A fair-multicluster approach to clustering of categorical data

Abstract

In the last few years, the need of preventing classification biases due to race, gender, social status, etc. has increased the interest in designing fair clustering algorithms. The main idea is to ensure that the output of a cluster algorithm is not biased towards or against specific subgroups of the population. There is a growing specialized literature on this topic, dealing with the problem of clustering numerical data bases. Nevertheless, to our knowledge, there are no previous papers devoted to the problem of fair clustering of pure categorical attributes. In this paper, we show that the Multicluster methodology proposed by Santos and Heras (Interdiscip J Inf Knowl Manag 15:227–246, 2020. https://doi.org/10.28945/4643) for clustering categorical data, can be modified in order to increase the fairness of the clusters. Of course, there is a tradeoff between fairness and efficiency, so that an increase in the fairness objective usually leads to a loss of classification efficiency. Yet it is possible to reach a reasonable compromise between these goals, since the methodology proposed by Santos and Heras (2020) can be easily adapted in order to get homogeneous and fair clusters.