EURO-Online login
- New to EURO? Create an account
- I forgot my username and/or my password.
- Help with cookies
(important for IE8 users)
1199. Pricing, Admission and Repositioning in Ride-Hailing Networks: A Deep Reinforcement Learning Approach
Invited abstract in session WC-55: Big data analysis and AI in transportation, stream Transportation.
Wednesday, 12:30-14:00Room: S02 (building: 101)
Authors (first author is the speaker)
1. | Thomas De Munck
|
Center for Operations Research and Econometrics, UCLouvain | |
2. | Jean-Sébastien Tancrez
|
CORE - Louvain School of Management, Université catholique de Louvain | |
3. | Philippe Chevalier
|
Louvain Institute of Data Analytics and Modelling, UCLouvain |
Abstract
We consider the problem of a ride-hailing platform (e.g., Uber, Lyft) that connects supply with demand over a network of locations. To this aim, the platform makes pricing, customer admission, and driver repositioning decisions. The customers are impatient and have distinct willingness to pay. The drivers can be repositioned by the platform, or can choose to relocate to other locations according to their own choice. We formulate this problem as a discrete-time Markov decision process, and find a near-optimal policy by combining mathematical optimization with deep reinforcement learning. Following our approach, we first find a heuristic policy by solving repeatedly an optimization problem based on the expected values. Then, we train two neural networks to replicate the heuristic policy, and learn the associated value function. We finally improve the policy further through the Proximal Policy Optimization algorithm. By applying our model to real data from New York City, we demonstrate the effectiveness of our approach in comparison with alternative methods. In addition, we explore the interplay between pricing, customer admission, and driver repositioning, and evaluate the effectiveness of these decisions in balancing supply and demand.
Keywords
- Transportation
- Stochastic Models
- Artificial Intelligence
Status: accepted
Back to the list of papers