Data-Driven Simulation of Ride-Hailing Services using Imitation and Reinforcement Learning

  • 2021-04-06 16:49:26
  • Haritha Jayasinghe, Tarindu Jayatilaka, Ravin Gunawardena, Uthayasanker Thayasivam
  • 0


The rapid growth of ride-hailing platforms has created a highly competitivemarket where businesses struggle to make profits, demanding the need for betteroperational strategies. However, real-world experiments are risky and expensivefor these platforms as they deal with millions of users daily. Thus, a needarises for a simulated environment where they can predict users' reactions tochanges in the platform-specific parameters such as trip fares and incentives.Building such a simulation is challenging, as these platforms exist withindynamic environments where thousands of users regularly interact with oneanother. This paper presents a framework to mimic and predict user,specifically driver, behaviors in ride-hailing services. We use a data-drivenhybrid reinforcement learning and imitation learning approach for this. First,the agent utilizes behavioral cloning to mimic driver behavior using areal-world data set. Next, reinforcement learning is applied on top of thepre-trained agents in a simulated environment, to allow them to adapt tochanges in the platform. Our framework provides an ideal playground forride-hailing platforms to experiment with platform-specific parameters topredict drivers' behavioral patterns.


Quick Read (beta)

loading the full paper ...