CEE-CSE PhD Thesis Defense: Chonghuan Wang

Abstract:

Experimental design is foundational across many disciplines, yet its application in operations research (OR) and operations management (OM) introduces unique complexities and opportunities that extend beyond classical statistical goals. This thesis explores why experimentation and its design are critical to OR/OM, identifies the challenges posed by operational and service systems to traditional experimental design, and argues that OR/OM researchers are particularly well positioned to address these challenges.

Specifically, this thesis advances experimental design by incorporating operational perspectives and addressing two key challenges: integrating operational objectives and leveraging operational models to improve experimentation. While traditional approaches—such as A/B testing—focus on statistical efficiency (e.g., reducing variance or bias), OR/OM applications often involve additional concerns, including welfare preservation, revenue optimization, risk control, and non-stationarity. We investigate fundamental trade-offs and interactions among these competing objectives. Additionally, we demonstrate how operational models—particularly Markov Decision Processes (MDPs)—can be employed to estimate long-term cumulative outcomes, such as customer lifetime value, using short-term experimental data.

Thesis Committee Members:

• Prof. David Simchi-Levi (Thesis Supervisor), Department of Civil and Environmental Engineering, MIT