Estimating the Impacts of Seasonal Variations of Streetscape on Dockless Bike Sharing Trip with Street View Images and Computer Vision

Abstract

AbstractA significant portion of the cycling experience is influenced by the streetscape, and this impact varies throughout the year. The temporal dynamic of streetscape has been neglected in most previous studies, including urban public mobility route choices. This paper examines the correlation between dockless bike sharing and streetscape as well as spatial elements in different seasons using a large amount of GPS bike trajectory data collected by LIME. The study shows that: (1) DBS volume is significantly influenced by seasonal streetscape factors such as roads, cars, sidewalks, tree, and vegetation color; (2) How significantly these seasonal factors affect DBS volume differs in summer and autumn; (3) In both summer and autumn models, non-seasonal factors like mixed land use score, street network connectivity, etc., are significant. Some non-seasonal factors only impact the DBS volume in one season; (4) When adding subjective perception to models of both seasons, model explanatory does get improved very slightly.