Using sample classifier to predict/validate classification in a different feature table

Hi Guys,

Is there a way to test the trained sample classifier (or regressor) to predict or validate classification (or continuous metadata) in a different feature table or data set? I’m looking for a way to test whether the trained model is generalizeable, and also need a sanity check to make sure it’s not the result of over-fitting.

Thanks!
Gil

Hi @Gil_Sharon

This is a duplicate of this thread. Short answer: not yet with the CLI, but this can be done in Artifact API.

1 Like

Thanks @Nicholas_Bokulich. Does the installation of the Artifact API change anything for the q2cli? I want to use them side by side (essentially using the API for specific tasks, and the q2cli for others). the installation of the Arifact API requires the feature-table to be replaced, and I want to make sure it doesnt change anything for the command line interface.

Thanks!

Hi @Gil_Sharon - I might be misunderstanding your question, but the short answer is that Interfaces in QIIME 2 are wholly independent! You can perform one step of an analysis in the Studio, then jump over to the CLI, and wrap up using the Artifact API!

The Artifact API is distributed with QIIME 2 (it is rolled into the Framework), so you actually don’t need to do anything to “install” it, just go ahead and open up a notebook or ipython session and get started! I would recommend reviewing the API docs before getting in too deep. Good luck!

3 Likes

Thanks! I should’ve tested before writing… plugins are imported with no problems.

1 Like

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