一种基于k-means聚类的航运信息孤立点分析算法  

Outlier detection algorithm for maritime information based on k-means clustering

作　　者：王天真[1] 刘萍[1] 汤天浩[1] 黄洪琼[1] 张艳[1] 

机构地区：[1]上海海事大学物流工程学院,上海201306

出　　处：《上海海事大学学报》2011年第3期54-57,共4页Journal of Shanghai Maritime University

基　　金：交通运输部应用基础研究项目(2009-329-810-030);上海市教育委员会重点学科建设项目(J50602);上海市教育委员会科研创新项目(11YZ140);上海海事大学校基金(20100136)

摘　　要：为了从大量含噪的航运信息中挖掘出有用的知识,研究孤立点探测方法在航运信息处理中的应用.首先在对k-means聚类算法分析的基础上,定义中位数的概念及孤立点的判断标准,提出一种基于k-means聚类的孤立点分析算法;然后,针对航运信息的特点,采用一维数据和多维数据异常点探测方法挖掘AIS数据库中的异常信息.实验结果证明,基于k-means聚类的算法对孤立点更敏感,能够探测到一维统计方法中探测不到的异常点.In order to find out useful knowledge from large quantities of maritime database with noise,the application of outlier detection method in maritime information processing is researched.Firstly,on the basis of analyzing k-means clustering algorithm,the concept of the median and the standard for the judgment of outlier are proposed,and an outlier analysis algorithm based on k-means is presented;then,according to the characteristics of the maritime information,the method for detecting abnormal points of one-dimensional and multi-dimensional data is used to find out the abnormal information from the AIS database.The experimental results show that the algorithm based on k-means clustering is more sensitive to outlier,and the algorithm can detect some heterogeneous points which are difficult to find by using one-dimensional statistical methods.

关 键 词：数据挖掘 K-MEANS 孤立点 航运信息 

分 类 号：U623.9[交通运输工程—船舶及航道工程] TP13[交通运输工程—船舶与海洋工程]