A dynamic region-division based pricing strategy in ride-hailing

In recent years, ride-hailing has played an important role in daily transportation. In the ride-hailing system, how to set the prices for orders is a crucial issue. Considering the variations in vehicle supply and order demand in different regions of the same city, it is necessary to divide regions according to the supply and demand in each region and develop differentiate pricing strategy. However, the traditional fixed region-division strategy is difficult to adapt to the dynamically changed supply and demand over the time. To address this issue, we propose a dynamic region-division based pricing strategy to set prices according to the demand and supply of different regions to maximize the long-term profit of the platform. Specifically, we first design a dynamic region-clustering algorithm based on Deep Q Network and K-Means algorithm to dynamically cluster small zones with similar demand and supply status and nearby locations into the same region. We then propose an adaptive multi-region dynamic pricing algorithm to set unit price for each clustered region to maximize the long-term profit of the ride-hailing platform. We further run extensive experiments based on a real-world dataset to demonstrate the effectiveness of our proposed algorithm. The experimental results show that the platform’s profit is increased under different pricing algorithms combined with dynamic region-clustering algorithm. Furthermore, we find that the combination of adaptive multi-region dynamic pricing algorithm with dynamic region-clustering algorithm can bring higher profits, serve more orders and have a higher service rate than benchmark approaches.