APR Seminar: Ali Aouad (Massachusetts Institute of Technology)
January 20, 2017 | 10:00am - 11:00am
APR Seminar: Ali Aouad (Massachusetts Institute of Technology)Mudd Hall 303
Title: Revenue Management in Face of Choice Heterogeneity
Abstract: Modern-day retailing involves complex customer choice behaviors. Modeling this choice heterogeneity strikes a delicate balance between explaining the data accurately and prescribing efficient operational policies. We propose a novel "consider-then-choose" modeling approach, borne out by the marketing literature. Experiments on a large purchase panel dataset demonstrate the predictive power of our models against common benchmarks. We develop a dynamic programming algorithm for assortment optimization and show, through a state space collapse, that our approach is tractable under various consider-then-choose models. Further, we study the joint assortment and inventory management problem, where customers exhibit a dynamic substitution behavior. We devise the first provably-good algorithms for practical models of choice, by revealing hidden submodular-like structure. Our approach is an order of magnitude faster than existing heuristics and increases revenue by 6% to 12% in experiments.
This work is based on several papers jointly with Profs. Vivek Farias,Retsef Levi and Danny Segev.
Bio: Ali Aouad is a 4th year PhD candidate in the Operations Research Center at MIT, co-advised by Profs. Vivek Farias and Retsef Levi. Before joining MIT, he earned a BS and an MS degree in Applied Mathematics from the Ecole Polytechnique (Paris) in 2013. His research focuses on the design and performance analysis of algorithms for contemporary problems in revenue management and data-driven analytics. His work was awarded the 2015 MIT ORC Best Student Paper prize and selected as finalist in the 2016 George Nicholson Prize Competition. Previous industry experiences include investment banking, strategic consulting and entrepreneurship.