Online ride-hailing platforms such as Uber have become popular recently. Here, a passenger is matched with a drive, and both can cancel the matching result. The accurate prediction of matches improves the real trading volume and provides both the passengers and drivers with a better user experience.
A recent study on arXiv.org proposes a model for matching success rate prediction. It thoroughly captures feature interactions from both passenger and driver sides and can retain knowledge about a city for future predictions. Furthermore, a learning scheme based on knowledge distillation is proposed. It allows transferring knowledge from other cities to the lightweight model designed for the target city.
The experimental results demonstrate the strength of the suggested model in terms of accuracy and scalability. It could be generalized for applications like friend-making websites and online marketplaces.
In recent years, online ride-hailing platforms have become an indispensable part of urban transportation. After a passenger is matched up with a driver by the platform, both the passenger and the driver have the freedom to simply accept or cancel a ride with one click. Hence, accurately predicting whether a passenger-driver pair is a good match turns out to be crucial for ride-hailing platforms to devise instant order assignments. However, since the users of ride-hailing platforms consist of two parties, decision-making needs to simultaneously account for the dynamics from both the driver and the passenger sides. This makes it more challenging than traditional online advertising tasks. Moreover, the amount of available data is severely imbalanced across different cities, creating difficulties for training an accurate model for smaller cities with scarce data. Though a sophisticated neural network architecture can help improve the prediction accuracy under data scarcity, the overly complex design will impede the model’s capacity of delivering timely predictions in a production environment. In the paper, to accurately predict the MSR of passenger-driver, we propose the Multi-View model (MV) which comprehensively learns the interactions among the dynamic features of the passenger, driver, trip order, as well as context. Regarding the data imbalance problem, we further design the Knowledge Distillation framework (KD) to supplement the model’s predictive power for smaller cities using the knowledge from cities with denser data and also generate a simple model to support efficient deployment. Finally, we conduct extensive experiments on real-world datasets from several different cities, which demonstrates the superiority of our solution.
Research paper: Wang, Y., Yin, H., Wu, L., Chen, T., and Liu, C., “Secure Your Ride: Real-time Matching Success Rate Prediction for Passenger-Driver Pairs”, 2021. Link: https://arxiv.org/abs/2109.07571