A partitive rough clustering algorithm

Abstract

Since rough sets were introduced by Pawlak about 25 years ago they have become a central part of soft computing. Recently Lingras presented a rough k-means clustering algorithm which assigns the data objects to lower and upper approximations of clusters. In our paper we introduce a rough k-medoids clustering algorithm and apply it to four different data sets (synthetic, colon cancer, forest and control chart data). We compare the results of these experiments to Lingras rough k-means and discuss the strengths and weaknesses of the rough k-medoids. © Springer-Verlag Berlin Heidelberg 2006.