Disclosing the Impact of Micro-level Environmental Characteristics on Dockless Bikeshare Trip Volume: A Case Study of Ithaca

Abstract

Although prior literature has examined the impact of the built environment on cycling behavior, the focus has been confined to macro-level environmental characteristics or limited objective features. The role of perceived qualities measured from visual surveys is largely unknown. Using a large amount of dockless bikeshare trajectories, this study maps the cycling trip volume at the street segment level. The research evaluates the micro-level objective features and perceived qualities along the cycling routes using street view imagery, computer vision, and machine learning. Through several regression models, the strengths of both micro-level environment characteristic groups are comprehensively analyzed to reveal their impacts on cycling volume at the street level. Overall, objective features exhibit higher predictive power than perceived qualities, while perceived qualities can complement objective features. The research justifies the significant impacts of micro-level environment characteristics on cycling route choices. It provides a valuable reference for urban planning toward a sustainable cycling-friendly city.