新冠疫情及建成环境对公交客流量的影响模型

为揭示新冠疫情背景下公交客流量变化的空间影响因素，以疫情前后公交站点层面客流变化量为因变量，以建成环境、病毒感染情况及病毒传播途径等指标为自变量，构建新冠疫情与建成环境对公交客流量共同影响的线性回归(Ordinary Least Squares, OLS)模型与梯度提升回归树(Gradient Boosting Regression Trees, GBRT)模型。以广州市为实证对象，基于公交IC卡数据、兴趣点数据(Point of Interest, POI)及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GBRT模型比OLS模型具有更好的拟合度；同时，常规公交站点的公交线路数量(22.02%)和到市中心距离(13.56%)是影响疫情背景下公交客流量变化的最重要因素，片区病毒感染与传播情况对疫情防控常态化时期的公交客流量作用有限，居民日常公交出行已经从疫情的影响下逐渐恢复。

Impact Model of COVID-19 and Built Environment on Bus Passenger Flow

In this study, the factors of COVID-19 and the built environment are used to examine variations in bus passenger flow. The study aims to reveal the influencing mechanism of bus passenger flow in the context of epidemic prevention and control, thereby providing strategic support for the quick recovery of bus passenger flow in the postepidemic period. This study focuses on Guangzhou City, and the data are collected from the bus IC card, point of interest (POI), and road network. The ordinary least squares (OLS) model and gradient boosting regression tree model (GBRT) are constructed to analyze the passenger flow of bus stops. The results show that the fitness of the GBRT model, which takes into account nonlinear effects, is superior to that of the OLS model. The key factors influencing changes in bus passenger flow during the epidemic period are the number of bus lines (which accounts for 22.02%) and the distance to the city center (which accounts for 13.56%). The findings indicate that the impact of COVID-19 on bus passenger flow is not crucial. With the normalization of epidemic prevention and control, people's demand for bus travel is recovering.