Abstract

Data mining is the discovery of patterns that may exist implicitly in a large database. In this paper, we study a combined model for cluster discovery. We build a multi dimensional data model (MDDM) and transform tuple of database into vector of MDDM then we use Kohonen's self-organizing algorithm to discover the potential clusters and Genetic Algorithm (GA) for validating these clusters. By combining MDDM with GA, we propose a heuristic for improving the efficiency of cluster discovery.