摘要
为解析建成环境对共享单车不同时段借还车量的非线性作用,本文以2016年上海摩拜共享单车和网络公开数据为基础,利用梯度提升决策树,建立共享单车工作日,非工作日及工作日早、晚高峰借还车量模型。研究结果显示:变量影响程度方面,街道至市中心距离对4个时段共享单车借还车量具有稳定且突出的作用,相对重要度超过17%;路网密度、非机动车道占比、人口密度影响次之,但在4个时段作用波动较大。非线性关系方面,街道至市中心距离、非机动车道占比、人口密度、就业POI(Point of Interest)密度与各时段共享单车借还车量呈复杂非线性关系,阈值效应显著;路网密度、住宅POI密度则与之总体负相关和正相关。同时段下,建成环境对早晚高峰时期的共享单车借还车量非线性作用差异明显,与高峰时期单车骑行的潮汐性特征相一致,非机动占比与街道中心邻近度、就业POI密度对高峰时段单车骑行作用有明显的协作性作用。
To investigate the nonlinear effect of built environment on bike-sharing ridership at different time periods,this study utilized 2016 data from Mobike in Shanghai,along with online public data.Using Gradient Boosting Decision Trees,prediction models for bike-sharing ridership during weekdays,weekends,and morning-evening peak hours were developed.The findings revealed that,regarding the importance of built environment,proximity to the city center had a consistent and significant influence on borrowed and returned bikes across all four time periods,with a relative importance of over 17%.Following that,road density,cycle-way ratio,and population density had substantial but varying influences over the four time periods.In terms of nonlinear effects,proximity to the city center,cycle-way ratio,population density,and job POI(Point of Interest)density all exhibited complicated nonlinear relationships with bike-sharing ridership and notable threshold effects.Meanwhile,bike usage is negatively related to road density and positively related to residence POI density.All built environment variables had varying nonlinear effects on bike borrowing and returning during morning and evening peak hours,consistent with the tidal features of bike riding.The cycle-way ratio along with the distance to CBD,and job POI density,have significant synergistic effects on peak-hour bike-sharing ridership.
作者
吴静娴
唐桂孔
李文翔
WU Jingxian;TANG Guikong;LI Wenxiang(Business School,University of Shanghai for Science and Technology,Shanghai 200093,China;Intelligent Emergency Management School,University of Shanghai for Science and Technology,Shanghai 200093,China)
出处
《交通运输系统工程与信息》
EI
CSCD
北大核心
2024年第1期290-298,310,共10页
Journal of Transportation Systems Engineering and Information Technology
基金
国家自然科学基金(52122215)
上海市曙光计划(22SG45)
上海市哲学社会科学规划课题(2022ZGL008)。
关键词
城市交通
建成环境
非线性作用
梯度提升树
共享单车
urban traffic
built environment
nonlinear effect
Gradient Boosting Decision Tree
bike-sharing service
