基于改进谱聚类的热点区域挖掘方法  

Hot Region Mining Approach Based on Improved Spectral Clustering

作　　者：梁卓灵 元昌安 覃晓 乔少杰[3] 韩楠 范勇强 LIANG Zhuoling;YUAN Changan;QIN Xiao;QIAO Shaojie;HAN Nan;FAN Yongqiang(Guangxi University,Nanning 530004,China;Nanning Normal University,Nanning 530001,China;Chengdu University of Information Technology,Chengdu 610225,China;Chengdu Environmental Protection Information Center,Chengdu 610015,China)

机构地区：[1]广西大学,南宁530004 [2]南宁师范大学,南宁530001 [3]成都信息工程大学,成都610225 [4]成都市环境保护信息中心,成都610015

出　　处：《重庆理工大学学报（自然科学）》2021年第1期129-137,共9页Journal of Chongqing University of Technology：Natural Science

基　　金：国家自然科学基金项目(61962006,61802035,61772091);广西科技开发项目(AA18118047,AD18126015);广西自然科学基金项目(2018GXNSFDA138005);四川省科技计划项目(2018JY0448,2019YFG0106,2019YFS0067);广西高校中青年教师能力提升项目(ky2016yb276)。

摘　　要：移动轨迹数据的热点区域挖掘在城市交通管理、道路规划和基于位置的服务中具有重要的作用。传统数据挖掘方法K-means、DBSCAN等算法,其参数选择困难、易影响聚类效果,针对在非凸数据集或密度不均匀、聚类间距差相差很大的数据集上聚类表现较差等问题,提出了基于改进谱聚类的热点区域挖掘算法(hot region mining algorithm based on improved spectral clustering,ISCRM)。实验结果表明:对比传统方法,ISCRM算法优势在于自适应选取参数,避免人工调试参数环节,且其适用于任意形状的样本空间,聚类质量更高。可准确获得各个聚类中心,从而识别出用户出行热点区域。Hot region mining of mobile trajectory data plays an important role in urban traffic management,road planning and location-based services.Traditional data mining methods,such as Kmeans and DBSCAN,are difficult to select their parameters and easy to affect the clustering effect.In order to solve the problem of poor clustering performance on non-convex data sets or data sets with uneven density and large difference in clustering spacing,a hot region mining algorithm based on improved spectral clustering,ISCRM is proposed.The experimental results show that,compared with the traditional method,the ISCRM algorithm has the advantage of adaptively selecting parameters,avoiding the step of manually debugging the parameters.It is suitable for any shape of sample space,and the clustering quality is higher.It can accurately calculate the center of each cluster,in order to identify the user’s travel hot region.

关 键 词：轨迹数据 热点区域 谱聚类 停留点 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]