Effects of built environment and weather on demands for transportation network company trips

Abstract

This paper investigates the effects of the built environment and weather on the demands for transportation network companies (TNC) in Toronto. The research is based on a historical dataset of Uber trips from September 2016 to September 2018 in Toronto. A wide range of built environments, sociodemographic, and weather data are generated at the dissemination area-level and fused with the monthly aggregated Uber dataset. To provide insight into the underlying factors that affect TNC demand, a series of aggregate demand models are estimated using log-transformed constant elasticity demand functions, with consideration of the seasonal lag effect. To capture the weather effect, an autoregressive moving average model is estimated for the downtown core of Toronto. The model results show that the influence of lagged ridership and seasonal lag effect have a positive correlation with TNC demand. The trip generation and attraction models reveal that TNC trips increase where when the commuting trip duration is longer than 60 min. It is found that the number of apartments in a dissemination area is positively correlated with TNC trip generation, while the number of single-detached houses has a negative correlation. The time-series model indicates that temperature and total daily precipitations are positively correlated with TNC demand. Due to the lack of comprehensive data sources on the Uber and Lyft ridership, the policymakers often struggle to make evidence-based policy recommendations to regulate such disruptive technologies. The series of models presented in this study will help us better understand the potential users of transportation network companies (TNC) and the effects of land use, built environment and weather on transportation network company trips.