可视化空间聚类挖掘算法及系统实现

空间聚类与传统聚类方法的区别之一在于空间聚类是对空间实体的集群性进行分析,在聚类过程中需考虑模式在空间分布上的一种或几种结构特征,如模式间的远近关系、拓扑关系、方位关系、疏密关系等。然而,传统聚类算法大多忽略空间结构特征对聚类结果的影响。同时,传统数据挖掘过程往往是“黑箱”作业,用户不论感兴趣与否都只能被动地接受挖掘结果,而且结果往往是抽象的、不易理解的。本文对基于MST的可视化空间数据聚类挖掘算法进行了研究,利用Delaunav三角网和MST最小生成树使得地理实体的邻接度与其他属性数据一起参与了空间聚类处理,同时用J2EE技术开发可视化空间聚类挖掘工具,为此类应用系统的建立提供了一种实用的可行方案。

The Study and Application on Visual Spatial Clustering

Traditional clustering approaches usually do not take into account the spatial structure, which refers to the distance between patterns, topology, density and other spatial distribution characteristics, and lack efficient level -of -detail strategy for visualization. Meanwhile the existing data mining methods tend to be "closed" and do not gear to allow the interaction needed to effectively support a human-led exploratory analysis. In this paper, we provide a new visual hierarchical clustering based on minimum spanning tree partitioning algorithm, which groups and visualizes cluster hierarchies consisting of both non-spatial and spatial attributes. Taking the spatial clustering as an example, an overall function for the J2EE-based visual spatial clustering system is designed in detail. At the same time, the key techniques are discussed. It provides a feasible solution to such kind of application system.