Ride-sharing Determinants: Spatial and Spatio-temporal Bayesian Analysis for Chicago Service in 2022

Abstract

The rapid expansion of ride-sharing services has caused significant disruptions in the transpor-tation industry and fundamentally altered the way individuals move from one place to another. Accurate estimation of ride-sharing improves service utilization and reliability and reduces travel time and traffic congestion. In this study, we employ two Bayesian models to estimate ride-sharing demand in the 77 Chicago community areas. We consider demographic, scoio-economic, transportation factors as well as land-use characteristics as explanatory variables. Our models assume conditional autoregression (CAR) prior for the explanatory variables. Moreover, the Bayesian frameworks estimate both the unstructured random error and the struc-tured errors for the spatial and the spatiotemporal correlation. We assessed the performance of the estimated models and the residuals of the spatial regression model have no left-over spatial structure. For the spatiotemporal model, the squared correlation between actual ride-shares and the fitted values is 0.95. Our analysis revealed that the demographic factors (populations size and registered crimes) positively impact the ride-sharing demand. Additionally, the ride-sharing demand increases with higher income and increase in the economically active propor-tion of the population as well as the residents with no cars. Moreover, the transit availability and the walkability indices are crucial determinants for the ridesharing in Chicago.