Fast Implementation of Dual Clustering Algorithm for Spatial Data Mining

Many applications scenarios require spatial clustering results in which a cluster has not only high proximity in geometrical domain but also high similarity in non-geometrical domain. Such clustering problem is called dual clustering. We proposed a new algorithm for solving such problem. We first implemented density-based sampling on spatial dataset to reduce data size, and then we partitioned the sample to different clusters in such a way that each cluster forms a compact region in geometrical domain while has the similarity in non-geometrical domain. The experimental results show our algorithm is very effective and efficient.