社区建成环境对小汽车使用行为的非线性影响模型

Non-linear impact model of community built environment on car usage behavior

为指导社区生活圈形成绿色出行建成环境,从社区建成环境层面定量解析小汽车使用行为的差异,通过停车场泊位利用率客观反映小汽车使用行为,构建了考虑非线性效应的梯度提升决策树(Gradient Boosting Decision Tree,GBDT)模型,以重庆市主城区停车调查数据中的居住区配建停车场为实证对象,构建以居住区为核心的社区生活圈作为建成环境的测度范围,实证结果表明:GBDT模型比线性假设的OLS模型具有更好的拟合度;建成环境指标贡献度方面,距离商圈的距离(14.30%)、人口密度(14.20%)、土地利用混合度(12.60%)对泊位利用率影响最大,反映其对小汽车使用行为的重要影响效应;在建成环境指标的非线性关系方面,建成环境因子与泊位利用率均具有非线性关系.

In order to guide the green travel and green built environment in the community living circle,this paper quantitatively analyzes the differences in car usage behavior in the community built environ⁃ment.Then the car usage behavior is objectively reflected by the parking space utilization rate,and the Gradient Boosting Decision Tree(GBDT)model is built that takes into account the nonlinear effect.The GBDT model takes the equipped parking lot data within the residential area of Chongqing’s main urban area in the parking survey as the empirical object,and builds a community living circle with the residential area as the core,which acts as the measurement range of the built environment.The empiri⁃cal results show that the GBDT model has a higher fit than the linear assumption OLS model;regard⁃ing the contribution of built environment indicators,the distance to the shopping area(14.30%),popu⁃lation density(14.20%),and land use mixture(12.60%)have significant effect on the parking space utilization rate,reflecting its great impact on the car usage behavior.Regarding the non-linear relation⁃ship between built environmental indicators,all built environmental factors have non-linear relation⁃ships with the parking space utilization rate.