Assortment Optimization Under Consider-Then-Choose Choice Models

Consider-then-choose models, borne out by empirical literature in marketing and psychology, explain that customers choose among alternatives two phases, first screening products to decide which consider then ranking them. In this paper, we develop a dynamic programming framework study the computational aspects of assortment optimization under consider-then-choose premises. Although nonparametric choice models generally lead computationally intractable problems, are able show for many empirically vetted assumptions on how choose, our resulting program is efficient. Our approach unifies subsumes several specialized settings analyzed previous literature. Empirically, demonstrate predictive power modeling combination synthetic real industry data sets, where prediction errors significantly reduced against common parametric models. experiments, algorithms practical computation schemes outperform state-of-the-art integer solver terms running time, parameter regimes interest. This paper was accepted Yinyu Ye, optimization.