Abstract
A typical decision making problem in practice is to select the best from a collection of alternatives. The value of each alternative, however, is unknown and can only be learned via expensive sampling. Motivated by the emerging popularity of personalized decision making in various areas including healthcare and e-commerce, we study the selection problem in the presence of contextual information, where the best alternative is not universal but depends on certain covariates. Given a limited sampling budget, we propose an adaptive sampling strategy to efficiently learn the decision rule that specifies the best alternative for a given value of the covariates. The sampling strategy is developed via a nonparametric Bayesian approach and is shown to be asymptotically optimal. We demonstrate the usefulness of our methodology via a case study in personalized medicine for selecting the best cancer treatment regimen.
Xiaowei Zhang
Associate Professor
My research research focuses on methodological advances in stochastic simulation and optimization, decision analytics, and reinforcement learning, with applications in service operations management, financial technology, and digital economy.