Heterogeneity Analysis of Metro Station Attributes Impact on Bike-sharing Trips
Based on the 2018 Shanghai bike-sharing order data,this paper uses the multi-scale geographically weighted regression(MGWR)model to analyze the influence mechanism between the metro built environment and the bike-sharing trips.It investigates the heterogeneous characteristics of the influence effects from two major attributes of the metro station type and spatial location.The results show that there are four different types of impacts on the attraction and generation of bike-sharing trips,including,all positive,all negative,double positive or negative,and positive and negative opposite.The main factors influencing bike-sharing trips vary significantly between different metro stations,and this variation becomes even more pronounced when considering trip generation.The distance between metro stations and the city center has a significant impact on the ranking of the key factors while there is marginal difference between transfer stations and non-transfer stations.The scenarios of"Metro Station Density>Population Density>Bus Station Density"and"Residential Land>Metro Station Density"are the most influential combinations of trip attraction and generation within 10 km of the city center.The scenarios of"Population Density>Bus Station Density>Residential Land"and"Land-Use Information Entropy>Population Density"correspond to the most important combinations of trip attraction and generation outside 10 km range from the city center.As the distance of metro stations from the city center increases,the influence coefficients of each factor show different gradual change patterns,such as"＜"type,"几"type and"S"type.This indicates that the impact of the built environment on bike-sharing change with the spatial location and there is a complex non-linear change,it is appropriate to adopt a place-specific strategy for bike-sharing around metro stations.
urban trafficheterogeneitymulti-scale geographically weighted regression modelbike-sharingmetro station
万方数据
维普
