Evaluation on the training set in example notebook

[zachschillaci27]

Looking at the example in 06-threshold-optimization.ipynb, it appears that you are fitting the RouteLayer model on a dataset and then evaluating the model on that same dataset. If that's indeed the case, then it's no surprise the accuracy increases dramatically. Did you perhaps also evaluate the trained model on an unseen dataset? I would be curious to know how the performance generalizes. Thanks!

[ChristianWeyer]

I actually had the exact same question...