Thesis Title: Pricing and Revenue Management in Supply Chain Networks and Service Systems
Dr. He Wang, School of Industrial and Systems Engineering, Georgia Tech
Dr. Pinar Keskinocak, School of Industrial and Systems Engineering, Georgia Tech
Dr. Anton Kleywegt, School of Industrial and Systems Engineering, Georgia Tech
Dr. Martin Savelsbergh, School of Industrial and Systems Engineering, Georgia Tech
Dr. Cong Shi, Department of Industrial and Operations Engineering, University of Michigan
Date and Time: 7:30-9:00pm EST (Thursday) April 29, 2021
Meeting URL: https://bluejeans.com/1152852591/
Meeting ID: 1152852591 (BlueJeans)
This thesis comprises of four topics focusing on pricing and revenue management. Pricing and revenue management is a powerful technique to enhance firm profitability through demand and supply management. Because of its efficacy, it has been adopted by small and big companies in broad range of industries.
In the first part of the thesis, joint work with He Wang, we consider a canonical quantity-based network revenue management problem, where a firm accepts or rejects incoming customer requests irrevocably in order to maximize expected revenue given limited resources. We design a new heuristic, which builds upon a family of re-solving heuristics that periodically re-optimize a deterministic approximation to the original problem. Our heuristic proves to have a strong theoretical performance guarantee and desirable numerical results.
The second part is joint work with Martin Savelsbergh and He Wang. We consider an integrated pricing and routing problem on a network, which is motivated by applications in freight transportation. We propose two algorithms for the solution of this problem: a Frank-Wolfe type algorithm, and a primal-dual algorithm using an online learning technique. Both algorithms prove to have desirable convergence rates. Numerical experiments show significant profit improvement of integrated pricing and routing decisions over independent pricing or routing strategies.
In the third part of this thesis, joint work with Sushil Mahavir Varma, Siva Theja Maguluri and He Wang, we study a two-sided queueing system under joint pricing and matching controls. This problem is motivated by applications in gig economy and online marketplaces. We propose a two-price and max-weight matching policy, which proves to achieve the optimal rate. The proposed algorithm also show promising numerical results when compared to other algorithms in similar settings.
The fourth part of the thesis is joint work with He Wang. We study a discrete choice model. The problem is motivated by the violation of regularity property, which states that adding an option to a choice set cannot increase the choice probability for any of the original choice options. This property can be observed in real-world applications but cannot be represented by a widely-used random utility maximization (RUM) model. We propose a more general choice model -- a model based on a neural network framework -- which can explain any choice phenomenon. Numerical experiments show the model consistent outperforms other models for both synthetic and real datasets.