Behaviorally-Informed Discrete Choice Models in Operations Management

This thesis advances choice-based Revenue Management by integrating behavioral realism into analytical models of pricing and assortment optimization. Traditional discrete choice models (DCMs) simplify customer behavior by assuming that greater variety always increases purchase likelihood, that price sensitivity is homogeneous across different customer segments, and that each consumer purchases only a single product. I address these limitations through three studies, each capturing a distinct behavioral dimension that shapes customer choice. The first study formalizes the paradox of choice by modeling the no-purchase utility as a U-shaped function of assortment size and embedding this formulation within the Multinomial Logit (MNL), Nested Logit (NL), and Mixed Multinomial Logit (MMNL) frameworks. The second study examines the joint pricing and assortment problem under the MMNL model with capacity constraints, accounting for heterogeneous price sensitivities across customer segments. It develops a Fully Polynomial Time Approximation Scheme (FPTAS) that achieves near-optimal solutions efficiently. The third study introduces a discrete choice model that captures multi-product and multi-unit purchases and provides exact and approximate algorithms for assortment and pricing optimization. Collectively, these studies enhance the descriptive and prescriptive power of DCM-based Revenue Management by linking behavioral realism with analytical tractability, offering both theoretical insights and practical relevance.

Department dc:contributor.department

Business

Year dc:date.issued

2026

Author dc:creator

• Milad Mirzaee

Advisor dc:contributor.supervisor

• Li, Guang

• Assortment Optimization
• Paradox of Choice
• Discrete Choice Model
• Multinomial Logit

Language dc:language.iso

eng

Handle dc:identifier.uri

https://hdl.handle.net/1974/36126

OAI identifier oai:identifier

oai:queensu.scholaris.ca:1974/36126