EURO-Online login
- New to EURO? Create an account
- I forgot my username and/or my password.
- Help with cookies
(important for IE8 users)
2341. Machine-learning-based policy for the dynamic pricing of substitutable itineraries
Invited abstract in session WC-27: Machine Learning and Ensemble Learning with optimization methods, stream Mathematical Optimization for XAI.
Wednesday, 12:30-14:00Room: 047 (building: 208)
Authors (first author is the speaker)
1. | Yue Su
|
Informatique, Université de Lille | |
2. | Axel Parmentier
|
Cermics | |
3. | Antoine Désir
|
ISEAD Business School |
Abstract
In the airline revenue management, dynamic pricing for substitutable itineraries presents significant challenges due to the stochastic nature of customer choices. The dynamic pricing problem can be conceptualized as a Markov decision process with the objective of maximizing the total expected revenue over a finite selling horizon. However, the complexity of this model, characterized by its multi-dimensional state and action spaces, makes it computationally prohibitive to solve exactly, even for for small-to-medium-sized instances. As a result, classic methods such as dynamic programming prove inefficient to address the dynamic pricing problem for larger instances. In this work, we introduce learning-based policies tailored to dynamic pricing problem of substitute itineraries with the same origin and destination. These policies are encoded by hybrid machine learning pipelines, while taking into account the heterogeneous customer choices. We proposed two different configurations of pipelines, where we have designed tailored training process for each of them. To demonstrate the performance of our learning-based policies, we first benchmark it against the optimal policy obtained from dynamic programming on small-sized instances. Then, we benchmark it against the state-of-the-art approximated policy heuristics on larger-sized instances. In addition, our method can be easily adapted to airline network scenarios, where itineraries of different origins and destinations are included.
Keywords
- Airline Applications
- Machine Learning
- Programming, Dynamic
Status: accepted
Back to the list of papers