摘要
为掌握乘客的精细化出行需求,进而提升公共交通系统的服务品质,以北京市为例,利用智能卡数据对乘客的出行特征进行分析。首先通过数据预处理获取乘客的完整出行链,然后从出行强度、出行时间和出行空间三个维度提取乘客出行特征指标,并利用主成分分析法(Principal Component Analysis, PCA)对其进行降维处理,最后基于降维后的特征利用不同的聚类算法对乘客进行分类。研究结果显示:K-means++算法的聚类效果最佳,聚类结果包含5类具有不同出行特征的乘客群体;其中类型一乘客的出行强度较大且出行时空特征稳定性较高,具有较明显的通勤出行特征,该类型乘客人数仅占总乘客数的18.4%,但其出行量占比超过55%,由于其公共交通依赖度较高,高峰期间应作为重点保障对象;类型二主要为生活类出行乘客,其出行时空稳定性较低,应深入挖掘此类乘客的个性化生活出行需求;类型三~类型五主要为低频或偶然出行乘客。根据乘客多天的出行链,进一步挖掘类型一乘客的居住地和工作地,结果显示乘客居住地主要分布在回龙观、天通苑以及黄村等区域,工作地主要分布在国贸、中关村和望京等区域,与北京市现状相符。
In order to understand the refined travel needs of passengers and improve the service quality of the public transport system,this paper analyzed the travel characteristics of passengers using smart card data in Beijing.Firstly,the complete trip chain of passengers was obtained after data preprocessing.Secondly,passenger travel feature indicators were extracted from three dimensions:travel intensity,travel time,and travel space,and Principal Component Analysis(PCA)was used to reduce the dimensionality of the travel feature indicators.Finally,different clustering algorithms were used to classify passengers based on the features after dimensionality reduction.The results showed that the K-Means++algorithm achieved the best performance,and there were 5 types of passenger clusters with different travel characteristics.Type 1 passengers showed greater travel intensity and higher stability of travel spatiotemporal characteristics,which indicated significant commuting attribute.Although the number of type 1 passengers only accounted for 18.4%,the number of their trips accounted for more than 55%.The public transport dependency degree of passengers with this type was high,whose travel needs should be guaranteed during peak hours.Type 2 mainly referred to lifestyle travel passengers,which had low spatio-temporal stability.Therefore,it was necessary to deeply explore the personalized lifestyle travel needs of passengers with this type.Type 3 to 5 passengers mainly travelled at low frequency or by chance.Based on the multi-day trip chains,the residence and working locations of type 1 passengers were further explored.The residence locations were mainly distributed in Huilongguan,Tiantongyuan,and Huangcun,and the working locations were mainly distributed in Guomao,Zhongguancun,and Wangjing,which is consistent with the real situation in Beijing.
作者
张开婷
王子帆
陈艳艳
蔺陆洲
王欣
ZHANG Kai-ting;WANG Zi-fan;CHEN Yan-yan;NIN Lu-zhou;WANG Xin(Quantutong Location Network Co.,Ltd.,Beijing 100176,China;Beijing Key Laboratory of Traffic Engineering,Beijing University of Technology,Beijing 100124,China)
出处
《交通运输研究》
2023年第2期72-81,共10页
Transport Research
基金
国家重点研发计划项目(2020YFB1600703)。
关键词
城市公共交通
智能卡数据
聚类算法
出行特征
通勤出行
urban public transport
smart card data
clustering algorithm
travel characteristics
commuter travel
