Feature Importance from classification on distance matrix

I’m using the pipeline classify-sample-from-dist of the sample classifier plugin (2018.8 version). Is there a way to get the feature importance artifact (as possible in classify-samples)?
The output of the plugin includes predictions.qza and accuracy_results.qzv.

Hi @priscilla_b,
No, that classifier does not generate feature importances because it uses a KNN classifier, which cannot be used to estimate feature importances. You will need to use classify-samples with an appropriate classifier to extract feature importances.

I hope that helps!

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.