Are raw results available?

[PGijsbers]

Great work creating an easy-to-use benchmark and a common interface to many different algorithms 👏

I just read the paper and am interested in working with:

• the (auroc) score of each individual configuration (i.e., each hyperparameter configuration from the grid search), instead of just the aggregate performances shown in the paper
• the decision function values for each individual configuration (to be able to compute metrics)

I wasn't able to find either, are they available or could they be made available? Thanks!

[kayzliu]

Thanks for your attention! Unfortunately, we have not stored these intermediate data, but it will not be too hard to obtain them with slightly modification on our benchmark script. The hyperparameters can be retrieved at init_model in utils.py, and the decision scores are available in score in each script.