optuna 0.18.0

v0.18.0

on Python PyPI

This is the release note of v0.18.0. See here for the complete list of solved issues and merged PRs.

Upcoming Python 2 Support Drop

Due to the end-of-life (EOL) of Python 2 in January 2020, Optuna will drop Python 2 support in December 2019.

This decision was made considering the following facts:

• Python 2 will become end-of-life (EOL) in January 2020.
• Many scientific computation packages, including NumPy, which is one of the core dependency of Optuna, are planning or already started to drop support for Python 2.

We plan to drop Python 2 support in the first release in December 2019.

Compatibility

• Remove OptunaConfig. (#653)
• Catch no exceptions in optimize by default. (#638)

New Features

• Add function for high-dimensional parameter relationships visualization. (#594, thanks @suecharo!)
• Add function for parameter relationships visualization as slice plot. (#540, thanks @suecharo!)
• Add function for parameter relationships visualization as contour plot. (#539, thanks @suecharo!)
• Add a PyTorch Ignite handler for pruning. (#561)
• Add PyTorch Lightning pruning integration. (#597)
• Add FastAI Callback for pruning. (#585, thanks @crcrpar!)
• Add public APIs for checking visualization availability. (#607)
• [Experimental] Add a prototype version of automatic LightGBM tuner with stepwise logic. (#549)

Enhancements

• Improve handling of reported values. (#659)
• Avoid iterating through all trials unnecessarily. (#641, thanks @dwiel!)
• Remove TrialModel.where_study_id. (#640)
• Record all error raising trials as failures. (#637)
• Speed up Parzen Estimator. (#554, thanks @oda!)
• Cache best trial_id. (#535, thanks @oda!)
• Add an option to force_garbage_collection for every trial. (#533, thanks @oda!)
• Fix plot_optimization_history. (#631, thanks @cafeal!)

Bug Fixes

• Add type casting to float. (#644)
• Add missing call to check function. (#598)

Documents

• Add PULL_REQUEST_TEMPLATE.md file. (#651)
• Update issue template. (#652)
• Update CONTRIBUTING.md file. (#649)
• Update the experimental warning of OptunaSearchCV. (#639)
• Update BaseStudy.trials doc to specify order constraint. (#634)
• Add visualization functions to API reference doc. (#615)
• Apply a small fix to the docstring of Study. (#591)
• Add links to nbviewer. (#590, thanks @upura!)

Examples

• Add a distributed optimization example on Kubernetes. (#643, thanks @AnesBenmerzoug!)
• Add --pruning option to the PyTorch Ignite example. (#633)
• Fix an example for tensorflow==1.15.0. (#620)
• Add pruning to Chainer example. (#614, thanks @Crissman!)
• Enclose plot_slice() of slice plot example in backquotes. (#608, thanks @crcrpar!)
• Add PyTorch Lightning example. (#584)

Tests

• Fix assertions in test_optimize_with_catch. (#650)
• Fix the module of type variables. (#636)
• Add ValueError test to test_get_contour_plot(). (#616, thanks @crcrpar!)

Continuous Integration

• Skip XGBoost tests and examples with Python 2.7 in CircleCI. (#654)
• Hide pip progress bar from CircleCI log. (#648)
• Align CircleCI install-examples step name to anchor. (#618)

Code Fixes

• Use indices to align with the others. (#621, thanks @crcrpar!)
• Replace useless None check with assertion. (#610)