vanderschaarlab / autoprognosis

A system for automating the design of predictive modeling pipelines tailored for clinical prognosis.
https://www.autoprognosis.vanderschaar-lab.com/
Apache License 2.0
97 stars 26 forks source link

feat: add option for stratified CV #35

Closed HLasse closed 1 year ago

HLasse commented 1 year ago

Adds the id parameter to Classifier|RegressionStudy which is passed downwards to use StratifiedGroupKFold for CV if id is specified. Fixes #33

Handling is done in dataframe_preprocess.

Added a simple test for integration mainly.

Hope this helps! Very open to any and all comments, @bcebere!

bcebere commented 1 year ago

Thank you for the contribution, @HLasse !

I just moved the group_id parameter to the search method. After all the tests pass, I will merge the PR! Thanks!