A data-driven newsvendor problem: From data to decision

摘要: Abstract Retailers that offer perishable items are required to make ordering decisions for hundreds of products on a daily basis. This task is non-trivial because the risk too much or little associated with overstocking costs and unsatisfied customers. The well-known newsvendor model captures essence this trade-off. Traditionally, problem solved based demand distribution assumption. However, in reality, true hardly ever known decision maker. Instead, large datasets available enable use empirical distributions. In paper, we investigate how exploit data making better decisions. We identify three levels which can generate value, assess their potential. To end, present data-driven solution methods Machine Learning Quantile Regression do not require assumption specific distribution. provide an evaluation these point-of-sales German bakery chain. find approaches substantially outperform traditional if dataset enough. also benefit improved forecasting dominates other potential benefits methods.