Categorical data analysis using multiobjective differential evolution based fuzzy clustering

During the last one decade, rapid proliferation of categorical data attracts computer scientists and engineers to analyse the categorical data. In this regard, unsupervised technique, such as clustering has been used by inception of new algorithms or modification of the existing ones. These methods are basically optimizing single objective function to get the partitions. However, optimization of multiple conflicting objectives simultaneously may evolve better clustering results as the multiobjective optimization techniques have been successfully applied in various fields of engineering and science. Hence, in this article, Multiobjective Differential Evolution based Fuzzy Clustering for Categorical Data is proposed. For this purpose, differential evolution is used as an underlying optimization technique. Moreover, the index encoding scheme is used to encode the vector in differential evolution and after the mutation operation in differential evolution, scaling is introduced to adjust the encoded index value within the permissible range of categorical objects. The performance of the proposed method has been demonstrated by comparing it with the widely used state-of-the-art methods for two synthetic and two real life data sets. Finally, statistical test has been conducted to judge the superiority of the proposed method.