Skip to content

Conversation

@jvaquet
Copy link

@jvaquet jvaquet commented May 16, 2025

Hey there,
I got in touch (#1677) a month ago about implementing the SAM-kNN Classifier for river. I think the implementation is pretty much ready, let me know what you think!

A couple remarks:

  • In order to pass the tests, I disabled check_emerging_features, check_disappearing_features and check_radically_disappearing_features in 05d2ea2/ 96a5618 as the classifier only works with a consistent feature set. Is this the intended way to go about this or did I miss something?
  • Currently the classifier only works for real-valued features. I would like to try adding categorical feature support by adding an option to select the LTM compression method between kmeans and kmodes. kmeans is the currently used default method and uses the sklearn implementation. kmodes would be another library that is not a dependency of river. Can this be implemented using a conditional import only when the option is selected? What is the project's policy on that?

Thanks everyone!

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.

1 participant