ADAPT-pricing

Abstract

Ridesharing platforms use dynamic pricing as a means to control the network's supply and demand at different locations and times (e.g., Lyft's Prime Time and Uber's Surge Pricing) to increase revenue. These algorithms only consider the network's current supply and demand only at a ride's origin to adjust the price of the ride. In this work, we show how we can increase the platform's revenue while lowering the prices as compared to state-of-the-art algorithms, by considering the network's future demand. Furthermore, we show if rather than setting the price of a ride only based on the supply and demand at its origin, we use predictive supply and demand at both the ride's origin and destination, we can further increase the platform's overall revenue. Using a real-world data set from New York City, we show our pricing method can increase the revenue by up to 15% while reducing the price of the rides by an average of 5%. Furthermore, we show that our methods are resilient to up to 25% error in future demand prediction.