Spring 2016 Seminar Series

DATE: 3/10/16
LOCATION: E51-325
TIME: 4:15pm
Reception immediately following

SPEAKER:
Ali Aouad

TITLE
Assortment Optimization Under Consider-Then-Choose Choice Models

ABSTRACT
Assortment planning is a major operational issue that arises in many industries, such as retailing, airlines and consumer electronics. Given a set of products (or services) that are differentiated by price, quality and possibly other attributes, one has to decide on the subset of products and the respective quantities that will be stocked and offered in the face of heterogeneous customers who exhibit complex substitution behaviors.

 

We study several assortment models under the general family of consider-then-rank choice models, and show how different assumptions on the underlying customer choice model translate to the complexity of the model. In particular, we establish that for many practical and empirically vetted assumptions on how customers consider and then rank, the resulting assortment problem is computationally tractable. Our approach relies on a divide-and-conquer decomposition scheme in a graph-based representation of the problem. For generic instances, we derive tight inapproximability results, and prove that a common sense heuristic, used by practitioners, in fact yields the best-possible approximations. In computational experiments, our algorithms runs in time comparable to a commercial MIP solver in several parameter regimes, and even outperforms it in some cases of practical interest.

 

Joint work with Profs. Vivek Farias and Retsef Levi.