Rough Clustering Using an Evolutionary Algorithm

Abstract

Cluster analysis is a fundamental technique in traditional data analysis and many clustering methods have been identified, including the commonly used k-means approach, which requires the number of clusters to be specified in advance and is dependent on initial starting points. We present an evolutionary-based rough clustering algorithm, which is designed to overcome these limitations. Rough clusters are defined in a similar manner to Pawlak's rough set concept, with a lower and upper approximation, allowing multiple cluster membership for objects in the data set. The paper describes the template, the data structure used to describe rough clusters. It also provides an overview of the evolutionary algorithm used to develop viable cluster solutions, consisting of an optimal number of templates providing descriptions of the clusters. This algorithm was tested on a small data set and a large data set.