A Dynamic Approach to Rough Clustering

R rough set theory is utilized to deal with uncertainty and an engineering like approach to dynamic clustering that is based on rough k-means is suggested.

Abstract

Many projects in data mining face, besides others, the following two challenges. On the one hand concepts to deal with uncertainty - like probability, fuzzy set or rough set theory - play a major role in the description of real life problems. On the other hand many real life situations are characterized by constant change - the structure of the data changes. For example, the characteristics of the customers of a retailer may change due to changing economical parameters (increasing oil prices etc.). Obviously the retailer has to adapt his customer classification regularly to the new situations to remain competitive. To deal with these changes dynamic data mining has become increasingly important in several practical applications. In our paper we utilize rough set theory to deal with uncertainty and suggest an engineering like approach to dynamic clustering that is based on rough k-means.