Network-based analysis of spatial unbalance in urban bike-sharing systems: A case study of four cities in the United States

Bike-sharing systems are a fascinating mode of transportation, but they face the challenge of spatial unbalance, with bicycles overcrowded at some stations and scarce at others. This study investigates the Pearson correlation coefficient between bike-sharing unbalance and maximum temperature, traffic volume, and average traveled distance across four U.S. cities. Furthermore, using multivariate logistic regression, a strong correlation was observed between station-level unbalance, quantified by the difference in bicycle inflows and outflows at a given station, and clustering coefficient and disparity. Moreover, we conduct a log-likelihood ratio analysis and detect that station-level unbalance is better characterized by a truncated power-law or lognormal distribution rather than a power-law distribution. As a result, this study reveals the factors associated with spatial unbalance and proposes alternative distributions that better capture unbalance patterns, thereby contributing to a deeper understanding of its causes.