A framework of fuzzy partition based on Artificial Bee Colony for categorical data clustering

Abstract

Fuzzy k-partition (FkP) is an effective clustering technique, which is mathematical model based. Thus, the objective function of FkP is a nonlinear function. Membership random selection is featured by an iterative process, which results in local optima traps easily. It is important to find global optimal consider to nonlinear objective function of the problem. Moreover, Artificial Bee colony (ABC) has ability and efficiently used for multivariable, multinomial function optimization. To this, this paper proposes the hybridization of FkP based on Artificial Bee colony (ABC) a population based algorithm. Some of benchmarks data sets have been elaborated to test the proposed approach. The experiment shows that FkP ABC obtains better results in term of the dun index validity clustering as compared to the baseline algorithm.