【Academic Seminar】Omnichannel Assortment Optimization under the Multinomial Logit Model with a Features Tree
Topic：Omnichannel Assortment Optimization under the Multinomial Logit Model with a Features Tree
Speaker：Prof. Venus LO, City University of Hong Kong
Time & Date：10:30 am - 11:30 am, Wednesday, November 25, 2020
Venue：Zoom, Meeting ID: 559 916 3678 (Password: 962062)
Consider the assortment optimization problem faced by an omnichannel retailer that operates a physical store and an online store. Customers who purchase from the online store will first visit the physical store to test out some of the products. Online products may share features with in-store products, and hence customers revise their preferences for online products after examining the products in the physical store. Unlike traditional assortment optimization, the in-store assortment affects revenue from both the online store and the physical store. The full assortment is offered online, and the goal is to select an assortment for the physical store in order to maximize the retailer’s total expected revenue.
We use a tree to describe how products are related by features. The non-leaf vertices on the tree correspond to features, and the leaf vertices correspond to products. The ancestors of a leaf correspond to features of the product. Customers choose to purchase from either the online store or the physical store, and decide to purchase a product from their chosen store according to the multinomial logit model. We consider the setting where all customers purchase online after examining the in-store assortment, and the setting where some customers purchase from each store. When all customers purchase online, we give an efficient algorithm to find the optimal assortment to display in the physical store. We characterize conditions under which it is optimal to display expensive products with under-rated features, and expose inexpensive products with over-rated features. With a mix of customers, the problem becomes NP-hard and we give a fully polynomial-time approximation scheme. We numerically demonstrate that we can approximate the case where products have arbitrary combinations of features, and that our FPTAS performs remarkably well.
Venus LO is an assistant professor at the City University of Hong Kong. She received her PhD from the School of Operations Research and Information Engineering at Cornell University under the supervision of Professor Huseyin Topaloglu. Prior to Cornell, she studied accounting and optimization at the University of Waterloo. Her current research focuses on designing efficient algorithms with provable performance guarantees for assortment optimization problems under new and interesting choice models.