地球信息科学学报

地球信息科学学报 ›› 2019, Vol. 21 ›› Issue (3): 337-345.doi: 10.12082/dqxxkx.2019.180380

• 地球信息科学理论与方法 • 上一篇    下一篇

顾及空间异质性的出租载客与公交客流回归分析

邬群勇1,2,*(), 张良盼1,2, 吴祖飞1,2   

  1. 1. 福州大学 卫星空间信息技术国家地方联合工程研究中心,福州 350000
    2. 空间数据挖掘与信息共享教育部重点实验室,福州 350000
  • 收稿日期:2018-08-16 修回日期:2018-12-19 出版日期:2019-03-15 发布日期:2019-03-15
  • 通讯作者: 邬群勇 E-mail:qywu@fzu.edu.cn
  • 作者简介:

    作者简介：邬群勇（1973-）,男,山东诸城人,博士,研究员,研究方向为时空大数据分析、地理信息服务。E-mail：qywu@fzu.edu.cn

  • 基金资助:
    国家自然科学基金项目（41471333）;中央引导地方科技发展专项项目（2017L3012）

Regression Analysis of Taxi Pick-up and Bus Passenger Flow Considering the Spatial Heterogeneity

Qunyong WU1,2,*(), Liangpan ZHANG1,2, Zufei WU1,2   

  1. 1. National & Local Joint Engineering Research Center of satellite-spatial Information Technology, Fuzhou University, Fuzhou 350000, China
    2. Key Laboratory of Spatial Data Mining & Information Sharing of MOE, Fuzhou 350000, China
  • Received:2018-08-16 Revised:2018-12-19 Online:2019-03-15 Published:2019-03-15
  • Contact: Qunyong WU E-mail:qywu@fzu.edu.cn
  • Supported by:
    National Natural Science Foundation of China, No.41471333;The Central Guided Local Development of Science and Technology Project, No.2017L3012.

RichHTML

PDF (PC)

赞

被引次数

2

摘要：

出租车一直以来被看作公共交通的补充,但是以往研究多侧重于出租客流与公交客流的独立研究,对于二者的关联关系分析没有足够得到关注。预测出租车载客热点区域不仅能够实时的了解城市交通热点区域,还能够很好地指引出租车司机,帮助出租车司机快速寻客。出租车载客热点常发生在人流密集并且交通出行需求较高的区域,公交乘客IC卡数据能够实时的反映城市中的交通需求。因此,本文使用厦门岛出租车GPS轨迹数据与公共交通运输系统运营数据,利用核密度估计法和地理加权回归模型分析了早晚高峰时段出租车载客与公交上下车（OD）客流之间的时空分布关系。研究发现,出租乘客O点的核密度值在空间上存在分布不均衡性,聚集特征明显。在同一区域,公交乘客O点和公交乘客D点对出租乘客O点所产生的影响刚好相反;在不同区域,城市功能类型复杂的地区公交乘客O点对出租乘客O点产生负的影响,在城市功能类型单一的地区公交乘客O点对出租乘客O点产生正的影响,公交乘客D点则刚好相反。与普通线性回归模型相比,地理加权回归模型的拟合效果显著提高,早晚高峰拟合优度分别从0.13和0.11提升到了0.59和0.53。研究结果可为出租车载客数量的预测提供相关依据。

关键词: 乘客OD, 空间异质性, 地理加权回归, 核密度估计, 城市交通

Abstract:

Taxi traffic has always been regarded as a supplement to public transportation. However, this may be in part due to previous studies focusing on independent research of taxi and bus passenger flow. Research around the relationship between taxi and bus passenger flow has not yet been thoroughly investigated. Taxi passenger hotspots not only provide real-time understanding of urban traffic hotspots, but also guide taxi drivers and enable taxi companies to make effective dispatches. Taxi passenger hotspots tend to occur in areas where demand for transportation is high and in areas of intensive crowding. Bus passengers' IC card data can reflect real-time traffic demand within the city. This study used Xiamen Island taxi GPS trajectory data and public transportation system data, along with the kernel density estimation method and geographic weighted regression (GWR) model to analyze the OD (Origin-Destination) passenger flow in both morning and evening peak travel times. Results showed a significant spatial heterogeneity in the kernel density value of the taxi passenger O. However, within the same area, the impact of bus passenger O and bus passenger D on the taxi passenger O was found to be opposite; in various regions, the negative impact of bus passenger O on the taxi passenger O in areas with complex urban functional types, bus passenger O had a positive impact on the taxi O at a single function area, while the bus passenger D was just the opposite. Compared to the ordinary least squares (OLS) model, GWR provided a much better fit (with the goodness of fit increasing from 0.13 and 0.11 to 0.59 and 0.53 in the morning and evening peak hours, respectively). Results of this study could provide the basis to forecast the number of taxi passengers.

Key words: passenger OD, spatial heterogeneity, geographically weighted regression, kernel density estimation, urban traffic