Version 0.8.1

September 2020

• Fix GaussianProcessRegressor on sklearn 0.23 normalizes the variance to 1, which needs to reverted on predict.

Version 0.8.0

September 2020

skopt.Optimizer

• Enhancement n_jobs support was added to Optimizer and fixed for forest_minimize #884 by Holger Nahrstaedt based on #627 by JPN

skopt.plots

• Enhancement Allow dimension selection for plot_objective and plot_evaluations and add plot_histogram and plot_objective_2D. Plot code has been refactored. #848 by Holger Nahrstaedt based on #579 by Hvass-Labs

skopt.sampler

• Major Feature Initial sampling generation from latin hypercube, sobol, hammersly and halton is possible and can be set in all optimizers #835 by Holger Nahrstaedt

• Enhancement Improve sampler and add grid sampler #851 by Holger Nahrstaedt

skopt.searchcv

• Fix 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. :pr: 939 by Lucas Plagwitz

skopt.space

• Fix Fix Integer transform and inverse_transform for normalize #880 by Holger Nahrstaedt

• Enhancement Add is_constant property to dimension and n_constant_dimensions property to Space #883 by Holger Nahrstaedt

• Enhancement Skip constant dimensions for plot_objective and plot_evaluations to allow plots using BayesSearchCV #888 by Holger Nahrstaedt

skopt.utils

• Fix Fix Optimizer for full categorical spaces #874 by Holger Nahrstaedt

Miscellaneous

• Improve circle ci #852 by Holger Nahrstaedt

• Add project toml and adapt minimal numpy, scipy, pyyaml and joblib version in setup.py #850 by Holger Nahrstaedt