A new comparison framework to evaluate Fuzzy K-Means based clustering approaches

Fuzzy K-Means (FKM) is an important data clustering technique. FKM groups data objects into sets of disjoint clusters such that the objects within a cluster are highly similar and dissimilar with the objects in the other clusters. In recent years, many clustering approaches have been done based on the FKM algorithm and its variations. In order to determine how FKM-based clustering approaches have developed over recent years; this article reviews FKM-based clustering approaches through a survey of the literature from 2004 to 2018. To fulfill this goal, we introduce a new comparison framework, which is composed of 9 criteria that include: input source, representation schema, scale, scope, grouping algorithm, application, dynamicity, evaluation measure and language. Then using this comparison framework, we discuss the development of the FKM-based clustering in recent years and specify some important open problems that have emerged in the area of FKM-based clustering.