Processoptimizer

Changes

- Allow user defined bounds on noise level of WhiteKernel

Bugfixes

Changes

- Added the option to display uncertainty in 2d plots in plot_objective
- Initial efforts to streamline the input-structure of plot options
- Additional options for plots

Bugfixes

Changes

- Dependence modul now consistenty returns arrays instead of lists
- House-keeping on Github (contribution guidelines etc)

Bugfixes

- Bokeh_plot is repaired after we started returning the std to plots
- LHS is rewritten to ensure consistent returns in between real and integer
dimensions (integer types are ensure to return values "close" to those of a
corresponding real dimension)

Changes

- New plot-type to envision model coverage
- Kriging Believer now supports multiobjective opt
- Examples pruned to better reflect the purpose of ProcessOptimizer as a tool
for optimizing real world physical/chemical processes
- Expected_minimum can now return both maximum and minimum and can return the
expected std in the points. Works for both numerical and categorical dimensions
- QoL improvements with easy impoart of most used features through \_\_init\_\_
.py
- Add possibility to show ~95% credibility_intervals in plot_objective

Bugfixes

- More linting
- Supports Scikit-Learn 1.0.0
- plot_pareto works with partially categorical spaces
- Consolidate tests

Changes

- Kriging Believer added for batch-mode optimization
- Added Interactive Pareto plotting

Bugfixes

- LHS fixed to ensure randomization of dimensions
- More linting

Changes

- Added plot_expected_minimum_convergence
- Numerous format changes to satisfy Flake8
- Fixed deprecation warning from Numpy
- Merge keywords for batch-optimization
- Set LHS=True as default
- Extensive changes in test-suite

Bugfixes

- expected_minimum_sampling refactored
- Fix URL to present images on pypi
- Fixed Bokeh-plot to optional dependency
- Dependencies set (Bokeh optional, pyYAML added)