一种改进的DBSCAN聚类算法的研究与应用  

An Improved DBSCAN Clustering Algorithm

作　　者：刘军[1] 艾力.斯木吐拉[1] 马晓松[1] 

机构地区：[1]新疆农业大学,乌鲁木齐830052

出　　处：《交通与计算机》2008年第3期60-64,共5页Computer and Communications

基　　金：国家自然科学基金项目(批准号:50422286)资助

摘　　要：对DBSCAN(density based spatial clastering of applications with noise)聚类算法进行了研究,分析了它的特点、存在的问题和改进思想,提出了基于DBSCAN方法的交通事故多发点(段)的排查方法及其改进思路,并且结合J2EE技术,实现了一个Web环境下的交通事故数据挖掘系统,阐述了基于J2EE的交通事故挖掘系统的体系结构、设计及其应用。At first, a DBSCAN（density-based spatial clastering of applications with noise） clustering algorithm was discussed. Its characteristics, problems, and idea to improve it were analyzed. The identifying method of traffic accidents multiple-points based on the DBSCAN clustering algorithm and the idea to improve it were proposed. With the combination of J2EE technology, a web environment data mining system of traffic accidents was achieved. The architecture, design, and application of accidents data mining system based on J2EE were illustrated.

关 键 词：数据挖掘 聚类分析 DBSCAN算法 交通事故黑点(段) J2EE 

分 类 号：U491[交通运输工程—交通运输规划与管理] TP391.4[交通运输工程—道路与铁道工程]