【学术讲座】航空业数据分析：从辅营产品定价到稳健化飞机航线调配

Airline Analytics: From Ancillary Pricing to Robust Aircraft Routing

讲座信息 | INFOMATION

讲者：Prof. Chung-Piaw TEO, National University of Singapore

时间：2026年6月18日（周四） 15时

地点:   同济大厦A座501教室

讲座摘要| ABSTRACT

Modern airlines compete on two fronts: how to price what customers buy, and how to operate the fleet that carries them. Both are decision problems made under sparse, noisy, and shifting data, and both demand methods that are not only theoretically sound but transparent and stable enough to deploy. This seminar presents two recent studies, conducted with major Asia-Pacific carriers, that bring modern optimization and analytics to bear on each front.

The first study tackles ancillary revenue management through the pricing of checked-baggage tiers — close substitutes sold across thousands of route-segments where demand data are too sparse to estimate fully separate models. We adopt a rational-inattention demand specification, in which customers ration scarce attention across options, and calibrate it via a linear-programming estimator that pools information across booking features. The framework yields a semi-closed-form optimal price ladder, an explicit decomposition of estimation error into merging bias, approximation bias, and statistical error, and a quadratic revenue-loss bound linking sample size to economic performance. In a randomized field experiment spanning 1.41 million booking sessions, the method raised baggage revenue by roughly 8% —while lowering average posted prices, since the recommended price distribution sits stochastically below the incumbent's.

This is based on joint work with Jin Xiao, Ren Junjie from NUS, Liu Changchun from Xian Jiaotong University, and Sun Hailong from Shanghai Jiaotong Univeristy. 

The second study turns to operational resilience in aircraft routing (tail assignment), the airline's last lever to absorb disruptions once the schedule is published. We show that controlling cascading delays is not about adding slack but about placing it: concentrating existing ground time into large buffer walls can cut mean propagated delay by over 50% with no schedule change. To find such structures robustly under shifting, post-pandemic delay patterns, we develop a distributionally robust routing model that needs only three summary statistics per flight (mean, mean absolute deviation, and support) and a provably exact column-generation pricing algorithm. Evaluated on 25 months of operational data from a hub-and-spoke carrier, the method reduces expected delay cost over current practice, with its advantage growing sharply when delay patterns are contaminated.

 This is based on joint work with Li Jingying from NUS, and Shan Wenxuan from Beijing Jiaotong University.

Together, the two studies illustrate a common methodological arc — recover structure from limited data, build tractable and interpretable optimization models around it, and validate on real operations — and argue that the frontier of airline analytics lies in coupling rigorous OR theory with deployment-ready decision rules on both the revenue and operational sides of the business.

嘉宾简介| About the speaker:

Chung-Piaw TEO is Stephen Riady Professor and Executive Director of the Institute of Operations Research and Analytics (IORA) in the National University of Singapore. Prior to the current appointments, he was a Head of Department, Acting Deputy Dean, Vice-Dean of the Research and Ph.D Program as well as Chair of the Ph.D Committee in the NUS Business School. He serves as a department editor for Management Science (Optimization and Decision Analytics), and a former area editor for Operations Research (Operations and Supply Chains).

He studied issues in service and manufacturing operations, supply chain management, discrete optimization, and machine learning. He has also served on several international committees such as the Chair of the Nicholson Paper Competition (INFORMS, US), member of the LANCHESTER and IMPACT Prize Committee (INFORMS, US), Fudan Prize Committee on Outstanding Contribution to Management (China). He was selected as INFORMS Fellow on the year of 2019.