Spatial Clustering for Data Mining with Genetic Algorithms

This paper develops a genetic search heuristic for solving medoid based clustering problems based on Random Assorting Recombination and results show improvements on the geneticsearch heuristic.

Abstract

Spatial data mining is the discovery of interesting relationships and characteristics that may exist implicitly in spatial databases. The identification of clusters in spatially referenced data provides a means of generalization of the spatial component of the data associated with a Geographical Information System. A variety of clustering formulations exists. A non-hierarchical approach in Data-mining applications is to use a medoid based version. This approach has robust behavior with respect to outliers and many heuristics have been developed that find near optimal partitions. This paper develops a genetic search heuristic for solving medoid based clustering problems. We base our genetic recombination upon Random Assorting Recombination. A comparison is made with previous solution approaches. Results show improvements on the genetic search heuristic. Keywords: Data Mining, Spatial data sets, Genetic Algorithms, Clustering. 2 Estivill-Castro & Murray 1 Introduction Geographical In...