Dynamic Assortment Planning under Various Discrete Choice Models
Thursday, October 4, 2018 - 4:45pm - 5:30pm
In this talk, we study the dynamic assortment planning problem under various popular discrete choice models, including the multinomial-logit (MNL) model, the nested logit model, and a contextual MNL model. For each arriving customer, the seller offers an assortment of substitutable products, and then the customer makes the purchase according to a pre-specified choice model. Since all the utility parameters of customers are unknown, the seller needs to simultaneously learn customers' choice behavior and make dynamic decisions on assortments based on the current knowledge. For each discrete choice model, we develop a computationally efficient dynamic policy and establish the corresponding regret bound. For MNL and contextual MNL models, our policies achieve the optimal regret, which is independent of the total number of products.