A Study of Alternative Quantile Estimation Methods in Newsboy-Type Problems

摘要

The newsboy problem solutions under the conditions of risk and of uncertainty about demand are well known. The former is the case where the distribution of demand is known or estimated, and the latter is the case where it is not known but the range of demand is given. Minimizing the expected cost under risk and the minimax approach under uncertainty are well-known methods to solve the problem. The situation is considered where demand frequency data is acquired, one observation per decision period. For the expected value solution, this study presents five candidate estimators based on the order statistics and evaluates them by comparing costs with those achieved using the minimax rule under uncertainty. This is done by simulation; the results also provide information about the switching period from the minimax rule to the proposed quantile estimators. (Author)