Analyzing And Evaluating Spatiotemporal Heterogeneous Effects Of Ride-hailing On Urban Public Transit Demand

As a newly emerging mode of urban traffic,ride-hailing services have gained increasing popularity over the last decade.Although this system attracts tremendous attention,the understanding of the effects of ride-hailing on urban traffic is still limited due to the unavailability of practical data.The lack of impact assessment of the ride-hailing services will limit its integration with public transit systems.To fill the research gap,the thesis analyzed how ride-hailing services affect the bus and subway ridership and proposed the optimal ride-hailing scheduling strategies to enhance the attractiveness of public transit system using vehicle GPS trajectories,smart card data,and built environments,collected in Shenzhen,in May 2019.First,the mixed geographically weighted regression and spatiotemporal weighted regression models are applied to capture the spatial and temporal effects of ride-hailing services on public transit ridership.The results show that the substitution and supplement of the newly mode with traditional transit modes are not constant and they vary significantly at different times of the day and in different areas.The spatiotemporal variations are highly determined by the distribution of travel demand,transit coverage,the function of the subway station(transfer/other),and user’s aged group(youth/older people)become the key factors affecting the effects of ride-hailing services.The follow-up analysis proposes several ride-hailing-related strategies involving a dynamic search range for ride-hailing orders,"P+R" facilities,bus rapid route,and real-time information on bus arrivals to lead the ride-hailing services to improve the attractiveness of the transit system.Second,the thesis develops a ride-hailing scheduling strategy evaluation framework combining machine learning models and spatial statistics to identify the effectiveness of the strategies by estimating the predicted value of bus and subway ridership.Two types of strategies,i.e.,real-time scheduling based on proximity principle and driver-fixed-based remote scheduling,are applied to identify the optimal strategies.The simulating experiments are conducted under bus-ride-hailing on weekday nights and subway-ride-hailing on weekend scenarios.The results indicate that the transfer of ride-hailing services from city centers to peripheral areas is going to enhance the city-level transit ridership.The real-time scheduling based on proximity principle generates more robust effects than the driver-fixed-based remote scheduling does.These two types of strategies both make differences for the transit system in the city periphery,while they have a chequered past in the city center.There are 34 figures,21 tables and 74 references.