Skip to content
This repository was archived by the owner on Feb 28, 2024. It is now read-only.

[MRG] compatibility update #939

Merged
holgern merged 7 commits into from
Sep 3, 2020
Merged

[MRG] compatibility update #939

holgern merged 7 commits into from
Sep 3, 2020

Conversation

@lucasplagwitz
Copy link
Contributor

@lucasplagwitz lucasplagwitz commented Sep 1, 2020 •
edited
Loading

Change few lines to make it work with newest versions of scikit-learn and numpy. Short overview:

  • new GP in scikit-learn >=0.23: add y_normalize=False
  • new GP in scikit-learn >= 0.23: often throws nan error -> increase initial runs
  • numpy MaskArray -> numpy.ma.array
  • set matplotlib backend to Agg via export for conf.py

@rkern rkern mentioned this pull request Sep 1, 2020
Merged
@pep8speaks
Copy link

pep8speaks commented Sep 1, 2020 •
edited
Loading

Hello @lucasplagwitz! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻

Comment last updated at 2020-09-01 22:26:15 UTC

@lucasplagwitz
Copy link
Contributor Author

lucasplagwitz commented Sep 2, 2020

The scikit-learn NaN (normalize_y=True and one point => std=0) error with GP version >= 0.23 is now mentioned in the issue board: scikit-learn#18318. In future versions the n_initial_run increasment might not be necessary anymore, but I don't think they do any harm.

@holgern
Copy link
Contributor

holgern commented Sep 3, 2020

Thanks @lucasplagwitz , can you add [MRG] into the title of your pull request, when your PR is ready for review?

As soon as you added [MRG], I will start to review. Thanks again.

@lucasplagwitz
Copy link
Contributor Author

lucasplagwitz commented Sep 3, 2020

Thanks @holgern for your hint. Add the tag immediately. In case of merge, please be so kind and make a release so we can close #931, too. I saw the instruction on your readme but am not totally sure if it should be new 0.8....
Thanks for reviewing - hope it helped to bring scikit-optimize back to just work after pip/conda install.

@lucasplagwitz lucasplagwitz changed the title compatibility update [MRG] compatibility update Sep 3, 2020
holgern
holgern approved these changes Sep 3, 2020
View reviewed changes
Copy link
Contributor

@holgern holgern left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Everything looks good to me, thanks.

@holgern holgern merged commit 08431c6 into scikit-optimize:master Sep 3, 2020
This was referenced Sep 3, 2020
Closed
Closed
Closed
HunterMcGushion added a commit to HunterMcGushion/hyperparameter_hunter that referenced this pull request Oct 20, 2020
@HunterMcGushion
Pin SKLearn to 0.22.2 to fix Bayesian optimization NaN errors with si…
4ee6203 
…ngle point

- See:
  - scikit-optimize/scikit-optimize#939
  - scikit-learn/scikit-learn#15782

- Alternative could be to change `y_normalize` given to `GaussianProcessRegressor` in `optimization.backends.skopt.engine` (https://github.com/HunterMcGushion/hyperparameter_hunter/blob/cf4a13765a6c871909418e59596260a7f0628fce/hyperparameter_hunter/optimization/backends/skopt/engine.py#L769) to False instead of True
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Reviewers

1 more reviewer

@holgern holgern holgern approved these changes

Reviewers whose approvals may not affect merge requirements

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@lucasplagwitz @pep8speaks @holgern