交通运输系统工程与信息

交通运输系统工程与信息 ›› 2021, Vol. 21 ›› Issue (4): 197-203.DOI: 10.16097/j.cnki.1009-6744.2021.04.024

所属专题： 2021年英文专栏

• 系统工程理论与方法 • 上一篇    下一篇

建成环境对城市停车需求影响的非线性模型

陈坚*1，刘柯良1，邸晶2，彭涛1   

  1. 1. 重庆交通大学，交通运输学院，重庆 400074；2. 保定市城市设计院，河北 保定 071000
  • 收稿日期:2021-05-27 修回日期:2021-06-13 接受日期:2021-07-05 出版日期:2021-08-25 发布日期:2021-08-23
  • 作者简介:陈坚(1985- )，男，江西赣州人，教授，博士。
  • 基金资助:
    重庆市社会科学规划重点项目

Nonlinear Model of Impact of Built Environment on Urban Parking Demand

CHEN Jian*1 , LIU Ke-liang1 , DI Jing2 , PENG Tao1   

  1. 1. School of Traffic & Transportation, Chongqing Jiaotong University, Chongqing 400074, China; 2. Urban Design Academe of Baoding, Baoding 071000, Hebei, China
  • Received:2021-05-27 Revised:2021-06-13 Accepted:2021-07-05 Online:2021-08-25 Published:2021-08-23
  • Supported by:
    The Key Project of Social Science Foundation of Chongqing, China(2020ZCZX04)。

PDF

PDF(English version)

133

摘要： 为精细化把握城市建设项目在微观空间尺度下的停车需求规律，从空间视角探究停车需 求与建成环境之间的关系。通过高峰小时建筑物单位面积的停车生成数表征停车需求，以土地 利用混合度、路网密度、公交服务水平等9个因子描述建成环境，分别构建建成环境对停车需求影 响的普通最小二乘(Ordinary Least Squares，OLS)模型与梯度提升迭代决策树(Gradient Boosting Decision Tree，GBDT)模型。以保定市主城区停车调查数据中的商业类配建停车场为对象，基于 停车调查数据、兴趣点数据(Point of Interst，POI)、道路网络数据等多源异构数据进行模型实证分 析。结果表明，考虑非线性效应的GBDT模型比OLS模型具有更好的拟合度。从影响贡献度来 看，配建指标(18.92%)与区位(15.23%)是影响停车需求的最重要建成环境因素，交叉口密度 (5.19%)贡献度最小；在非线性关系方面，建成环境因子与停车需求均具有非线性关系与阈值效 应，除交叉口密度及人口密度与停车需求呈现U型关系，其余因素与停车需求的关系整体上保持 正相关或负相关。

关键词: 城市交通, 建成环境, 停车需求, 梯度提升迭代决策树, 非线性关系

Abstract: This paper investigates the relationship between parking demand and the built environment from a spatial perspective to understand the parking demand trend from the micro- spatial scale of urban construction projects. The parking demand is represented by the number of parking generations per unit area of the building during peak hours, and the built environment is characterized by 9 factors including the degree of mixed land use, road density, and service level of urban public transport. The Ordinary least- squares (OLS) model and the gradient boosting decision tree (GBDT) model are developed to describe the impact of built environment on parking demand. Based on the parking data of commercial parking lots in the main urban area in Baoding, China, this study conducted an empirical analysis of the model with the multi-source heterogeneous data including parking survey data, the Point of Interest (POI) data and road network data. The results show that the GBDT model considering the non-linear effect has a better fitting degree than the OLS model. From the perspective of impact contribution, construction indicators and location are two built environmental factors significantly affect parking demand, the contributions are respectively18.92% and 15.23%. The intersection density has the least contribution, which is 5.19%. In terms of non-linear relationship, both built environmental factors and parking demand have non- linear relationship and threshold effect. In addition to the Ushaped relationship of intersection density and population density with parking demand, the relationship of other factors and parking demand overall remains positive or negative correlations.

Key words: urban traffic, built environment, parking demand, gradient boosting decision tree, non-linear relationship

中图分类号: