Closed danielhollas closed 1 year ago
Patch and project coverage have no change.
Comparison is base (
d644768
) 79.63% compared to head (741e15c
) 79.63%.
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Do you have feedback about the report comment? Let us know in this issue.
@unkcpz thanks! I'll merge this to unblock the other PRs.
We should have probably put scikit-learn~1.0
to have more freedom on the versions
I've done this on purpose to make sure things work for now. As explained in the OP I want to get rid of it altogether soon.
scikit-learn build for version 0.24 started failing due to new version of Cython so we bump the version to 1.0. Per docs, this version does not bring any breaking changes. https://scikit-learn.org/1.0/auto_examples/release_highlights/plot_release_highlights_1_0_0.html
We also only use the sklearn.PCA in SMILES widget, I've verified that it still works and checked the release notes that there are not API changes to this method. https://scikit-learn.org/1.0/whats_new/v1.0.html#changes-1-0
In the near future I'd like to get rid of scikit-learn altogether since it is a very heavy dependency and we only use it for PCA to reorient the molecules generated from SMILES by RDkit. I have a suspicion that this step is not really needed, and if it is, we can simply vendor the PCA function.