Thanks to visit codestin.com
Credit goes to github.com

Skip to content

Conversation

@joaquindas
Copy link
Contributor

Sklearn deprecated force_all_finite param for check_array(). Docs here. There is a deprecation warning when fitting the model. check_array function is used twice in hdbscan_.py. Changed both to new param naming ensure_all_finite, functionality is identical.

@JelmerBot
Copy link
Collaborator

This change also increases the minimum required scikit-learn version from 0.2 to 1.6. Maybe it is worth testing which version is available at run-time to keep support for older scikit-learn versions?

@joaquindas
Copy link
Contributor Author

If dependencies were declared in pyproject.toml, it would be straightforward to specify scikit-learn version for given python versions. Is there interest in setting that up instead of requirements.txt?

@betatim
Copy link

betatim commented Dec 9, 2025

scikit-learn 1.8 will probably be released today. Depending on how many past versions of scikit-learn you want to support it might be worth investigating https://github.com/sklearn-compat/sklearn-compat#validate_data-function to make it easier to deal with the various deprecations.

@chrishalcrow
Copy link

Hello, we use (and love) hdbscan in the spikeinterface package. Would be great if this, or a similar PR, could make hdbscan compatible with scikit-learn v1.8 :)

@lmcinnes lmcinnes merged commit 26906d9 into scikit-learn-contrib:master Dec 11, 2025
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants