Add option to retrieve fold info

Simple addition to the HepNet predict() function to retrieve information about which event was train/val/test in which fold.

@fsauerbu , please let me know whether you think this addition is useful or not. Basically, we wanted to easily retrieve this information in order to make plots for each fold separately. Maybe there is another way I'm overseeing. Cheers!

By Benjamin Paul Jaeger on 2020-10-19T02:56:01 (imported from GitLab)

added 1 commit

• d1b96ab1 - Add option to retrieve fold info in HepNet.predict

Compare with previous version

By Benjamin Paul Jaeger on 2020-10-19T02:57:36 (imported from GitLab)

added Type::Feature request label

Hi @bejaeger,

that's a good idea. You could use the cross-validator object directly to select subsets of the dataframe and plot only a subset. I wonder, if one should move the retrieve_fold_info logic to the CV object and simply call it from predict().

added 1 commit

• 176299b5 - move fold info retrieval to cross validator class

Compare with previous version

By Benjamin Paul Jaeger on 2020-10-21T23:57:49 (imported from GitLab)

Hi @fsauerbu , yes that makes total sense to me. I changed the implementation and added another function to select the cross validator set. Let me know if this is what you were envisioning, happy to adapt it further!

By Benjamin Paul Jaeger on 2020-10-22T00:01:08 (imported from GitLab)

assigned to @fsauerbu

added 2 commits

• a7701cdb - Add tests for fold info retrieval
• 8cf2a8e8 - Fix net naming scheme for unit tests

Compare with previous version

I think this is great. Thank you. I've added unit tests for the new functions. Once the pipeline succeeds, I'll merge it.

merged