Download This PaperModeling determinants of ridesourcing usage: A census tract-level analysis of Chicago
Ghaffar, A., Mitra, S., Hyland, M., 2020. Modeling determinants of ridesourcing usage: A census tract-level analysis of Chicago. Transp. Res. Part C Emerg. Technol. 119, 102769. https://doi.org/10.1016/j.trc.2020.102769
31 Pages Posted: 4 May 2020 Last revised: 4 Sep 2020
Date Written: April 7, 2020
Abstract
Ridesourcing services provided by companies like Uber, Lyft, and Didi have grown rapidly over the past decade and now serve a sizable portion of trips in many metropolitan areas. An understanding of these services (e.g. to whom, where, when, and for what purposes do they provide service?) is critical for regulating, planning, and managing urban multi-modal transportation systems effectively. Unfortunately, little is known about ridesourcing travel because private companies providing ridesourcing services were not previously subject to data sharing requirements. Fortunately, the city of Chicago recently collected and released spatially (census tract) and temporally (15-minute interval) aggregated data on ridesourcing trips collected from private companies. This study analyzes the Chicago ridesourcing data to examine factors influencing ridesourcing usage. The study employs a random-effects negative binomial (RENB) regression approach to model ridesourcing usage. Determinants considered in the model include weekend vs. weekday and weather variables as well as census tract socio-demographics and commute characteristics, land-use variables, places of interest, transit supply, parking features, and crime. The model results indicate ridesourcing demand is higher on days when temperatures are lower, there is less precipitation, and on the weekend, as well as in census tracts with (i) higher household incomes, (ii) a higher percentage of workers who carpool or take transit to work, (iii) a higher percentage of households with zero vehicles, (iv) higher population and employment density, (v) higher land-use diversity, (vi) fewer parking spots and higher parking rates, (vii) more restaurants, and (viii) more homicides. The results also demonstrate a non-linear (and insightful) relationship between ridesourcing demand and transit supply variables. The paper discusses the implications of these model results to inform transportation planning and policymaking as well as future research.
Keywords: Mobility-on-Demand, Shared Mobility, Transportation Network Companies, Ride-hailing, Random-effects Negative Binomial, Demand Model, Parking, Transit
Suggested Citation: Suggested Citation