Abstract
With the rapid development of mobile applications, the ride-hailing services such as Uber in America and Didi-taxi in China have been very popular all over the world as they provide convenience to the users. A key factor that makes the ride-hailing service successful is the user experience, which is highly related to the passenger service time. In our work, we define service time as the passenger wait time plus the time the driver takes the passenger to the destination. Many studies have been done on the research of conventional taxis. However, few existing works have been done to comprehensively dissect the passenger service time in ride-hailing services due to the complex real-world factors, e.g., trip origin, trip destination, and weather, etc. In this paper, we firstly analyze the impact factors of service time based on 36.6 million ride-hailing trips. Then we propose an improved XGBoost model BO-XGBoost, which combines with the Bayesian Optimization method, to predict the service time. Comprehensive experiments on real datasets show that our BO-XGBoost achieves better prediction accuracies than other methods.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call