All fairness algorithms have some notion of a protected group based on the sensitive attribute. In case of facility location, the sensitive attribute is population density. To find the exact range of population densities that belong to one protected group, Bell's curve and z score can be leveraged.
To Do
From the dataset available in the repository, a histogram can be plotted based on population density. It is expected that, if right scale and ranges are chosen, a normal distribution would be obtained. Once the normal distribution is obtained, ± k z's should naturally output the protected groups
Description
All fairness algorithms have some notion of a protected group based on the sensitive attribute. In case of facility location, the sensitive attribute is population density. To find the exact range of population densities that belong to one protected group, Bell's curve and z score can be leveraged.
To Do
From the dataset available in the repository, a histogram can be plotted based on population density. It is expected that, if right scale and ranges are chosen, a normal distribution would be obtained. Once the normal distribution is obtained, ± k z's should naturally output the protected groups
Example