Statistical learning in inventory management

We survey the recent development on inventory control problems through the lens of statistical learning. This survey chapter consists of two parts. In the first part, we survey on research works on the single period setting, which is also known as the newsvendor model. We first showcase the performance guarantee of the sample average approximation (SAA) method in the sampling based setting. Then, we survey on other non-SAA approaches. Finally, we also survey on a generalization of the sampling based setting, based on the supervised learning model. In the second part, we survey on multi-period models in the sampling based setting. We survey on existing works that explore the intersection of statistical learning and multi-period inventory control in multiple settings. For example, we review research works that consider capacity constraints on ordering quantity, pricing decisions, serial systems, fixed cost model, and censored demand data.