Credibilistic Isodata Algorithm

In this paper, we propose a clustering algorithm meant for data sets consisting of imprecise or fuzzy information. The proposed algorithm is an analogue of Iterative Self Organizing Data Analysis Technique (ISODATA) algorithm meant for crisp data sets. The algorithm uses measures related to imprecise information developed using the Credibility Theory [6]. An illustrative numerical example based on a bench mark data set is given to demonstrate its utility. The proposed algorithm is compared with the credibilistic means algorithm [7] in terms of certain cluster validity measures and found to perform well