdaBoostClassifier` with `algorithm="SAMME"`
where the decision function of each weak learner should be symmetric (i.e.
the sum of the scores should sum to zero for a sample).
:pr:`26521` by :user:`Guillaume Lemaitre <glemaitre>`.
:mod:`sklearn.feature_selection`
................................
- |Fix| :func:`feature_selection.mutual_info_regression` now correctly computes the
result when `X` is of integer dtype. :pr:`26748` by :user:`Yao Xiao <Charlie-XIAO>`.
:mod:`sklearn.impute`
.....................
- |Fix| :class:`impute.KNNImputer` now correctly adds a missing indicator column in
``transform`` when ``add_indicator`` is set to ``True`` and missing values are observed
during ``fit``. :pr:`26600` by :user:`Shreesha Kumar Bhat <Shreesha3112>`.
:mod:`sklearn.metrics`
......................
- |Fix| Scorers used with :func:`metrics.get_scorer` handle properly
multilabel-indicator matrix.
:pr:`27002` by :user:`Guillaume Lemaitre <glemaitre>`.
:mod:`sklearn.mixture`
......................
- |Fix| The initialization of :class:`mixture.GaussianMixture` from user-provided
`precisions_init` for `covariance_type` of `full` or `tied` was not correct,
and has been fixed.
:pr:`26416` by :user:`Yang Tao <mchikyt3>`.
:mod:`sklearn.neighbors`
........................
- |Fix| :meth:`neighbors.KNeighborsClassifier.predict` no longer raises an
exception for `pandas.DataFrames` input.
:pr:`26772` by :user:`Jérémie du Boisberranger <jeremiedbb>`.
- |Fix| Reintroduce `sklearn.neighbors.BallTree.valid_metrics` and
`sklearn.neighbors.KDTree.valid_metrics` as public class attributes.
:pr:`26754` by :user:`Julien Jerphanion <jjerphan>`.
- |Fix| :class:`sklearn.model_selection.HalvingRandomSearchCV` no longer raises
when the input to the `param_distributions` parameter is a list of dicts.
:pr:`26893` by :user:`Stefanie Senger <StefanieSenger>`.
- |Fix| Neighbors based estimators now correctly work when `metric="minkowski"` and the
metric parameter `p` is in the range `0 < p < 1`, regardless of the `dtype` of `X`.
:pr:`26760` by :user:`Shreesha Kumar Bhat <Shreesha3112>`.
:mod:`sklearn.preprocessing`
............................
- |Fix| :class:`preprocessing.LabelEncoder` correctly accepts `y` as a keyword
argument. :pr:`26940` by `Thomas Fan`_.
- |Fix| :class:`preprocessing.OneHotEncoder` shows a more informative error message
when `sparse_output=True` and the output is configured to be pandas.
:pr:`26931` by `Thomas Fan`_.
:mod:`sklearn.tree`
...................
- |Fix| :func:`tree.plot_tree` now accepts `class_names=True` as documented.
:pr:`26903` by :user:`Thomas Roehr <2maz>`
- |Fix| The `feature_names` parameter of :func:`tree.plot_tree` now accepts any kind of
array-like instead of just a list. :pr:`27292` by :user:`Rahil Parikh <rprkh>`.
.. _changes_1_3:
Version 1.3.0
=============
**June 2023**
Changed models
--------------
The following estimators and functions, when fit with the same data and
parameters, may produce different models from the previous version. This often
occurs due to changes in the modelling logic (bug fixes or enhancements), or in
random sampling procedures.
- |Enhancement| :meth:`multiclass.OutputCodeClassifier.predict` now uses a more
efficient pairwise distance reduction. As a consequence, the tie-breaking
strategy is different and thus the predicted labels may be different.
:pr:`25196` by :user:`Guillaume Lemaitre <glemaitre>`.
- |Enhancement| The `fit_transform` method of :class:`decomposition.DictionaryLearning`
is more efficient but may produce different results as in previous versions when
`transform_algorithm` is not the same as `fit_algorithm` and the number of iterations
is small. :pr:`24871` by :user:`Omar Salman <OmarManzoor>`.
- |Enhancement| The `sample_weight` parameter now will be used in centroids
initialization for :class:`cluster.KMeans`, :class:`cluster.BisectingKMeans`
and :class:`cluster.MiniBatchKMeans`.
This change will break backward compatibility, since numbers generated
from same random seeds will be different.
:pr:`25752` by :user:`Gleb Levitski <glevv>`,
:user:`Jérémie du Boisberranger <jeremiedbb>`,
:user:`Guillaume Lemaitre <glemaitre>`.
- |Fix| Treat more consistently small values in the `W` and `H` matrices during the
`fit` and `transform` steps of :class:`decomposition.NMF` and
:class:`decomposition.MiniBatchNMF` which can produce different results than previous
versions. :pr:`25438` by :user:`Yotam Avidar-Constantini <yotamcons>`.
- |Fix| :class:`decomposition.KernelPCA` may produce different results through
`inverse_transform` if `gamma` is `None`. Now it will be chosen correctly as
`1/n_features` of the data that it is fitted on, while previously it might be
incorrectly chosen as `1/n_features` of the data passed to `inverse_transform`.
A new attribute `gamma_` is provided for revealing the actual value of `gamma`
used each time the kernel is called.
:pr:`26337` by :user:`Yao Xiao <Charlie-XIAO>`.
Changed displays
----------------
- |Enhancement| :class:`model_selection.LearningCurveDisplay` displays both the
train and test curves by default. You can set `score_type="test"` to keep the
past behaviour.
:pr:`25120` by :user:`Guillaume Lemaitre <glemaitre>`.
- |Fix| :class:`model_selection.ValidationCurveDisplay` now accepts passing a
list to the `param_range` parameter.
:pr:`27311` by :user:`Arturo Amor <ArturoAmorQ>`.
Changes impacting all modules
-----------------------------
- |Enhancement| The `get_feature_names_out` method of the following classes now
raises a `NotFittedError` if the instance is not fitted. This ensures the error is
consistent in all estimators with the `get_feature_names_out` method.
- :class:`impute.MissingIndicator`
- :class:`feature_extraction.DictVectorizer`
- :class:`feature_extraction.text.TfidfTransformer`
- :class:`feature_selection.GenericUnivariateSelect`
- :class:`feature_selection.RFE`
- :class:`feature_selection.RFECV`
- :class:`feature_selection.SelectFdr`
- :class:`feature_selection.SelectFpr`
- :class:`feature_selection.SelectFromModel`
- :class:`feature_selection.SelectFwe`
- :class:`feature_selection.SelectKBest`
- :class:`feature_selection.SelectPercentile`
- :class:`feature_selection.SequentialFeatureSelector`
- :class:`feature_selection.VarianceThreshold`
- :class:`kernel_approximation.AdditiveChi2Sampler`
- :class:`impute.IterativeImputer`
- :class:`impute.KNNImputer`
- :class:`impute.SimpleImputer`
- :class:`isotonic.IsotonicRegression`
- :class:`preprocessing.Binarizer`
- :class:`preprocessing.KBinsDiscretizer`
- :class:`preprocessing.MaxAbsScaler`
- :class:`preprocessing.MinMaxScaler`
- :class:`preprocessing.Normalizer`
- :class:`preprocessing.OrdinalEncoder`
- :class:`preprocessing.PowerTransformer`
- :class:`preprocessing.QuantileTransformer`
- :class:`preprocessing.RobustScaler`
- :class:`preprocessing.SplineTransformer`
- :class:`preprocessing.StandardScaler`
- :class:`random_projection.GaussianRandomProjection`
- :class:`random_projection.SparseRandomProjection`
The `NotFittedError` displays an informative message asking to fit the instance
with the appropriate arguments.
:pr:`25294`, :pr:`25308`, :pr:`25291`, :pr:`25367`, :pr:`25402`,
by :user:`John Pangas <jpangas>`, :user:`Rahil Parikh <rprkh>` ,
and :user:`Alex Buzenet <albuzenet>`.
- |Enhancement| Added a multi-threaded Cython routine to the compute squared
Euclidean distances (sometimes followed by a fused reduction operation) for a
pair of datasets consisting of a sparse CSR matrix and a dense NumPy.
This can improve the performance of following functions and estimators:
- :func:`sklearn.metrics.pairwise_distances_argmin`
- :func:`sklearn.metrics.pairwise_distances_argmin_min`
- :class:`sklearn.cluster.AffinityPropagation`
- :class:`sklearn.cluster.Birch`
- :class:`sklearn.cluster.MeanShift`
- :class:`sklearn.cluster.OPTICS`
- :class:`sklearn.cluster.SpectralClustering`
- :func:`sklearn.feature_selection.mutual_info_regression`
- :class:`sklearn.neighbors.KNeighborsClassifier`
- :class:`sklearn.neighbors.KNeighborsRegressor`
- :class:`sklearn.neighbors.RadiusNeighborsClassifier`
- :class:`sklearn.neighbors.RadiusNeighborsRegressor`
- :class:`sklearn.neighbors.LocalOutlierFactor`
- :class:`sklearn.neighbors.NearestNeighbors`
- :class:`sklearn.manifold.Isomap`
- :class:`sklearn.manifold.LocallyLinearEmbedding`
- :class:`sklearn.manifold.TSNE`
- :func:`sklearn.manifold.trustworthiness`
- :class:`sklearn.semi_supervised.LabelPropagation`
- :class:`sklearn.semi_supervised.LabelSpreading`
A typical example of this performance improvement happens when passing a sparse
CSR matrix to the `predict` or `transform` method of estimators that rely on
a dense NumPy representation to store their fitted parameters (or the reverse).
For instance, :meth:`sklearn.neighbors.NearestNeighbors.kneighbors` is now up
to 2 times faster for this case on commonly available laptops.
:pr:`25044` by :user:`Julien Jerphanion <jjerphan>`.
- |Enhancement| All estimators that internally rely on OpenMP multi-threading
(via Cython) now use a number of threads equal to the number of physical
(instead of logical) cores by default. In the past, we observed that using as
many threads as logical cores on SMT hosts could sometimes cause severe
performance problems depending on the algorithms and the shape of the data.
Note that it is still possible to manually adjust the number of threads used
by OpenMP as documented in :ref:`parallelism`.
:pr:`26082` by :user:`Jérémie du Boisberranger <jeremiedbb>` and
:user:`Olivier Grisel <ogrisel>`.
Experimental / Under Development
--------------------------------
- |MajorFeature| :ref:`Metadata routing <metadata_routing>`'s related base
methods are included in this release. This feature is only available via the
`enable_metadata_routing` feature flag which can be enabled using
:func:`sklearn.set_config` and :func:`sklearn.config_context`. For now this
feature is mostly useful for third party developers to prepare their code
base for metadata routing, and we strongly recommend that they also hide it
behind the same feature flag, rather than having it enabled by default.
:pr:`24027` by `Adrin Jalali`_, :user:`Benjamin Bossan <BenjaminBossan>`, and
:user:`Omar Salman <OmarManzoor>`.
Changelog
---------
..
Entries should be grouped by module (in alphabetic order) and prefixed with
one of the labels: |MajorFeature|, |Feature|, |Efficiency|, |Enhancement|,
|Fix| or |API| (see whats_new.rst for descriptions).
Entries should be ordered by those labels (e.g. |Fix| after |Efficiency|).
Changes not specific to a module should be listed under *Multiple Modules*
or *Miscellaneous*.
Entries should end with:
:pr:`123456` by :user:`Joe Bloggs <joeongithub>`.
where 123456 is the *pull request* number, not the issue number.
`sklearn`
.........
- |Feature| Added a new option `skip_parameter_validation`, to the function
:func:`sklearn.set_config` and context manager :func:`sklearn.config_context`, that
allows to skip the validation of the parameters passed to the estimators and public
functions. This can be useful to speed up the code but should be used with care
because it can lead to unexpected behaviors or raise obscure error messages when
setting invalid parameters.
:pr:`25815` by :user:`Jérémie du Boisberranger <jeremiedbb>`.
:mod:`sklearn.base`
...................
- |Feature| A `__sklearn_clone__` protocol is now available to override the
default behavior of :func:`base.clone`. :pr:`24568` by `Thomas Fan`_.
- |Fix| :class:`base.TransformerMixin` now currently keeps a namedtuple's class
if `transform` returns a namedtuple. :pr:`26121` by `Thomas Fan`_.
:mod:`sklearn.calibration`
..........................
- |Fix| :class:`calibration.CalibratedClassifierCV` now does not enforce sample
alignment on `fit_params`. :pr:`25805` by `Adrin Jalali`_.
:mod:`sklearn.cluster`
......................
- |MajorFeature| Added :class:`cluster.HDBSCAN`, a modern hierarchical density-based
clustering algorithm. Similarly to :class:`cluster.OPTICS`, it can be seen as a
generalization of :class:`cluster.DBSCAN` by allowing for hierarchical instead of flat
clustering, however it varies in its approach from :class:`cluster.OPTICS`. This
algorithm is very robust with respect to its hyperparameters' values and can
be used on a wide variety of data without much, if any, tuning.
This implementation is an adaptation from the original implementation of HDBSCAN in
`scikit-learn-contrib/hdbscan <https://github.com/scikit-learn-contrib/hdbscan>`_,
by :user:`Leland McInnes <lmcinnes>` et al.
:pr:`26385` by :user:`Meekail Zain <micky774>`
- |Enhancement| The `sample_weight` parameter now will be used in centroids
initialization for :class:`cluster.KMeans`, :class:`cluster.BisectingKMeans`
and :class:`cluster.MiniBatchKMeans`.
This change will break backward compatibility, since numbers generated
from same random seeds will be different.
:pr:`25752` by :user:`Gleb Levitski <glevv>`,
: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.
:pr:`25251` by :user:`Gleb Levitski <glevv>`.
- |API| The `Xred` argument in :func:`cluster.FeatureAgglomeration.inverse_transform`
is renamed to `Xt` and will be removed in v1.5. :pr:`26503` by `Adrin Jalali`_.
:mod:`sklearn.compose`
......................
- |Fix| :class:`compose.ColumnTransformer` raises an informative error when the individual
transformers of `ColumnTransformer` output pandas dataframes with indexes that are
not consistent with each other and the output is configured to be pandas.
:pr:`26286` by `Thomas Fan`_.
- |Fix| :class:`compose.ColumnTransformer` correctly sets the output of the
remainder when `set_output` is called. :pr:`26323` by `Thomas Fan`_.
:mod:`sklearn.covariance`
.........................
- |Fix| Allows `alpha=0` in :class:`covariance.GraphicalLasso` to be
consistent with :func:`covariance.graphical_lasso`.
:pr:`26033` by :user:`Genesis Valencia <genvalen>`.
- |Fix| :func:`covariance.empirical_covariance` now gives an informative
error message when input is not appropriate.
:pr:`26108` by :user:`Quentin Barthélemy <qbarthelemy>`.
- |API| Deprecates `cov_init` in :func:`covariance.graphical_lasso` in 1.3 since
the parameter has no effect. It will be removed in 1.5.
:pr:`26033` by :user:`Genesis Valencia <genvalen>`.
- |API| Adds `costs_` fitted attribute in :class:`covariance.GraphicalLasso` and
:class:`covariance.GraphicalLassoCV`.
:pr:`26033` by :user:`Genesis Valencia <genvalen>`.
- |API| Adds `covariance` parameter in :class:`covariance.GraphicalLasso`.
:pr:`26033` by :user:`Genesis Valencia <genvalen>`.
- |API| Adds `eps` parameter in :class:`covariance.GraphicalLasso`,
:func:`covariance.graphical_lasso`, and :class:`covariance.GraphicalLassoCV`.
:pr:`26033` by :user:`Genesis Valencia <genvalen>`.
:mod:`sklearn.datasets`
.......................
- |Enhancement| Allows to overwrite the parameters used to open the ARFF file using
the parameter `read_csv_kwargs` in :func:`datasets.fetch_openml` when using the
pandas parser.
:pr:`26433` by :user:`Guillaume Lemaitre <glemaitre>`.
- |Fix| :func:`datasets.fetch_openml` returns improved data types when
`as_frame=True` and `parser="liac-arff"`. :pr:`26386` by `Thomas Fan`_.
- |Fix| Following the ARFF specs, only the marker `"?"` is now considered as a missing
values when opening ARFF files fetched using :func:`datasets.fetch_openml` when using
the pandas parser. The parameter `read_csv_kwargs` allows to overwrite this behaviour.
:pr:`26551` by :user:`Guillaume Lemaitre <glemaitre>`.
- |Fix| :func:`datasets.fetch_openml` will consistently use `np.nan` as missing marker
with both parsers `"pandas"` and `"liac-arff"`.
:pr:`26579` by :user:`Guillaume Lemaitre <glemaitre>`.
- |API| The `data_transposed` argument of :func:`datasets.make_sparse_coded_signal`
is deprecated and will be removed in v1.5.
:pr:`25784` by :user:`Jérémie du Boisberranger`.
:mod:`sklearn.decomposition`
............................
- |Efficiency| :class:`decomposition.MiniBatchDictionaryLearning` and
:class:`decomposition.MiniBatchSparsePCA` are now faster for small batch sizes by
avoiding duplicate validations.
:pr:`25490` by :user:`Jérémie du Boisberranger <jeremiedbb>`.
- |Enhancement| :class:`decomposition.DictionaryLearning` now accepts the parameter
`callback` for consistency with the function :func:`decomposition.dict_learning`.
:pr:`24871` by :user:`Omar Salman <OmarManzoor>`.
- |Fix| Treat more consistently small values in the `W` and `H` matrices during the
`fit` and `transform` steps of :class:`decomposition.NMF` and
:class:`decomposition.MiniBatchNMF` which can produce different results than previous
versions. :pr:`25438` by :user:`Yotam Avidar-Constantini <yotamcons>`.
- |API| The `W` argument in :func:`decomposition.NMF.inverse_transform` and
:class:`decomposition.MiniBatchNMF.inverse_transform` is renamed to `Xt` and
will be removed in v1.5. :pr:`26503` by `Adrin Jalali`_.
:mod:`sklearn.discriminant_analysis`
....................................
- |Enhancement| :class:`discriminant_analysis.LinearDiscriminantAnalysis` now
supports the `PyTorch <https://pytorch.org/>`__. See
:ref:`array_api` for more details. :pr:`25956` by `Thomas Fan`_.
:mod:`sklearn.ensemble`
.......................
- |Feature| :class:`ensemble.HistGradientBoostingRegressor` now supports
the Gamma deviance loss via `loss="gamma"`.
Using the Gamma deviance as loss function comes in handy for modelling skewed
distributed, strictly positive valued targets.
:pr:`22409` by :user:`Christian Lorentzen <lorentzenchr>`.
- |Feature| Compute a custom out-of-bag score by passing a callable to
:class:`ensemble.RandomForestClassifier`, :class:`ensemble.RandomForestRegressor`,
:class:`ensemble.ExtraTreesClassifier` and :class:`ensemble.ExtraTreesRegressor`.
:pr:`25177` by `Tim Head`_.
- |Feature| :class:`ensemble.GradientBoostingClassifier` now exposes
out-of-bag scores via the `oob_scores_` or `oob_score_` attributes.
:pr:`24882` by :user:`Ashwin Mathur <awinml>`.
- |Efficiency| :class:`ensemble.IsolationForest` predict time is now faster
(typically by a factor of 8 or more). Internally, the estimator now precomputes
decision path lengths per tree at `fit` time. It is therefore not possible
to load an estimator trained with scikit-learn 1.2 to make it predict with
scikit-learn 1.3: retraining with scikit-learn 1.3 is required.
:pr:`25186` by :user:`Felipe Breve Siola <fsiola>`.
- |Efficiency| :class:`ensemble.RandomForestClassifier` and
:class:`ensemble.RandomForestRegressor` with `warm_start=True` now only
recomputes out-of-bag scores when there are actually more `n_estimators`
in subsequent `fit` calls.
:pr:`26318` by :user:`Joshua Choo Yun Keat <choo8>`.
- |Enhancement| :class:`ensemble.BaggingClassifier` and
:class:`ensemble.BaggingRegressor` expose the `allow_nan` tag from the
underlying estimator. :pr:`25506` by `Thomas Fan`_.
- |Fix| :meth:`ensemble.RandomForestClassifier.fit` sets `max_samples = 1`
when `max_samples` is a float and `round(n_samples * max_samples) < 1`.
:pr:`25601` by :user:`Jan Fidor <JanFidor>`.
- |Fix| :meth:`ensemble.IsolationForest.fit` no longer warns about missing
feature names when called with `contamination` not `"auto"` on a pandas
dataframe.
:pr:`25931` by :user:`Yao Xiao <Charlie-XIAO>`.
- |Fix| :class:`ensemble.HistGradientBoostingRegressor` and
:class:`ensemble.HistGradientBoostingClassifier` treats negative values for
categorical features consistently as missing values, following LightGBM's and
pandas' conventions.
:pr:`25629` by `Thomas Fan`_.
- |Fix| Fix deprecation of `base_estimator` in :class:`ensemble.AdaBoostClassifier`
and :class:`ensemble.AdaBoostRegressor` that was introduced in :pr:`23819`.
:pr:`26242` by :user:`Marko Toplak <markotoplak>`.
:mod:`sklearn.exceptions`
.........................
- |Feature| Added :class:`exceptions.InconsistentVersionWarning` which is raised
when a scikit-learn estimator is unpickled with a scikit-learn version that is
inconsistent with the sckit-learn version the estimator was pickled with.
:pr:`25297` by `Thomas Fan`_.
:mod:`sklearn.feature_extraction`
.................................
- |API| :class:`feature_extraction.image.PatchExtractor` now follows the
transformer API of scikit-learn. This class is defined as a stateless transformer
meaning that it is note required to call `fit` before calling `transform`.
Parameter validation only happens at `fit` time.
:pr:`24230` by :user:`Guillaume Lemaitre <glemaitre>`.
:mod:`sklearn.feature_selection`
................................
- |Enhancement| All selectors in :mod:`sklearn.feature_selection` will preserve
a DataFrame's dtype when transformed. :pr:`25102` by `Thomas Fan`_.
- |Fix| :class:`feature_selection.SequentialFeatureSelector`'s `cv` parameter
now supports generators. :pr:`25973` by `Yao Xiao <Charlie-XIAO>`.
:mod:`sklearn.impute`
.....................
- |Enhancement| Added the parameter `fill_value` to :class:`impute.IterativeImputer`.
:pr:`25232` by :user:`Thijs van Weezel <ValueInvestorThijs>`.
- |Fix| :class:`impute.IterativeImputer` now correctly preserves the Pandas
Index when the `set_config(transform_output="pandas")`. :pr:`26454` by `Thomas Fan`_.
:mod:`sklearn.inspection`
.........................
- |Enhancement| Added support for `sample_weight` in
:func:`inspection.partial_dependence` and
:meth:`inspection.PartialDependenceDisplay.from_estimator`. This allows for
weighted averaging when aggregating for each value of the grid we are making the
inspection on. The option is only available when `method` is set to `brute`.
:pr:`25209` and :pr:`26644` by :user:`Carlo Lemos <vitaliset>`.
- |API| :func:`inspection.partial_dependence` returns a :class:`utils.Bunch` with
new key: `grid_values`. The `values` key is deprecated in favor of `grid_values`
and the `values` key will be removed in 1.5.
:pr:`21809` and :pr:`25732` by `Thomas Fan`_.
:mod:`sklearn.kernel_approximation`
...................................
- |Fix| :class:`kernel_approximation.AdditiveChi2Sampler` is now stateless.
The `sample_interval_` attribute is deprecated and will be removed in 1.5.
:pr:`25190` by :user:`Vincent Maladière <Vincent-Maladiere>`.
:mod:`sklearn.linear_model`
...........................
- |Efficiency| Avoid data scaling when `sample_weight=None` and other
unnecessary data copies and unexpected dense to sparse data conversion in
:class:`linear_model.LinearRegression`.
:pr:`26207` by :user:`Olivier Grisel <ogrisel>`.
- |Enhancement| :class:`linear_model.SGDClassifier`,
:class:`linear_model.SGDRegressor` and :class:`linear_model.SGDOneClassSVM`
now preserve dtype for `numpy.float32`.
:pr:`25587` by :user:`Omar Salman <OmarManzoor>`.
- |Enhancement| The `n_iter_` attribute has been included in
:class:`linear_model.ARDRegression` to expose the actual number of iterations
required to reach the stopping criterion.
:pr:`25697` by :user:`John Pangas <jpangas>`.
- |Fix| Use a more robust criterion to detect convergence of
:class:`linear_model.LogisticRegression` with `penalty="l1"` and `solver="liblinear"`
on linearly separable problems.
:pr:`25214` by `Tom Dupre la Tour`_.
- |Fix| Fix a crash when calling `fit` on
:class:`linear_model.LogisticRegression` with `solver="newton-cholesky"` and
`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
estimators consistent with the rest of estimators.
:pr:`25697` by :user:`John Pangas <jpangas>`.
:mod:`sklearn.manifold`
.......................
- |Fix| :class:`manifold.Isomap` now correctly preserves the Pandas
Index when the `set_config(transform_output="pandas")`. :pr:`26454` by `Thomas Fan`_.
:mod:`sklearn.metrics`
......................
- |Feature| Adds `zero_division=np.nan` to multiple classification metrics:
:func:`metrics.precision_score`, :func:`metrics.recall_score`,
:func:`metrics.f1_score`, :func:`metrics.fbeta_score`,
:func:`metrics.precision_recall_fscore_support`,
:func:`metrics.classification_report`. When `zero_division=np.nan` and there is a
zero division, the metric is undefined and is excluded from averaging. When not used
for averages, the value returned is `np.nan`.
:pr:`25531` by :user:`Marc Torrellas Socastro <marctorsoc>`.
- |Feature| :func:`metrics.average_precision_score` now supports the
multiclass case.
:pr:`17388` by :user:`Geoffrey Bolmier <gbolmier>` and
:pr:`24769` by :user:`Ashwin Mathur <awinml>`.
- |Efficiency| The computation of the expected mutual information in
:func:`metrics.adjusted_mutual_info_score` is now faster when the number of
unique labels is large and its memory usage is reduced in general.
:pr:`25713` by :user:`Kshitij Mathur <Kshitij68>`,
:user:`Guillaume Lemaitre <glemaitre>`, :user:`Omar Salman <OmarManzoor>` and
:user:`Jérémie du Boisberranger <jeremiedbb>`.
- |Enhancement| :class:`metrics.silhouette_samples` nows accepts a sparse
matrix of pairwise distances between samples, or a feature array.
:pr:`18723` by :user:`Sahil Gupta <sahilgupta2105>` and
:pr:`24677` by :user:`Ashwin Mathur <awinml>`.
- |Enhancement| A new parameter `drop_intermediate` was added to
:func:`metrics.precision_recall_curve`,
:func:`metrics.PrecisionRecallDisplay.from_estimator`,
:func:`metrics.PrecisionRecallDisplay.from_predictions`,
which drops some suboptimal thresholds to create lighter precision-recall
curves.
:pr:`24668` by :user:`dberenbaum`.
- |Enhancement| :meth:`metrics.RocCurveDisplay.from_estimator` and
:meth:`metrics.RocCurveDisplay.from_predictions` now accept two new keywords,
`plot_chance_level` and `chance_level_kw` to plot the baseline chance
level. This line is exposed in the `chance_level_` attribute.
:pr:`25987` by :user:`Yao Xiao <Charlie-XIAO>`.
- |Enhancement| :meth:`metrics.PrecisionRecallDisplay.from_estimator` and
:meth:`metrics.PrecisionRecallDisplay.from_predictions` now accept two new
keywords, `plot_chance_level` and `chance_level_kw` to plot the baseline
chance level. This line is exposed in the `chance_level_` attribute.
:pr:`26019` by :user:`Yao Xiao <Charlie-XIAO>`.
- |Fix| :func:`metrics.pairwise.manhattan_distances` now supports readonly sparse datasets.
:pr:`25432` by :user:`Julien Jerphanion <jjerphan>`.
- |Fix| Fixed :func:`metrics.classification_report` so that empty input will return
`np.nan`. Previously, "macro avg" and `weighted avg` would return
e.g. `f1-score=np.nan` and `f1-score=0.0`, being inconsistent. Now, they
both return `np.nan`.
:pr:`25531` by :user:`Marc Torrellas Socastro <marctorsoc>`.
- |Fix| :func:`metrics.ndcg_score` now gives a meaningful error message for input of
length 1.
:pr:`25672` by :user:`Lene Preuss <lene>` and :user:`Wei-Chun Chu <wcchu>`.
- |Fix| :func:`metrics.log_loss` raises a warning if the values of the parameter
`y_pred` are not normalized, instead of actually normalizing them in the metric.
Starting from 1.5 this will raise an error.
:pr:`25299` by :user:`Omar Salman <OmarManzoor`.
- |Fix| In :func:`metrics.roc_curve`, use the threshold value `np.inf` instead of
arbitrary `max(y_score) + 1`. This threshold is associated with the ROC curve point
`tpr=0` and `fpr=0`.
:pr:`26194` by :user:`Guillaume Lemaitre <glemaitre>`.
- |Fix| The `'matching'` metric has been removed when using SciPy>=1.9
to be consistent with `scipy.spatial.distance` which does not support
`'matching'` anymore.
:pr:`26264` by :user:`Barata T. Onggo <magnusbarata>`
- |API| The `eps` parameter of the :func:`metrics.log_loss` has been deprecated and
will be removed in 1.5. :pr:`25299` by :user:`Omar Salman <OmarManzoor>`.
:mod:`sklearn.gaussian_process`
...............................
- |Fix| :class:`gaussian_process.GaussianProcessRegressor` has a new argument
`n_targets`, which is used to decide the number of outputs when sampling
from the prior distributions. :pr:`23099` by :user:`Zhehao Liu <MaxwellLZH>`.
:mod:`sklearn.mixture`
......................
- |Efficiency| :class:`mixture.GaussianMixture` is more efficient now and will bypass
unnecessary initialization if the weights, means, and precisions are
given by users.
:pr:`26021` by :user:`Jiawei Zhang <jiawei-zhang-a>`.
:mod:`sklearn.model_selection`
..............................
- |MajorFeature| Added the class :class:`model_selection.ValidationCurveDisplay`
that allows easy plotting of validation curves obtained by the function
:func:`model_selection.validation_curve`.
:pr:`25120` by :user:`Guillaume Lemaitre <glemaitre>`.
- |API| The parameter `log_scale` in the class
:class:`model_selection.LearningCurveDisplay` has been deprecated in 1.3 and
will be removed in 1.5. The default scale can be overridden by setting it
directly on the `ax` object and will be set automatically from the spacing
of the data points otherwise.
:pr:`25120` by :user:`Guillaume Lemaitre <glemaitre>`.
- |Enhancement| :func:`model_selection.cross_validate` accepts a new parameter
`return_indices` to return the train-test indices of each cv split.
:pr:`25659` by :user:`Guillaume Lemaitre <glemaitre>`.
:mod:`sklearn.multioutput`
..........................
- |Fix| :func:`getattr` on :meth:`multioutput.MultiOutputRegressor.partial_fit`
and :meth:`multioutput.MultiOutputClassifier.partial_fit` now correctly raise
an `AttributeError` if done before calling `fit`. :pr:`26333` by `Adrin
Jalali`_.
:mod:`sklearn.naive_bayes`
..........................
- |Fix| :class:`naive_bayes.GaussianNB` does not raise anymore a `ZeroDivisionError`
when the provided `sample_weight` reduces the problem to a single class in `fit`.
:pr:`24140` by :user:`Jonathan Ohayon <Johayon>` and :user:`Chiara Marmo <cmarmo>`.
:mod:`sklearn.neighbors`
........................
- |Enhancement| The performance of :meth:`neighbors.KNeighborsClassifier.predict`
and of :meth:`neighbors.KNeighborsClassifier.predict_proba` has been improved
when `n_neighbors` is large and `algorithm="brute"` with non Euclidean metrics.
:pr:`24076` by :user:`Meekail Zain <micky774>`, :user:`Julien Jerphanion <jjerphan>`.
- |Fix| Remove support for `KulsinskiDistance` in :class:`neighbors.BallTree`. This
dissimilarity is not a metric and cannot be supported by the BallTree.
:pr:`25417` by :user:`Guillaume Lemaitre <glemaitre>`.
- |API| The support for metrics other than `euclidean` and `manhattan` and for
callables in :class:`neighbors.NearestNeighbors` is deprecated and will be removed in
version 1.5. :pr:`24083` by :user:`Valentin Laurent <Valentin-Laurent>`.
:mod:`sklearn.neural_network`
.............................
- |Fix| :class:`neural_network.MLPRegressor` and :class:`neural_network.MLPClassifier`
reports the right `n_iter_` when `warm_start=True`. It corresponds to the number
of iterations performed on the current call to `fit` instead of the total number
of iterations performed since the initialization of the estimator.
:pr:`25443` by :user:`Marvin Krawutschke <Marvvxi>`.
:mod:`sklearn.pipeline`
.......................
- |Feature| :class:`pipeline.FeatureUnion` can now use indexing notation (e.g.
`feature_union["scalar"]`) to access transformers by name. :pr:`25093` by
`Thomas Fan`_.
- |Feature| :class:`pipeline.FeatureUnion` can now access the
`feature_names_in_` attribute if the `X` value seen during `.fit` has a
`columns` attribute and all columns are strings. e.g. when `X` is a
`pandas.DataFrame`
:pr:`25220` by :user:`Ian Thompson <it176131>`.
- |Fix| :meth:`pipeline.Pipeline.fit_transform` now raises an `AttributeError`
if the last step of the pipeline does not support `fit_transform`.
:pr:`26325` by `Adrin Jalali`_.
:mod:`sklearn.preprocessing`
............................
- |MajorFeature| Introduces :class:`preprocessing.TargetEncoder` which is a
categorical encoding based on target mean conditioned on the value of the
category. :pr:`25334` by `Thomas Fan`_.
- |Feature| :class:`preprocessing.OrdinalEncoder` now supports grouping
infrequent categories into a single feature. Grouping infrequent categories
is enabled by specifying how to select infrequent categories with
`min_frequency` or `max_categories`. :pr:`25677` by `Thomas Fan`_.
- |Enhancement| :class:`preprocessing.PolynomialFeatures` now calculates the
number of expanded terms a-priori when dealing with sparse `csr` matrices
in order to optimize the choice of `dtype` for `indices` and `indptr`. It
can now output `csr` matrices with `np.int32` `indices/indptr` components
when there are few enough elements, and will automatically use `np.int64`
for sufficiently large matrices.
:pr:`20524` by :user:`niuk-a <niuk-a>` and
:pr:`23731` by :user:`Meekail Zain <micky774>`
- |Enhancement| A new parameter `sparse_output` was added to
:class:`preprocessing.SplineTransformer`, available as of SciPy 1.8. If
`sparse_output=True`, :class:`preprocessing.SplineTransformer` returns a sparse
CSR matrix. :pr:`24145` by :user:`Christian Lorentzen <lorentzenchr>`.
- |Enhancement| Adds a `feature_name_combiner` parameter to
:class:`preprocessing.OneHotEncoder`. This specifies a custom callable to
create feature names to be returned by
:meth:`preprocessing.OneHotEncoder.get_feature_names_out`. The callable
combines input arguments `(input_feature, category)` to a string.
:pr:`22506` by :user:`Mario Kostelac <mariokostelac>`.
- |Enhancement| Added support for `sample_weight` in
:class:`preprocessing.KBinsDiscretizer`. This allows specifying the parameter
`sample_weight` for each sample to be used while fitting. The option is only
available when `strategy` is set to `quantile` and `kmeans`.
:pr:`24935` by :user:`Seladus <seladus>`, :user:`Guillaume Lemaitre <glemaitre>`, and
:user:`Dea María Léon <deamarialeon>`, :pr:`25257` by :user:`Gleb Levitski <glevv>`.
- |Enhancement| Subsampling through the `subsample` parameter can now be used in
:class:`preprocessing.KBinsDiscretizer` regardless of the strategy used.
:pr:`26424` by :user:`Jérémie du Boisberranger <jeremiedbb>`.
- |Fix| :class:`preprocessing.PowerTransformer` now correctly preserves the Pandas
Index when the `set_config(transform_output="pandas")`. :pr:`26454` by `Thomas Fan`_.
- |Fix| :class:`preprocessing.PowerTransformer` now correctly raises error when
using `method="box-cox"` on data with a constant `np.nan` column.
:pr:`26400` by :user:`Yao Xiao <Charlie-XIAO>`.
- |Fix| :class:`preprocessing.PowerTransformer` with `method="yeo-johnson"` now leaves
constant features unchanged instead of transforming with an arbitrary value for
the `lambdas_` fitted parameter.
:pr:`26566` by :user:`Jérémie du Boisberranger <jeremiedbb>`.
- |API| The default value of the `subsample` parameter of
:class:`preprocessing.KBinsDiscretizer` will change from `None` to `200_000` in
version 1.5 when `strategy="kmeans"` or `strategy="uniform"`.
:pr:`26424` by :user:`Jérémie du Boisberranger <jeremiedbb>`.
:mod:`sklearn.svm`
..................
- |API| `dual` parameter now accepts `auto` option for
:class:`svm.LinearSVC` and :class:`svm.LinearSVR`.
:pr:`26093` by :user:`Gleb Levitski <glevv>`.
:mod:`sklearn.tree`
...................
- |MajorFeature| :class:`tree.DecisionTreeRegressor` and
:class:`tree.DecisionTreeClassifier` support missing values when
`splitter='best'` and criterion is `gini`, `entropy`, or `log_loss`,
for classification or `squared_error`, `friedman_mse`, or `poisson`
for regression. :pr:`23595`, :pr:`26376` by `Thomas Fan`_.
- |Enhancement| Adds a `class_names` parameter to
:func:`tree.export_text`. This allows specifying the parameter `class_names`
for each target class in ascending numerical order.
:pr:`25387` by :user:`William M <Akbeeh>` and :user:`crispinlogan <crispinlogan>`.
- |Fix| :func:`tree.export_graphviz` and :func:`tree.export_text` now accepts
`feature_names` and `class_names` as array-like rather than lists.
:pr:`26289` by :user:`Yao Xiao <Charlie-XIAO>`
:mod:`sklearn.utils`
....................
- |FIX| Fixes :func:`utils.check_array` to properly convert pandas
extension arrays. :pr:`25813` and :pr:`26106` by `Thomas Fan`_.
- |Fix| :func:`utils.check_array` now supports pandas DataFrames with
extension arrays and object dtypes by return an ndarray with object dtype.
:pr:`25814` by `Thomas Fan`_.
- |API| `utils.estimator_checks.check_transformers_unfitted_stateless` has been
introduced to ensure stateless transformers don't raise `NotFittedError`
during `transform` with no prior call to `fit` or `fit_transform`.
:pr:`25190` by :user:`Vincent Maladière <Vincent-Maladiere>`.
- |API| A `FutureWarning` is now raised when instantiating a class which inherits from
a deprecated base class (i.e. decorated by :class:`utils.deprecated`) and which
overrides the `__init__` method.
:pr:`25733` by :user:`Brigitta Sipőcz <bsipocz>` and
:user:`Jérémie du Boisberranger <jeremiedbb>`.
:mod:`sklearn.semi_supervised`
..............................
- |Enhancement| :meth:`semi_supervised.LabelSpreading.fit` and
:meth:`semi_supervised.LabelPropagation.fit` now accepts sparse metrics.
:pr:`19664` by :user:`Kaushik Amar Das <cozek>`.
Miscellaneous
.............
- |Enhancement| Replace obsolete exceptions `EnvironmentError`, `IOError` and
`WindowsError`.
:pr:`26466` by :user:`Dimitri Papadopoulos ORfanos <DimitriPapadopoulos>`.
.. rubric:: Code and documentation contributors
Thanks to everyone who has contributed to the maintenance and improvement of
the project since version 1.2, including:
2357juan, Abhishek Singh Kushwah, Adam Handke, Adam Kania, Adam Li, adienes,
Admir Demiraj, adoublet, Adrin Jalali, A.H.Mansouri, Ahmedbgh, Ala-Na, Alex
Buzenet, AlexL, Ali H. El-Kassas, amay, András Simon, André Pedersen, Andrew
Wang, Ankur Singh, annegnx, Ansam Zedan, Anthony22-dev, Artur Hermano, Arturo
Amor, as-90, ashah002, Ashish Dutt, Ashwin Mathur, AymericBasset, Azaria
Gebremichael, Barata Tripramudya Onggo, Benedek Harsanyi, Benjamin Bossan,
Bharat Raghunathan, Binesh Bannerjee, Boris Feld, Brendan Lu, Brevin Kunde,
cache-missing, Camille Troillard, Carla J, carlo, Carlo Lemos, c-git, Changyao
Chen, Chiara Marmo, Christian Lorentzen, Christian Veenhuis, Christine P. Chai,
crispinlogan, Da-Lan, DanGonite57, Dave Berenbaum, davidblnc, david-cortes,
Dayne, Dea María Léon, Denis, Dimitri Papadopoulos Orfanos, Dimitris
Litsidis, Dmitry Nesterov, Dominic Fox, Dominik Prodinger, Edern, Ekaterina
Butyugina, Elabonga Atuo, Emir, farhan khan, Felipe Siola, futurewarning, Gael
Varoquaux, genvalen, Gleb Levitski, Guillaume Lemaitre, gunesbayir, Haesun
Park, hujiahong726, i-aki-y, Ian Thompson, Ido M, Ily, Irene, Jack McIvor,
jakirkham, James Dean, JanFidor, Jarrod Millman, JB Mountford, Jérémie du
Boisberranger, Jessicakk0711, Jiawei Zhang, Joey Ortiz, JohnathanPi, John
Pangas, Joshua Choo Yun Keat, Joshua Hedlund, JuliaSchoepp, Julien Jerphanion,
jygerardy, ka00ri, Kaushik Amar Das, Kento Nozawa, Kian Eliasi, Kilian Kluge,
Lene Preuss, Linus, Logan Thomas, Loic Esteve, Louis Fouquet, Lucy Liu, Madhura
Jayaratne, Marc Torrellas Socastro, Maren Westermann, Mario Kostelac, Mark
Harfouche, Marko Toplak, Marvin Krawutschke, Masanori Kanazu, mathurinm, Matt
Haberland, Max Halford, maximeSaur, Maxwell Liu, m. bou, mdarii, Meekail Zain,
Mikhail Iljin, murezzda, Nawazish Alam, Nicola Fanelli, Nightwalkx, Nikolay
Petrov, Nishu Choudhary, NNLNR, npache, Olivier Grisel, Omar Salman, ouss1508,
PAB, Pandata, partev, Peter Piontek, Phil, pnucci, Pooja M, Pooja Subramaniam,
precondition, Quentin Barthélemy, Rafal Wojdyla, Raghuveer Bhat, Rahil Parikh,
Ralf Gommers, ram vikram singh, Rushil Desai, Sadra Barikbin, SANJAI_3, Sashka
Warner, Scott Gigante, Scott Gustafson, searchforpassion, Seoeun
Hong, Shady el Gewily, Shiva chauhan, Shogo Hida, Shreesha Kumar Bhat, sonnivs,
Sortofamudkip, Stanislav (Stanley) Modrak, Stefanie Senger, Steven Van
Vaerenbergh, Tabea Kossen, Théophile Baranger, Thijs van Weezel, Thomas A
Caswell, Thomas Germer, Thomas J. Fan, Tim Head, Tim P, Tom Dupré la Tour,
tomiock, tspeng, Valentin Laurent, Veghit, VIGNESH D, Vijeth Moudgalya, Vinayak
Mehta, Vincent M, Vincent-violet, Vyom Pathak, William M, windiana42, Xiao
Yuan, Yao Xiao, Yaroslav Halchenko, Yotam Avidar-Constantini, Yuchen Zhou,
Yusuf Raji, zeeshan lone