Speaker: Yufeng Liu
John D. MacArthur Professor of Statistics
Department of Statistics
University of Michigan
Title: Low-Rank Online Dynamic Assortment with Dual Contexts
Abstract: In e-commerce, delivering real-time personalized recommendations from vast catalogs poses a critical challenge for retail platforms. Maximizing revenue requires careful consideration of both individual customer characteristics and available item features to optimize assortments over time. In this talk, we consider the dynamic assortment problem with dual contexts - user and item features. In high-dimensional scenarios, the quadratic growth in dimensionality complicates computation and estimation. To tackle this challenge, we introduce a new low-rank dynamic assortment model to transform this problem into a manageable scale. We then propose an efficient algorithm that estimates the intrinsic subspaces and uses an upper-confidence-bound (UCB) approach to address the exploration-exploitation trade-off in online decision-making. Some theoretical properties, as well as simulations and an application to Expedia’s hotel recommendation, will be discussed.
This talk is based on joint work with Seong Jin Lee at UNC-Chapel Hill and Will Wei Sun at Purdue University.