基于图论的EMST聚类算法的适用性研究

Study on The Applicability of EMST Clustering Algorithm based on Graph Theory

针对空间聚类算法EMST在实际应用中适用性不清,具有很大局限性的问题,提出了通过对比实验来确定EMST算法在空间数据库中的性能的方法。实验从簇的空间形态(类球形和任意形状)、簇的空间密度差异、簇的空间邻近和噪声点(孤立点)的影响来分析EMST算法的优缺点,同时分析不同参数对EMST算法聚类结果的影响。通过实验分析总结,提高了EMST算法的实际应用价值。

Aiming at the problem of unclear applicability and great limitation of spatial clustering algorithm EMST in practical application,a method is proposed to determine the performance of EMST algorithm in spatial database through comparative experiments.In this experiment,the advantages and disadvantages of EMST algorithm were analyzed from the aspects of spatial morphology(spherical and arbitrary shapes)of clusters,spatial density difference of clusters,spatial proximity of clusters and the influence of noise points(outludes),as well as the influence of different parameters on the clustering results of EMST algorithm.Through experimental analysis and summary,the practical application value of EMST algorithm is improved.