Incorporating advanced behavioral models in integer optimization
Modern demand models rely on discrete choice models. These models are able to predict the choice behavior of individuals in a detailed way, accounting not only for the price and quality of the goods, but also for the characteristics of the customers. Consequently, they allow to capture the heterogeneity of the behavioral patterns in the population, that generate the demand. Unfortunately, these models are highly non linear and non convex in the variables of interest, and are therefore difficult to include in an integer optimization framework. In this talk, we propose a new modeling framework that leads to a linear formulation of any discrete choice model, allowing to include them in an integer optimization framework. As the project is at an early stage, the fundamental concepts will be presented, together with some illustrative examples. The pros and cons of the approach will also be discussed.
Record created on 2015-09-15, modified on 2017-02-16