Nonparametric Prediction Intervals for Sample Medians in the General Case
研究了在不假设总体分布的情况下，为任意大小的未来样本的中位数构造非参数预测区间，并给出了覆盖概率的最佳下界，这些下界可用超几何和二项分布的尾部概率计算。
Abstract This article deals with nonparametric prediction intervals for the median(s) of a future sample of arbitrary size without any assumption about the parent distribution. The best possible distribution-free lower bounds for the relevant coverage probabilities are derived. Furthermore, these sharp lower bounds are given in terms of lower tail probabilities of hypergeometric and binomial distributions for which extensive tables and good approximations are available. Key Words: NonparametricPrediction intervalsSample medianGeneral distributionsSharp probability bounds