bramreinders97

The idea behind the `validate_and_reformat` function is the following: The keys of `threshold_dict` represent the values of the sensitive feature column(s). If there is just one sensitive feature, these values...

> Instead of using tuples of strings as dict keys I'd suggest utilizing something similar to the existing functionality that maps such tuples to a single string with the difference...

>Where do I find this code? This is a screenshot so I couldn't copy. The tests all ran fine (after the import adjustments mentioned in other comments) for me locally...

> I am struggling to see the difference between what you're proposing and what is already implemented in reformat_threshold_dict_keys(). The main reason I proposed using tuples as dict keys is...

In an offline discussion with @hildeweerts this morning, we decided that we'd add a keyword argument `default_threshold` to which we compare groups that are not mentioned in `threshold_dict`. This seems...

In the last commit I've added a first version of the user guide for `Thresholder`. Please let me know if you miss something/anything is unclear/other feedback @adrinjalali @romanlutz @hildeweerts. There...

> Aren't these both .4? > I'm conflicted as to whether we should allow this... > These are the same. How come this does anything different at all? I will...

>To check if I understand correctly, if we would go for this option I should: > - Move the short description I have here to the dataset user guide >...

>Don't forget to add an entry to https://fairlearn.org/main/user_guide/installation_and_version_guide/v0.7.1.html I've added two bullet points, is that all that you meant I should do for this?

I have added a first version of `RejectOptionClassifier` to the postprocessing package, and included a user guide in the mitigation section. Feedback is very welcome :) Some questions/remarks I have:...