Confidence Sets Having the Shape of a Half-Space
针对p维正态分布均值估计问题，提出了以观测向量为法线的超平面界定的半空间置信区域，并给出了精确覆盖概率的置信区域构造方法和表格。
SUMMARY For the problem of estimating the mean of a p-dimensional normal distribution, p > 1, confidence regions based on half-spaces bounded by a hyperplane having the vector of observations as normal are proposed. Confidence regions with exact probability of coverage are constructed. Tables are provided.