Choice-based optimization and its application to revenue management
UE-EIG.00244
Teacher(s): Pacheco Paneque Meritxell |
Level: Master |
Type of lesson: Lecture |
ECTS: 4.5 |
Language(s): English |
Semester(s): SS-2025 |
Choice-based optimization enable planners to make decisions while taking into account individual’s choice behavior. This behavior is modeled with discrete choice models, which correspond to the state-of-the-art of the disaggregate mathematical representation of the demand. They assume that each individual associates a utility with each alternative they can choose from and selects the one with the highest alternative.
Choice-based optimization problems are receiving increasing attention because they allow to explicitly capture the interplay between the planner’s decisions and the expected demand provided that the decisions are explanatory variables of the discrete choice model. In this course, we first introduce discrete choice models by reviewing their assumptions, mathematical formulations and applications. We then discuss how they can be integrated in an optimization problem by expressing the discrete choice model as a set of linear constraints that can be embedded in a mixed-integer linear formulation (i.e., linear model with both integer and continuous variables). We then apply the introduced modelling techniques in the context of revenue management, one of the application areas where the demand representation plays a very important role in the associated decision-making processes. Revenue management refers to the pricing and revenue optimization as a quantitative approach to set and update pricing and product availability decisions in a consistent and effective fashion. It has proven particularly successful in the airline industry, where fares and ticket offerings dynamically change as a function of the number of free seats, forecast of future demand and specific request characteristics. We introduce the basics on pricing with and without a capacity constraint, then discuss price differentiation aspects and look at revenue optimization problems including customer segmentation. From these single resource problems, we move to the network case, in which multiple resources are used to provide a service. We look at these revenue management topics from a choice-based optimization perspective concerning their modeling, solving and interpretation.
Training aims
• Become familiar with discrete choice models, understand the underlying assumptions and learn how to formulate them.
• Understand the characterization of discrete choice models as a set of linear constraints and how it can be embedded into an optimization problem formulated as a mixed-integer linear programming model.
• Gain a good understanding of the main issues and opportunities in pricing and revenue optimization and learn to identify and exploit these opportunities in different business contexts.
Documentation
• M. Ben-Akiva and S. Lerman, Discrete Choice Analysis: Theory and Application to Travel Demand, MIT Press (1985)
• Pricing and Revenue Optimization, Robert L. Phillips, Stanford University Press (2005)