南京市城市轨道交通站点周边地区建成环境对居民活动的影响_基于梯度提升决策树与SHAP解释模型的分析  

Influence of Built Environment on Residents'Activities in Nanjing Urban Rail Transit Station Areas:An Analysis Based on Gradient Boosting Decision Tree and SHAP Interpretation Model

作　　者：周扬[1] 邵天元 钱才云[1] ZHOU Yang;SHAO Tian-yuan;QIAN Cai-yun(College of Architecture,Nanjing Tech University,Nanjing 211816,China;Arts Group Co.,Ltd.,Suzhou 215000,China)

机构地区：[1]南京工业大学建筑学院,南京211816 [2]中衡设计集团股份有限公司,苏州215000

出　　处：《科学技术与工程》2023年第17期7509-7519,共11页Science Technology and Engineering

基　　金：2021年度教育部人文社会科学研究青年基金(21YJCZH248);2022江苏省科技厅自然科学基金面上项目(BK20221316);2021年度江苏省高校哲学社会科学研究一般项目(2021SJA0223)。

摘　　要：城市轨道交通作为低能耗、少污染、具有可持续属性的公共交通类型之一,其对沿线城市发展、居民生产生活产生深远影响。中国城市轨道交通建设目前仍处于高速发展阶段,部分站点周边地区面临空间利用率不匹配、潮汐客流趋势加重等问题。城市轨道交通站点周边地区的城市空间规划需关注城市居民的活动特征,以提升站点地区城市空间全时段活力。以南京市中心城区内轨道交通站点周边地区为例,基于城市空间开放数据、实地踏勘调研、互联网移动定位服务(location based service,LBS)数据,采集统计与评价建成环境现状与居民活动特征数据,并运用梯度提升决策树与SHAP(Shapley addictive explanation)解释分析站点地区建成环境与居民活动的非线性关系及建成环境要素之间的交互作用,在此基础上提出建成环境要素适宜区间及协同优化条件,为城市轨道交通站点周边地区空间规划与优化提供建议。As one of the types of public transit with low energy consumption,less pollution,and sustainable attributes,urban rail transit has a profound impact on the development of cities along the route and residents’life.The construction of urban rail transit in China is still in the stage of high-speed development,and some station areas are facing problems such as mismatch of space utilization and aggravated tidal passenger flow trend.The urban space planning of rail transit station areas should pay more attention to the characteristics of residents’activities to enhance the full-time vitality of urban space in the station areas.Taking the area around the rail transit station in the downtown area of Nanjing as an example,based on the urban space open data,field survey and survey,and the internet mobile location based service(LBS)data,the current situation of the built environment and the characteristic data of residents'activities were collected,counted and evaluated.And nonlinear relationships between built environment and resident activities at stations and interactions between built environment factors have been analyzed using gradient ascension decision trees and Shapley addictive explanation(SHAP).Based on these explanations,appropriate ranges of built environment factors and collaborative optimization conditions have been proposed.The research provides suggestions for the spatial planning and optimization of the surrounding area of urban rail transit stations.

关 键 词：轨道交通站点周边地区 建成环境要素 居民活动 梯度提升决策树 SHAP 

分 类 号：TU984.191[建筑科学—城市规划与设计]