Fuzzy clustering of incomplete data based on cluster dispersion

Abstract

Clustering algorithms are used to identify groups of similar data objects within large data sets. Since traditional clustering methods were developed to analyse complete data sets, they cannot be applied to many practical problems, e.g. on incomplete data. Approaches proposed for adapting clustering algorithms for dealing with missing values work well on uniformly distributed data sets. But in real world applications clusters are generally differently sized. In this paper we present an extension for existing fuzzy c-means clustering algorithms for incomplete data, which uses the information about the dispersion of clusters. In experiments on artificial and real data sets we show that our approach outperforms other clustering methods for incomplete data. © 2010 Springer-Verlag Berlin Heidelberg.