Rough Clustering using Credibilistic Critical Values

In this paper, the effectiveness of using credibilistic critical values in crisp conversion of fuzzy data is studied. The conversion assumes significance when fuzzy data are considered for the purpose of clustering. In this paper, performance of the well-known clustering algorithm namely rough k-means is evaluated under credibilistic critical value crisp conversion. Comparative studies have been carried out with the help of two artificial data sets describing different environments. In the comparative study, two clustering algorithms namely Lingras and West Rough k-means algorithm and Peters’ Rough k-means algorithm have been considered. The well-known rough clustering validity measure,namely, David Bouldin index is employed in this study.