Station-Level Forecasting of Bike Sharing Ridership: Station Network Effects in Three U.S. Systems

摘要

This study investigates the effects of demographic and built environment characteristics nearbike sharing stations on bike sharing ridership levels in three operational U.S. systems. Whileprevious studies have focused on the analysis of a single system, the increasing availability ofstation-level ridership data creates the opportunity to compare experiences across systems;particular attention is paid to data quality and consistency issues raised by a multi-city analysis.This project also expands on previous studies by including the network effects of the size andspatial distribution of the bike sharing station network, contributing to a more robust regressionmodel for predicting station ridership.The regression analysis identifies a number of variables as having statistically significantcorrelations with station-level bike sharing ridership: population density; retail job density; bike,walk, and transit commuters; median income; education; presence of bikeways; non-whitepopulation (negative association); days of precipitation (negative association); and proximity to anetwork of other bike sharing stations. Proximity to a greater number of other bike sharingstations exhibits a strong positive correlation with ridership in a variety of model specificationsand while controlling for the other demographic and built environment variables, suggesting thataccess to a comprehensive network of stations is a critical factor supporting ridership. Relative toprevious models, this model will be more widely applicable to a diverse range of communitiesand help those interested in adopting bike sharing systems to predict potential levels of ridershipand identify station locations that will serve the greatest number of riders.