基于密度分区的出租车载客热点区域聚类分析  

CLUSTER ANALYSIS OF HOT SPOTS FOR TAXI PASSENGERS BASED ON DENSITY ZONING

作　　者：任丹萍[1,2] 刘琳 陈湘国 Ren Danping;Liu Lin;Chen Xiangguo(School of Information and Electrical Engineering,Hebei University of Engineering,Handan 056038,Hebei,China;Hebei Key Laboratory of Security&Protection Information Sensing and Processing,Handan 056038,Hebei,China)

机构地区：[1]河北工程大学信息与电气工程学院,河北邯郸056038 [2]河北省安防信息感知与处理重点实验室,河北邯郸056038

出　　处：《计算机应用与软件》2023年第10期83-89,共7页Computer Applications and Software

基　　金：河北省自然科学基金项目(F2018402198)。

摘　　要：基于城市居民出行的随机性和出租车行驶的机动性,对出租车轨迹数据进行载客热点区域的挖掘,得到城市居民出行规律。由于出租车轨迹数据密度分布不均匀,应用一般的聚类方法效果不佳,因此提出一种基于密度分区的聚类算法。该算法通过求取每个出租车上车点位置数据的局部密度,得到密度峰值点作为簇中心,实现对轨迹数据集基于密度的快速划分,得到不同密度的轨迹数据集,在此基础上进行二次聚类。实验结果表明,该算法可以有效识别不同密度的出租车载客热点区域,提高聚类结果的精确度。Based on the randomness of urban residents' travel and the mobility of taxi driving,the hot spots of passenger carrying were excavated from the taxi trajectory data,and the travel rules of urban residents were obtained.Due to the uneven density distribution of taxi trajectory data,the general clustering method had a low clustering accuracy,so a clustering algorithm based on density zoning was proposed.By calculating the local density of the taxi pickup point location data,the algorithm obtained the peak density point as the cluster center,and realized the fast partition of the trajectory data set based on the density,and obtained the trajectory data set with different densities.On this basis,secondary clustering was carried out.The experimental result show that the clustering algorithm based on density partition can effectively identify the hot spots of taxi passengers with different densities and improve the accuracy of clustering result.

关 键 词：出租车GPS轨迹数据 时空特征分析 密度聚类 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]