88Version 1.5
99===========
1010
11- ..
12- -- UNCOMMENT WHEN 1.5.0 IS RELEASED --
13- For a short description of the main highlights of the release, please refer to
14- :ref:`sphx_glr_auto_examples_release_highlights_plot_release_highlights_1_5_0.py`.
11+ For a short description of the main highlights of the release, please refer to
12+ :ref: `sphx_glr_auto_examples_release_highlights_plot_release_highlights_1_5_0.py `.
1513
1614.. include :: changelog_legend.inc
1715
@@ -20,7 +18,7 @@ Version 1.5
2018Version 1.5.0
2119=============
2220
23- **In Development **
21+ **May 2024 **
2422
2523Security
2624--------
@@ -59,6 +57,10 @@ Changed models
5957Changes impacting many modules
6058------------------------------
6159
60+ - |Fix | Raise `ValueError ` with an informative error message when passing 1D
61+ sparse arrays to methods that expect 2D sparse inputs.
62+ :pr: `28988 ` by :user: `Olivier Grisel <ogrisel> `.
63+
6264- |API | The name of the input of the `inverse_transform ` method of estimators has been
6365 standardized to `X `. As a consequence, `Xt ` is deprecated and will be removed in
6466 version 1.7 in the following estimators: :class: `cluster.FeatureAgglomeration `,
@@ -67,10 +69,6 @@ Changes impacting many modules
6769 :class: `pipeline.Pipeline ` and :class: `preprocessing.KBinsDiscretizer `.
6870 :pr: `28756 ` by :user: `Will Dean <wd60622> `.
6971
70- - |Fix | Raise `ValueError ` with an informative error message when passing 1D
71- sparse arrays to methods that expect 2D sparse inputs.
72- :pr: `28988 ` by :user: `Olivier Grisel <ogrisel> `.
73-
7472Support for Array API
7573---------------------
7674
@@ -82,8 +80,8 @@ See :ref:`array_api` for more details.
8280**Functions: **
8381
8482- :func: `sklearn.metrics.r2_score ` now supports Array API compliant inputs.
85- :pr: `27904 ` by :user: `Eric Lindgren <elindgren> `, `Franck Charras <fcharras> `,
86- `Olivier Grisel <ogrisel> ` and `Tim Head <betatim> `.
83+ :pr: `27904 ` by :user: `Eric Lindgren <elindgren> `, :user: `Franck Charras <fcharras> `,
84+ :user: `Olivier Grisel <ogrisel> ` and :user: `Tim Head <betatim> `.
8785
8886**Classes: **
8987
@@ -103,8 +101,8 @@ Unless we discover a major blocker, setuptools support will be dropped in
103101scikit-learn 1.6. The 1.5.x releases will support building scikit-learn with
104102setuptools.
105103
106- Meson support for building scikit-learn was added in :pr: `28040 ` by :user: ` Loïc
107- Estève <lesteve> `
104+ Meson support for building scikit-learn was added in :pr: `28040 ` by
105+ :user: ` Loïc Estève <lesteve>
108106
109107Metadata Routing
110108----------------
@@ -120,7 +118,8 @@ more details.
120118 now support metadata routing. The fit methods now
121119 accept ``**fit_params `` which are passed to the underlying estimators
122120 via their `fit ` methods.
123- :pr: `28432 ` by :user: `Adam Li <adam2392> ` and :user: `Benjamin Bossan <BenjaminBossan> `.
121+ :pr: `28432 ` by :user: `Adam Li <adam2392> ` and
122+ :user: `Benjamin Bossan <BenjaminBossan> `.
124123
125124- |Feature | :class: `linear_model.RidgeCV ` and
126125 :class: `linear_model.RidgeClassifierCV ` now support metadata routing in
@@ -144,8 +143,8 @@ more details.
144143
145144- |Feature | :class: `pipeline.FeatureUnion ` now supports metadata routing in its
146145 ``fit `` and ``fit_transform `` methods and route metadata to the underlying
147- transformers' ``fit `` and ``fit_transform ``. :pr: ` 28205 ` by :user: ` Stefanie
148- Senger <StefanieSenger> `.
146+ transformers' ``fit `` and ``fit_transform ``.
147+ :pr: ` 28205 ` by :user: ` Stefanie Senger <StefanieSenger>
149148
150149- |Fix | Fix an issue when resolving default routing requests set via class
151150 attributes.
@@ -156,8 +155,8 @@ more details.
156155 :pr: `28651 ` by `Adrin Jalali `_.
157156
158157- |FIX | Prevent a `RecursionError ` when estimators with the default `scoring `
159- param (`None `) route metadata. :pr: ` 28712 ` by :user: ` Stefanie Senger
160- <StefanieSenger> `.
158+ param (`None `) route metadata.
159+ :pr: ` 28712 ` by :user: ` Stefanie Senger <StefanieSenger>
161160
162161Changelog
163162---------
@@ -217,19 +216,20 @@ Changelog
217216:mod: `sklearn.cross_decomposition `
218217..................................
219218
220- - |API | Deprecates `Y ` in favor of `y ` in the methods fit, transform and inverse_transform of:
219+ - |Fix | The `coef_ ` fitted attribute of :class: `cross_decomposition.PLSRegression `
220+ now takes into account both the scale of `X ` and `Y ` when `scale=True `. Note that
221+ the previous predicted values were not affected by this bug.
222+ :pr: `28612 ` by :user: `Guillaume Lemaitre <glemaitre> `.
223+
224+ - |API | Deprecates `Y ` in favor of `y ` in the methods fit, transform and
225+ inverse_transform of:
221226 :class: `cross_decomposition.PLSRegression `.
222227 :class: `cross_decomposition.PLSCanonical `,
223228 :class: `cross_decomposition.CCA `,
224229 and :class: `cross_decomposition.PLSSVD `.
225230 `Y ` will be removed in version 1.7.
226231 :pr: `28604 ` by :user: `David Leon <davidleon123> `.
227232
228- - |Fix | The `coef_ ` fitted attribute of :class: `cross_decomposition.PLSRegression `
229- now takes into account both the scale of `X ` and `Y ` when `scale=True `. Note that
230- the previous predicted values were not affected by this bug.
231- :pr: `28612 ` by :user: `Guillaume Lemaitre <glemaitre> `.
232-
233233:mod: `sklearn.datasets `
234234.......................
235235
@@ -245,7 +245,8 @@ Changelog
245245 :func: `datasets.fetch_rcv1 `,
246246 and :func: `datasets.fetch_species_distributions `.
247247 By default, the functions will retry up to 3 times in case of network failures.
248- :pr: `28160 ` by :user: `Zhehao Liu <MaxwellLZH> ` and :user: `Filip Karlo Došilović <fkdosilovic> `.
248+ :pr: `28160 ` by :user: `Zhehao Liu <MaxwellLZH> ` and
249+ :user: `Filip Karlo Došilović <fkdosilovic> `.
249250
250251:mod: `sklearn.decomposition `
251252............................
@@ -350,13 +351,8 @@ Changelog
350351
351352- |Fix | :class: `linear_model.ElasticNet `, :class: `linear_model.ElasticNetCV `,
352353 :class: `linear_model.Lasso ` and :class: `linear_model.LassoCV ` now explicitly don't
353- accept large sparse data formats. :pr: `27576 ` by :user: `Stefanie Senger
354- <StefanieSenger> `.
355-
356- - |API | :class: `linear_model.RidgeCV ` and :class: `linear_model.RidgeClassifierCV `
357- will now allow `alpha=0 ` when `cv != None `, which is consistent with
358- :class: `linear_model.Ridge ` and :class: `linear_model.RidgeClassifier `.
359- :pr: `28425 ` by :user: `Lucy Liu <lucyleeow> `.
354+ accept large sparse data formats.
355+ :pr: `27576 ` by :user: `Stefanie Senger <StefanieSenger> `.
360356
361357- |Fix | :class: `linear_model.RidgeCV ` and :class: `RidgeClassifierCV ` correctly pass
362358 `sample_weight ` to the underlying scorer when `cv ` is None.
@@ -366,6 +362,11 @@ Changelog
366362 will now always be `None ` when `tol ` is set, as `n_nonzero_coefs ` is ignored in
367363 this case. :pr: `28557 ` by :user: `Lucy Liu <lucyleeow> `.
368364
365+ - |API | :class: `linear_model.RidgeCV ` and :class: `linear_model.RidgeClassifierCV `
366+ will now allow `alpha=0 ` when `cv != None `, which is consistent with
367+ :class: `linear_model.Ridge ` and :class: `linear_model.RidgeClassifier `.
368+ :pr: `28425 ` by :user: `Lucy Liu <lucyleeow> `.
369+
369370- |API | Passing `average=0 ` to disable averaging is deprecated in
370371 :class: `linear_model.PassiveAggressiveClassifier `,
371372 :class: `linear_model.PassiveAggressiveRegressor `,
@@ -382,7 +383,8 @@ Changelog
382383 :pr: `28703 ` by :user: `Christian Lorentzen <lorentzenchr> `.
383384
384385- |API | `store_cv_values ` and `cv_values_ ` are deprecated in favor of
385- `store_cv_results ` and `cv_results_ ` in `RidgeCV ` and `RidgeClassifierCV `.
386+ `store_cv_results ` and `cv_results_ ` in `~linear_model.RidgeCV ` and
387+ `~linear_model.RidgeClassifierCV `.
386388 :pr: `28915 ` by :user: `Lucy Liu <lucyleeow> `.
387389
388390:mod: `sklearn.manifold `
@@ -401,8 +403,15 @@ Changelog
401403 :pr: `27456 ` by :user: `Venkatachalam N <venkyyuvy> `, :user: `Kshitij Mathur <Kshitij68> `
402404 and :user: `Julian Libiseller-Egger <julibeg> `.
403405
406+ - |Feature | :func: `sklearn.metrics.check_scoring ` now returns a multi-metric scorer
407+ when `scoring ` as a `dict `, `set `, `tuple `, or `list `. :pr: `28360 ` by `Thomas Fan `_.
408+
409+ - |Feature | :func: `metrics.d2_log_loss_score ` has been added which
410+ calculates the D^2 score for the log loss.
411+ :pr: `28351 ` by :user: `Omar Salman <OmarManzoor> `.
412+
404413- |Efficiency | Improve efficiency of functions :func: `~metrics.brier_score_loss `,
405- :func: `~metrics .calibration_curve `, :func: `~metrics.det_curve `,
414+ :func: `~calibration .calibration_curve `, :func: `~metrics.det_curve `,
406415 :func: `~metrics.precision_recall_curve `,
407416 :func: `~metrics.roc_curve ` when `pos_label ` argument is specified.
408417 Also improve efficiency of methods `from_estimator `
@@ -411,9 +420,6 @@ Changelog
411420 :class: `~calibration.CalibrationDisplay `.
412421 :pr: `28051 ` by :user: `Pierre de Fréminville <pidefrem> `.
413422
414- - |Feature | :func: `sklearn.metrics.check_scoring ` now returns a multi-metric scorer
415- when `scoring ` as a `dict `, `set `, `tuple `, or `list `. :pr: `28360 ` by `Thomas Fan `_.
416-
417423- |Fix |:class: `metrics.classification_report ` now shows only accuracy and not
418424 micro-average when input is a subset of labels.
419425 :pr: `28399 ` by :user: `Vineet Joshi <vjoshi253> `.
@@ -422,8 +428,8 @@ Changelog
422428 computation. This is likely to affect neighbor-based algorithms.
423429 :pr: `28692 ` by :user: `Loïc Estève <lesteve> `.
424430
425- - |API | :func: `metrics.precision_recall_curve ` deprecated the keyword argument ` probas_pred `
426- in favor of `y_score `. `probas_pred ` will be removed in version 1.7.
431+ - |API | :func: `metrics.precision_recall_curve ` deprecated the keyword argument
432+ ` probas_pred ` in favor of `y_score `. `probas_pred ` will be removed in version 1.7.
427433 :pr: `28092 ` by :user: `Adam Li <adam2392> `.
428434
429435- |API | :func: `metrics.brier_score_loss ` deprecated the keyword argument `y_prob `
@@ -434,10 +440,6 @@ Changelog
434440 is deprecated and will raise an error in v1.7.
435441 :pr: `18555 ` by :user: `Kaushik Amar Das <cozek> `.
436442
437- - |Feature | :func: `metrics.d2_log_loss_score ` has been added which
438- calculates the D^2 score for the log loss.
439- :pr: `28351 ` by :user: `Omar Salman <OmarManzoor> `.
440-
441443:mod: `sklearn.mixture `
442444......................
443445
@@ -460,22 +462,22 @@ Changelog
460462 raises a warning when groups are passed in to :term: `split `. :pr: `28210 ` by
461463 `Thomas Fan `_.
462464
465+ - |Enhancement | The HTML diagram representation of
466+ :class: `~model_selection.GridSearchCV `,
467+ :class: `~model_selection.RandomizedSearchCV `,
468+ :class: `~model_selection.HalvingGridSearchCV `, and
469+ :class: `~model_selection.HalvingRandomSearchCV ` will show the best estimator when
470+ `refit=True `. :pr: `28722 ` by :user: `Yao Xiao <Charlie-XIAO> ` and `Thomas Fan `_.
471+
463472- |Fix | the ``cv_results_ `` attribute (of :class: `model_selection.GridSearchCV `) now
464473 returns masked arrays of the appropriate NumPy dtype, as opposed to always returning
465474 dtype ``object ``. :pr: `28352 ` by :user: `Marco Gorelli<MarcoGorelli> `.
466475
467- - |Fix | :func: `sklearn. model_selection.train_test_score
476+ - |Fix | :func: `model_selection.train_test_split ` works with Array API inputs.
468477 Previously indexing was not handled correctly leading to exceptions when using strict
469478 implementations of the Array API like CuPY.
470479 :pr: `28407 ` by :user: `Tim Head <betatim> `.
471480
472- - |Enhancement | The HTML diagram representation of
473- :class: `~model_selection.GridSearchCV `,
474- :class: `~model_selection.RandomizedSearchCV `,
475- :class: `~model_selection.HalvingGridSearchCV `, and
476- :class: `~model_selection.HalvingRandomSearchCV ` will show the best estimator when
477- `refit=True `. :pr: `28722 ` by :user: `Yao Xiao <Charlie-XIAO> ` and `Thomas Fan `_.
478-
479481:mod: `sklearn.multioutput `
480482..........................
481483
@@ -518,6 +520,10 @@ Changelog
518520:mod: `sklearn.utils `
519521....................
520522
523+ - |Fix | :func: `~utils._safe_indexing ` now works correctly for polars DataFrame when
524+ `axis=0 ` and supports indexing polars Series.
525+ :pr: `28521 ` by :user: `Yao Xiao <Charlie-XIAO> `.
526+
521527- |API | :data: `utils.IS_PYPY ` is deprecated and will be removed in version 1.7.
522528 :pr: `28768 ` by :user: `Jérémie du Boisberranger <jeremiedbb> `.
523529
@@ -529,15 +535,11 @@ Changelog
529535 `joblib.register_parallel_backend ` instead.
530536 :pr: `28847 ` by :user: `Jérémie du Boisberranger <jeremiedbb> `.
531537
532- - |API | Raise informative warning message in :func: `type_of_target ` when
533- represented as bytes. For classifiers and classification metrics, labels encoded
538+ - |API | Raise informative warning message in :func: `~utils.multiclass. type_of_target
539+ when represented as bytes. For classifiers and classification metrics, labels encoded
534540 as bytes is deprecated and will raise an error in v1.7.
535541 :pr: `18555 ` by :user: `Kaushik Amar Das <cozek> `.
536542
537- - |Fix | :func: `~utils._safe_indexing ` now works correctly for polars DataFrame when
538- `axis=0 ` and supports indexing polars Series.
539- :pr: `28521 ` by :user: `Yao Xiao <Charlie-XIAO> `.
540-
541543- |API | :func: `utils.estimator_checks.check_estimator_sparse_data ` was split into two
542544 functions: :func: `utils.estimator_checks.check_estimator_sparse_matrix ` and
543545 :func: `utils.estimator_checks.check_estimator_sparse_array `.
@@ -548,4 +550,29 @@ Changelog
548550Thanks to everyone who has contributed to the maintenance and improvement of
549551the project since version 1.4, including:
550552
551- TODO: update at the time of the release.
553+ 101AlexMartin, Abdulaziz Aloqeely, Adam J. Stewart, Adam Li, Adarsh Wase, Adrin
554+ Jalali, Advik Sinha, Akash Srivastava, Akihiro Kuno, Alan Guedes, Alexis
555+ IMBERT, Ana Paula Gomes, Anderson Nelson, Andrei Dzis, Arnaud Capitaine, Arturo
556+ Amor, Aswathavicky, Bharat Raghunathan, Brendan Lu, Bruno, Cemlyn, Christian
557+ Lorentzen, Christian Veenhuis, Cindy Liang, Claudio Salvatore Arcidiacono,
558+ Connor Boyle, Conrad Stevens, crispinlogan, davidleon123, DerWeh, Dipan Banik,
559+ Duarte São José, DUONG, Eddie Bergman, Edoardo Abati, Egehan Gunduz, Emad
560+ Izadifar, Erich Schubert, Filip Karlo Došilović, Franck Charras, Gael
561+ Varoquaux, Gönül Aycı, Guillaume Lemaitre, Gyeongjae Choi, Harmanan Kohli,
562+ Hong Xiang Yue, Ian Faust, itsaphel, Ivan Wiryadi, Jack Bowyer, Javier Marin
563+ Tur, Jérémie du Boisberranger, Jérôme Dockès, Jiawei Zhang, Joel Nothman,
564+ Johanna Bayer, John Cant, John Hopfensperger, jpcars, jpienaar-tuks, Julian
565+ Libiseller-Egger, Julien Jerphanion, KanchiMoe, Kaushik Amar Das, keyber,
566+ Koustav Ghosh, kraktus, Krsto Proroković, ldwy4, LeoGrin, lihaitao, Linus
567+ Sommer, Loic Esteve, Lucy Liu, Lukas Geiger, manasimj, Manuel Labbé, Manuel
568+ Morales, Marco Edward Gorelli, Maren Westermann, Marija Vlajic, Mark Elliot,
569+ Mateusz Sokół, Mavs, Michael Higgins, Michael Mayer, miguelcsilva, Miki
570+ Watanabe, Mohammed Hamdy, myenugula, Nathan Goldbaum, Naziya Mahimkar, Neto,
571+ Olivier Grisel, Omar Salman, Patrick Wang, Pierre de Fréminville, Priyash
572+ Shah, Puneeth K, Rahil Parikh, raisadz, Raj Pulapakura, Ralf Gommers, Ralph
573+ Urlus, Randolf Scholz, Reshama Shaikh, Richard Barnes, Rodrigo Romero, Saad
574+ Mahmood, Salim Dohri, Sandip Dutta, SarahRemus, scikit-learn-bot, Shaharyar
575+ Choudhry, Shubham, sperret6, Stefanie Senger, Suha Siddiqui, Thanh Lam DANG,
576+ thebabush, Thomas J. Fan, Thomas Lazarus, Thomas Li, Tialo, Tim Head, Tuhin
577+ Sharma, VarunChaduvula, Vineet Joshi, virchan, Waël Boukhobza, Weyb, Will
578+ Dean, Xavier Beltran, Xiao Yuan, Xuefeng Xu, Yao Xiao
0 commit comments