288 / 2024-04-24 10:05:55

Data Driven Dynamic Pricing and Scheduling in Ride-hailing Platform with Temporal-spatial Information

ride-hailing; temporal-spatial information; dynamic pricing; vehicle scheduling; deep reinforcement learning

As a new urban transportation service mode, ride-hailing has a significant impact on residents' travel behavior, and even gradually changes the structure of urban transportation network. Compared with the traditional travel mode, ride-hailing is unified scheduled and operated by a specific ride-hailing service platform, which have higher matching efficiency of supply and demand. However, there is still a lot of room for improvement in the overall market efficiency of ride-hailing. It is mainly reflected in the lack of foresight and flexibility in the dynamic pricing and scheduling strategies in the current ride-hailing market. In recent years, the ride-hailing platform provides users with convenient travel services through smartphone applications, but also accumulates a lot of temporal-spatial data. By virtue of a large amount of temporal-spatial travel information, how to explain the impact of the synergistic mechanism of dynamic pricing and scheduling strategies to the ride-hailing market is a crucial pain point to further improve the operational efficiency of the ride-hailing market. Therefore, based on the temporal-spatial travel data of yellow taxis in Manhattan, New York, this paper proposes a deep reinforcement learning model considering the benefits of ride-hailing platform, passengers and drivers to optimize the dynamic pricing and scheduling behavior of the ride-hailing platform simultaneously. The results show that, compared with the baseline model, the model proposed in this paper can optimizing the revenue of the ride-hailing platform while improving order response rate, vehicle occupancy rate, and the overall social welfare.