A Prescriptive Analytics Approach for a Multi-Item Inventory Replenishment Problem Considering Demand Correlation

With the rapid development of online shopping in China, efficient inventory replen- ishment is crucial to ensure fast product delivery. This paper addresses a multi- period multi-item inventory replenishment problem in an e-commerce company. Different from the traditional predict-then-optimize methods, a prescriptive ana- lytics approach (PAA) is applied to prescribe item replenishment decision based on previous transaction data and historical scenario data. We prove that the pro- posed prescriptive analytics approach considering demand correlation converges to the asymptotically optimality w.p.1. Numerical experiments are done to compare the performances of different configurations of prescriptive analytics approach. Sen- sitivity analyses further provide valuable managerial insights for decision makers