Releases: scikit-optimize/scikit-optimize
v0.9.0
What's Changed
- [MRG]Add hollow iterations early stopper by @ludaavics in #917
- Doc line, logspace call, set_transformer default parameter misstype, … by @GCidd in #921
- Revert change to plot_histogramm by @holgern in #942
- [MRG] Fixes GaussianProcessRegressor for sklearn >= 0.23 by @holgern in #943
- [MRG] Refactor Sobol' API by @tupui in #955
- docs: fix simple typo, stategies -> strategies by @timgates42 in #979
- [MRG] Correct documented default is_int for Normalize. by @gcp in #1019
- [MRG] add ThresholdStopper() by @GuillaumeSimo in #1000
- [MRG] Check whether prior values are valid. by @gcp in #1020
- Fix numpy Deprecation errors by @xmatthias in #1023
- Remove BayesSearchCV(iid=) parameter deprecated in sklearn 0.24 by @kernc in #988
- Re-add
BayesSearchCV.best_score_
needed by some examples by @kernc in #1031 - Revert "Re-add
BayesSearchCV.best_score_
needed by some examples" by @kernc in #1032 - Use deepcopy to prevent reference cycles in Optimizer by @freddyaboulton in #1029
- [MRG] Fixed grammatical error in docs: 'less' vs 'fewer' by @samuelstevens in #1040
- [MRG] update packges in circleci deps by @QuentinSoubeyran in #1069
- [MRG] Fix GaussianProcessRegressor for scikit-learn 1.0 by @QuentinSoubeyran in #1063
- Fix remaining sklearn 1.0
return_X_y=
error by @kernc in #1078 - MNT: Fix CircleCI builds by unpinning + full docs for PRs by @kernc in #1079
- MNT: Fix "LaTeX Error: File 'tgtermes.sty' not found." on CircleCI by @kernc in #1080
New Contributors
- @ludaavics made their first contribution in #917
- @GCidd made their first contribution in #921
- @tupui made their first contribution in #955
- @timgates42 made their first contribution in #979
- @GuillaumeSimo made their first contribution in #1000
- @xmatthias made their first contribution in #1023
- @kernc made their first contribution in #988
- @freddyaboulton made their first contribution in #1029
- @samuelstevens made their first contribution in #1040
- @QuentinSoubeyran made their first contribution in #1069
Full Changelog: v0.8.1...v0.9.0
v0.8.1
v0.8.0
-
n_jobs support was added to Optimizer and fixed for forest_minimize #884
-
Allow dimension selection for plot_objective and plot_evaluations and add plot_histogram and plot_objective_2D. Plot code has been refactored. #848
-
Initial sampling generation from latin hypercube, sobol, hammersly and halton is possible and can be set in all optimizers #835
-
Improve sampler and add grid sampler #851
-
Fix library for scikit-learn >= 0.23. numpy MaskArray is replaced by numpy.ma.array. y_normalize=False has been added and initial runs has been increased. #939
-
Integer transform and inverse_transform for normalize #880
-
Add is_constant property to dimension and n_constant_dimensions property to Space #883
-
Skip constant dimensions for plot_objective and plot_evaluations to allow plots using BayesSearchCV #888
-
Fix Fix Optimizer for full categorical spaces #874
-
Improve circle ci #852
-
Add project toml and adapt minimal numpy, scipy, pyyaml and joblib version in setup.py #850
v0.7.4
v0.7.3
v0.7.2
Version 0.7.2
New features
- Add expected_minimum_random_sampling
- New plot examples
- Add more parameter to plot_objective
- Return ordereddict in point_asdict
- update_next() and get_results() added to Optimize
Bug fixes
- Fix searchcv rank (issue #831)
- Fix random forest regressor (issue #766)
- Fix doc examples
- Fix integer normalize by using round()
- Fix random forest regressor (Add missing min_impurity_decrease)
Maintenance
- Fix license detection in github
- Add doctest to CI
v0.7.1
Version 0.7.1
New features
- Sphinx documentation
- notebooks are replaced by sphinx-gallery
- New StringEncoder, can be used in Categoricals
- Remove string conversion in Identity
- dtype can be set in Integer and Real
Bug fixes
Maintenance
- Old pdoc scripts are removed and replaced by sphinx
v0.7
Version 0.7
New features
- Models queue has now a customizable size (model_queue_size).
- Add log-uniform prior to Integer space
- Support for plotting categorical dimensions
Bug fixes
- Allow BayesSearchCV to work with sklearn 0.21
- Reduce the amount of deprecation warnings in unit tests
Maintenance
- joblib instead of sklearn.externals.joblib
- Improve travis CI unit tests (Different sklearn version are checked)
- Added
versioneer
support, to keep things simple and to fix pypi deploy