一种调整簇阀值的加速聚类分析算法及其应用  

Accelerated Clustering Analysis Algorithm Based on Cluster Threshold-Adjusted and Its Application

作　　者：高迪驹[1] 王天真[1] 刘远[1] 刘洋[1] 

机构地区：[1]上海海事大学物流工程学院,上海200135

出　　处：《数据采集与处理》2012年第3期287-293,共7页Journal of Data Acquisition and Processing

基　　金：国家交通部科研基金(2009-329-810-030)资助项目;国家自然科学基金(60804064)资助项目;上海市教委科研创新基金(11YZ139)资助项目;上海海事大学科研基金(20120103)资助项目

摘　　要：针对航运信息中大样本聚类问题,根据k-means聚类过程中大部分簇的调整发生在初始迭代阶段的特性,提出了一种调整簇阀值的加速聚类方法,并对该算法进行实例测试,实验结果证明了该方法在满足原有的聚类精度的基础上,减少了聚类的计算量。本文将该方法应用到船舶航线设计中。Considering most adjustments of clusters happen in the initial stage of iteration in the clustering process of k-means algorithm, an accelerated clustering analysis algorithm is proposed to perform the clustering for large amounts of samples based on cluster threshold adjusted. Experiments prove that the improved algorithm can obtain the good clustering result, and accelerate the speed of clustering. Finally, an application of clustering is given for designing routes from the marine database.

关 键 词：数据挖掘 聚类分析 簇阀值 计算复杂度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]