A bikeshare station area typology to forecast the station-level ridership of system expansion

• Main Authors: Steven R. Gehrke, Timothy F. Welch
• Format: Article
• Language: English
• Published: University of Minnesota 2019-04-01
• Series: Journal of Transport and Land Use
• Subjects: bikeshare; built environment; latent class analysis; bike ridership; machine learning
• Online Access: https://www.jtlu.org/index.php/jtlu/article/view/1395

The continuous introduction and expansion of docked bikeshare systems with publicly available origin-destination data have opened exciting avenues for bikeshare research. In response, a flux of recent studies has examined the sociodemographic determinants and safety or natural environment deterrents of system ridership. An increasing abundance of disaggregate spatial data has also spurred recent calls for research aimed at extending the utility of these contextual data to model bikeshare demand and trip patterns. As planners and operators seek to expand bikeshare services into underserved areas, a need exists to provide a data-driven understanding of the spatial dynamics of bikeshare use. This study of the Washington, DC, metro region’s Capital Bikeshare (CaBi) program answers this call by performing a latent class cluster analysis to identify five bikeshare station area types based on variation in a set of land development pattern, urban design, and transportation infrastructure features. This typology is integrated into a planning application exploring the potential for system expansion into nearby jurisdictions and forecasting the associated trip-making potential between existing and proposed station locations.