Robust Assortment Optimization Under A Heuristic Demand Estimation Methodology

This paper studied a single-period assortment optimization problem in which demand is uncertain and is only known to belong to some computationally tractable uncertainty sets which are derived from a heuristic demand estimation model based on Multinomial Logit Choice model.Researchers in the area of Business hold that the assortment of single retailer cannot meet customers' purchase demand independently.In fact,it is commercial agglomeration which consists of a bunch of retailers who are of complementary assortment can 'meet customers' purchase demand.Therefore,we consider such a retailer which is not isolated but in a commercial agglomeration,which means its assortment planning and product demand are affected by other retailers in this commercial agglomeration.Based on such assumption,we develop a heuristic to estimate the products demand for this specific retailer.As a result,we can obtain the uncertainty sets which represent the demand.After that,we consider two scenarios according to whether assortment is given or not.When the assortment is given,our goal is to set optimal inventory levels for each product the retailer operates in order to obtain maximum profit.In face of the uncertainty of the demand,we choose to use robust optimization methodology to obtain optimal ordering quantity for each product in a specific period.When the assortment is changeable,we first develop a recursive algorithm to derive optimal assortment from potential products.And then we can obtain the optimal inventory levels for this assortment by means of te same methodology proposed above.The numerical experiment indicates that this optimization methodology can successfully trade-off the profit and risks for retailers.