地球信息科学学报 ›› 2021, Vol. 23 ›› Issue (8): 1461-1472.doi: 10.12082/dqxxkx.2021.200602

• 地理空间分析综合应用 • 上一篇    下一篇

基于出行时空数据的分时租赁汽车与网约车出行场景比较研究

许研1,*(), 纪雪洪1, 叶玫2   

  1. 1.北方工业大学经济管理学院,北京 100144
    2.广东科学技术职业学院计算机工程技术学院,广州 510640
  • 收稿日期:2020-10-14 修回日期:2021-03-23 出版日期:2021-08-25 发布日期:2021-10-25
  • 通讯作者: 许研
  • 作者简介:许 研(1983— ),女,辽宁锦州人,副教授,主要从事共享出行行为复杂性研究。E-mail: bnuxuyan@126.com
  • 基金资助:
    北京社会科学基金项目(18GLC080)

Travel Scenes Comparison of Time-Sharing and Car-Hailing based on Traveling Spatiotemporal Data

XU Yan1,*(), JI Xuehong1, YE Mei2   

  1. 1. Economic and Management School, North China University of Technology, Beijing 100144, China
    2. Computer Engineering Technical College, Guangdong Polytechnic of Science and Technology, Guangzhou 510640, China
  • Received:2020-10-14 Revised:2021-03-23 Online:2021-08-25 Published:2021-10-25
  • Contact: XU Yan
  • Supported by:
    Beijing Social Science Fund(18GLC080)

摘要:

分时租赁和网约车同属共享汽车,但规模对比悬殊。找到差异化的出行场景有利于分时租赁在网约车主导的共享汽车市场中谋求立足之地。本文以北京地区某分时租赁公司2017年5月1日—30日的出行订单和2018年4月23日—29日的网约车出行订单为研究对象,结合城市兴趣点数据,利用地理信息层次聚类、关联规则等方法挖掘两共享汽车的典型出行场景,并进行比较分析。研究表明:① 网约车主要服务于“通勤出行”和“市内商务区之间的出行”,2种出行场景分别占网约车订单总量的40.3%和28.7%;② 分时租赁主要服务非通勤出行,其特色出行场景是“往返城市旅游景区的出行”、“城市旅游景区之间的出行”和“外城住宅商务混合区的午夜出行”,分别占分时租赁订单总量的24.4%、6.9%和5.5%。③ 在分时租赁的特色出行场景中,分时租赁与网约车或传统租车等共享出行方式相比费用更低,仅占其费用的25%~35%,具有较大的竞争优势。本研究有关出行场景挖掘的方法和结论可以为北京市分时租赁的推广以及其他共享出行研究提供借鉴和参考。

关键词: 分时租赁, 网约车, 共享出行, 出行场景挖掘, 出行时空特征分析, 城市功能类型识别, 聚类分析, 关联规则

Abstract:

Both time-sharing rental cars and car-hailing are car sharing service, but with different scales. Finding differentiated travel scenes is conducive to time-sharing in seeking a foothold in the car sharing market dominated by car-hailing. This research uses car sharing records data combined with city's points of interest (POI) data to analyze the spatial temporal characteristics, and typical travel scenes of time-sharing and car-hailing in Beijing. Firstly, hierarchical clustering method was used to define the most distinguished clusters of city's grid cells based on POI data. Dunn index examined the optimal cluster number. Secondly, Origin and Destination (OD) locations of each car-sharing trip were labelled by the cluster types. The users' preferences were observed from OD cluster pairs appearing more frequently than others in the records. Thirdly, typical travel scenes were extracted by analyzing association rules of these cluster pairs. In the end, spatiotemporal patterns of typical travel scenes were tested. The findings of this study can be divided into three portions. Firstly, car-hailing mainly serves commuters and travels among business districts within the city. Secondly, time-sharing mainly serves non-commuter travels, and the representative travel scenes are short-distance city travel for tourism and Midnight travel in suburban areas. Many of these observed relationships are interpretable. For example, a short-distance city travel for tourism usually lasts half a day. Renting behaviors avoid the morning and evening rush hours of commuting. A midnight travel in suburban areas happens outside the city center, which usually lasts less than an hour since public transport is not available during that time. Thus, this travel scene seems to be related with urgent travel demands. Thirdly, the cost of time-sharing is far lower than that of other alternative modes and constitutes only 30%~50% cost of others, indicating that car sharing is beneficial when compared with other modes in these scenarios obtained in our research. These findings can serve as references and suggestions for time-sharing's promotion and operation process. The travel scenes mining method proposed in this study can be repeated in other car sharing researches.

Key words: time-sharing rental cars, car-hailing, car sharing service, travel scenes mining, travels spatial temporal characteristics analysis, city functional cluster types, hierarchical clustering analysis, association rules