Samuel Hoffman

Can you explain a little more what you mean by that? What kind of fairness metric are you interested in? Satisfying group fairness requires sensitive attributes to partition the samples...

@monindersingh I have a question about MEPS. In the pre-processing, you included `'EDUCYR'` and `'HIDEG'`: https://github.com/Trusted-AI/AIF360/blob/963df2e4eea807bd5765fee9f1c594500bdcbb5b/aif360/datasets/meps_dataset_panel19_fy2015.py#L45 but in the `features_to_keep` arg of `__init__` they're not there: https://github.com/Trusted-AI/AIF360/blob/963df2e4eea807bd5765fee9f1c594500bdcbb5b/aif360/datasets/meps_dataset_panel19_fy2015.py#L80-L85 Essentially this means...

I don't really have skeletons for these but I thought examples would be enough: For the MEPS dataset, a good place to start might be the [COMPAS port](https://github.com/IBM/AIF360/blob/master/aif360/sklearn/datasets/compas_dataset.py). For differential...

Sorry! I got this confused with another issue. I have some work in progress for the first two but rich subgroup fairness is still outstanding. @mkrueger12

According to this, https://wires.onlinelibrary.wiley.com/doi/pdf/10.1002/widm.1452 (Section 3.1.5), we should be using: age60 vs 25

Did you clone the full repo from GitHub or just download the single demo notebook file? In the readme we recommend cloning and installing with `pip install -e '.[all]'` so...

Yes, this will allow DataFrames to be used directly with the algorithms. Reweighing is already implemented so you can try it out if you're comfortable using the master branch from...

@jimbudarz, if you encode your categorical data with `pd.get_dummies()`, or use `StandardDataset`, you will end up with `feature_names` that look like, e.g., `[..., native-country=United-States, native-country=Vietnam, native-country=Yugoslavia, ...]`. Then, if you...

> It might be helpful to explain what sep attribute does in the https://aif360.readthedocs.io/en/latest/modules/datasets.html documentation. That's a good point. Would you be willing to write a quick PR to that...

It looks like this is what sklearn does (e.g. [recall_score](https://scikit-learn.org/stable/modules/generated/sklearn.metrics.recall_score.html)) and what we do in the [sklearn-compatible metrics](https://github.com/Trusted-AI/AIF360/blob/master/aif360/sklearn/metrics/metrics.py#L167-L170) as well. However, it seems like a really breaking issue if your...