ASSORTMENT OPTIMIZATION AND PRICING PROBLEMS UNDER MULTI-STAGE MULTINOMIAL LOGIT MODELS

Abstract

In most E-commerce scenarios such as hotel booking and online shopping, products are not offered to customers simultaneously. Instead, they are divided into different webpages and presented to customers sequentially. In this thesis, we focus on solving a common problem faced by online retailers: when products are revealed to customers sequentially, which products should the retailers display at each stage and what prices should the retailers charge for each product so that the expected revenue can be maximized? To solve those problems, we generalize the classical multinomial logit model to capture the customer’s choice behavior under the sequential setting and present efficient algorithms for different generalized choice models and different operational constraints.