Problem:
The generated Zliobaite data samples the RESULTS of classifiers with different fairness levels.
We have to come up with a method to generate data ourselves, as Benjamin said, to train our own LVQ
classifiers (or others) to be able to compare it to the data.
Sorry. I misunderstood the paper.