查看文章

This study aims to investigate the impact of the emerging app-based for-hire vehicles on taxi industry through quantitative analyses of Uber and taxi demands for neighborhoods of New York City (NYC). Demand forecasting models, which can account for the spatial dependence of Uber and taxi trips are developed. In the empirical analysis, we explore the spatio-temporal patterns of Uber and taxi pick-up data. A high correlation between taxi and Uber pick-ups can be observed, especially in the central areas of the City. From 2014 to 2015, Uber trips increased dramatically by 10 million (223.3%), while taxi trips (include both yellow and green taxis) decreased slightly by 0.8 million (1.0%). The rate of growth of Uber is the lowest in Manhattan (201.2%), and the highest in the outer boroughs like Bronx (597.0%) and Staten Island (573.0%). Results of the Moran’s I test confirm the spatial dependence of both taxi and Uber demands. Linear models, spatial error models, and spatial lag models are developed to estimate the taxi and Uber demands of each neighborhood using socio-economical and transportation-related characteristics. The spatial error models are found to outperform the other two by capturing the spatial dependence via a spatially lagged dependent variable. Neighborhoods with lower transit access time (TAT), higher length of roadways, lower vehicle ownership, higher income and more job opportunities are associated with higher taxi/Uber demands.