Abstract：

In ranked recommendation systems such as hotel booking platforms, consumers often browse options sequentially in search of a satisfactory choice. While general sequential choice models can accurately capture this behavior, applying them to assortment optimization—selecting the best set of products to display—can be computationally intractable. To overcome this challenge, the cascade model, which assumes consumers make independent decisions after evaluating each item, is widely used in practice for its simplicity. In this talk, I will present both theoretical and empirical evidence supporting the cascade approximation. Specifically, I will show that the cascade model closely matches true choice probabilities and yields assortment decisions with provable revenue guarantees—even under known, adversarial, or random evaluation orders. Experiments using Expedia hotel booking data further demonstrate the model’s practical effectiveness in real-world settings.

Bio：

Dr. Pin Gao is an Assistant Professor at the School of Data Science, The Chinese University of Hong Kong, Shenzhen. He holds a B.S. in Physics from Wuhan University and an MPhil in Physics and Ph.D. in Operations Research from The Hong Kong University of Science and Technology. His research focuses on platform operations management, including developing simple models to approximate complex customer behavior, refining industry strategies to enhance performance both theoretically and empirically, and regulating platforms to promote social responsibility.