Rough Clustering and Regression Analysis

摘要

Since Pawlak introduced rough set theory in 1982 [1] it has gained increasing attention. Recently several rough clustering algorithms have been suggested and successfully applied to real data. Switching regression is closely related to clustering. The main difference is that the distance of the data objects to regression functions has to be minimized in contrast to the minimization of the distance of the data objects to cluster representatives in k-means and k-medoids. Therefore we will introduce rough switching regression algorithms which utilizes the concepts of rough clustering algorithms as introduced by Lingras at al. [2] and Peters [3].