一种面向GIS空间数据的聚类方法

一个好的聚类算法应该是用户输入参数少,对噪声不敏感,能够发现任意形状,可以处理高维数据,具有可解释性和可扩展性.将聚类分析应用于地理信息系统中,可以实现对GIS数据信息概括和综合.文中提出一种基于距离阈值相邻的聚类算法,通过距离阈值可达的方式逐个将对象加入到已知聚类中,可以发现任意形状的聚类并对噪声数据有很好的分离效果,实验中将该算法应用于地理信息系统中的数据挖掘实现上,结果证明此算法对于实现GIS聚类具有满意的效果.

A GIS Spatial Data Oriented Clustering Approach

A perfect clustering algorithm should have few parameter input by user and be insensitive to noise. It can discover the clusters with arbitrary shape and deal with high dimensional dataset, and also has good interpretability and expansibility. GIS data information can be generalized and synthesized through clustering analysis in Geographic Information System. This paper presents a clustering method based on distance threshold neighbor. The method adds objects to a cluster by distance reachability. It also can find arbitrary shape cluster and has good separation effect to noise. In our experiment, the proposed algorithm is applied to GIS data mining and the results show that the algorithm has the satisfactory performance for GIS clustering.