added demv preprocessing algorithm

[giordanoDaloisio]

We included an implementation of the Debiaser for Multiple Variables algorithm from the paper

Debiaser for Multiple Variables to enhance fairness in classification tasks G d’Aloisio, A D’Angelo, A Di Marco, G Stilo Information Processing & Management 60 (2), 103226

We included the implementation of the algorithm and a notebook showing an example usage

Signed-off-by: Giordano d'Aloisio giordano.daloisio@graduate.univaq.it

[hoffmansc]

Hi @giordanoDaloisio, thanks for your contribution! I have a couple quick suggestions:

• [ ] I think the notebook can be improved. Right now, it doesn't really explain clearly what's happening at each step. Users can always read the paper to understand exactly what's going on but the notebook should at least give a basic overview, explain the advantages compared to other algorithms, demonstrate interesting behavior, etc.
• [ ] There are no assert statements in the notebook so the test file isn't really doing much other than testing for runtime errors. In any case, I would suggest a true unit test file independent of the demo that more thoroughly tests edge cases, etc.
• [ ] Finally, it would be really appreciated if you could include a version compatible with the aif360.sklearn API as well. It looks like you're using DataFrames internally anyway so I hope this isn't too much trouble.

[giordanoDaloisio]

Hi @hoffmansc, thank you for your feedback! I have updated the notebook and the test scripts, and now I will work on implementing DEMV for the sklearn API. I hope the notebook is now more precise and more meaningful.

[giordanoDaloisio]

Hi @hoffmansc, here is the implementation of DEMV for the sklearn api