交通运输系统工程与信息

交通运输系统工程与信息 ›› 2023, Vol. 23 ›› Issue (5): 136-145.DOI: 10.16097/j.cnki.1009-6744.2023.05.015

• 系统工程理论与方法 • 上一篇    下一篇

建成环境和出租车需求对网约车出行需求影响的时空间分异模式

马健霄，赵飞燕，尹超英*，汤文蕴   

  1. 南京林业大学，汽车与交通工程学院，南京 210037
  • 收稿日期:2023-05-01 修回日期:2023-06-25 接受日期:2023-06-28 出版日期:2023-10-25 发布日期:2023-10-22
  • 作者简介:马健霄(1966- )，男，内蒙古赤峰人，教授，博士。
  • 基金资助:
    国家自然科学基金(72204114)；教育部人文社科项目 (22YJC630191)；江苏省研究生科研与实践创新计划项目 (SJCX23_0341)。

Spatial-temporal Heterogeneity Effects of Built Environment and Taxi Demand on Ride-hailing Demand

MA Jian-xiao,ZHAO Fei-yan,YIN Chao-ying*,TANG Wen-yun   

  1. College of Automobile and Traffic Engineering, Nanjing Forestry University, Nanjing 210037, China
  • Received:2023-05-01 Revised:2023-06-25 Accepted:2023-06-28 Online:2023-10-25 Published:2023-10-22
  • Supported by:
    National Natural Science Foundation of China (72204114)；Humanities and Social Sciences Fund of Ministry of Education of China (22YJC630191)； Postgraduate Research & Practice Innovation Program of Jiangsu Province (SJCX23_0341)。

PDF

摘要： 为探究考虑出租车出行需求影响下，建成环境与网约车出行需求之间的互动关系，本文基于南京市网约车和出租车订单数据，围绕密度、设计、多样性及公交临近度这4个维度构建城市建成环境指标，分别针对早高峰、晚高峰和平峰这3个时段建立考虑局部变化项和全局固定项的半参数地理加权回归模型(SGWR)，探究建成环境对网约车出行需求的时空异质性影响。结果表明：对比普通最小二乘回归(OLS)和传统地理加权回归模型(GWR)，SGWR模型的AICc值在3个时段分别下降了 2.44%与 0.15%，4.01%与 0.30%，1.89%与 0.27%；Adjusted R2 值分别提高了6.52%与0.11%，8.02%与0.55%，2.75%与0.11%，表明SGWR模型具有更好的拟合效果。局部变量的回归结果表明，不同的建成环境变量对网约车出行需求影响不同，具有时空异质性；全局变量的回归结果表明，土地利用混合度在早晚高峰时段对网约车出行需求影响显著，为负向影响。出租车和网约车之间呈现为合作关系，高密度的公司企业和公交站点数量会促进网约车出行需求的产生。本文研究结果可为合理配置网约车资源提供理论依据。

关键词: 城市交通, 时空异质性, 半参数地理加权回归, 网约车出行需求, 建成环境

Abstract: To study the interaction between the built environment and the travel demand of ride-hailing under the influence of taxi travel, this paper proposes urban built environment indicators around the four dimensions of density, design, diversity and distance to transit based on the data of ride- hailing and taxi orders in Nanjing city of China. A semi-parametric geographically weighted regression model (SGWR) considering local changes and global fixed terms is developed for the three periods of morning peak, evening peak and off peak to describe the spatial-temporal heterogeneity of the built environment on the travel demand of ride-hailing. The results show that: compared with ordinary least squares regression (OLS) and traditional geographically weighted regression (GWR), the AICc values of SGWR model decreased by 2.44% and 0.15% during morning peak, decreased by 4.01% and 0.30% during evening peak, and decreased by 1.89% and 0.27% in the off peak. Adjustment R2 increased by 6.52% and 0.11% during the morning peak, 8.02% and 0.55% in the evening peak, and 2.75% and 0.11% in the off peak, indicating that the SGWR model has better explanatory power and goodness of fit. The regression results of local variables show that different built environment variables have different effects on the travel demand of ride-hailing, with spatial-temporal heterogeneity. The regression results of the global variables show that the land use mix has a significant negative impact on the demand for ride-hailing in the morning and evening peak hours. There is a cooperative relationship between taxis and ride-hailing, and the number of high-density companies and bus stops will promote the demand for ride-hailing. This study can provide a theoretical basis for the rational allocation of ride-hailing resources.

Key words: urban traffic, spatial-temporal heterogeneity, semi-parametric geographically weighted regression, ride-hailing demand, built environment

中图分类号: