摘要
本文研究高维数据下两样本均值的检验问题.基于Hotelling's T^2检验,我们提出了适用于高维数据均值检验的复合Hotelling's T^2检验统计量,证明了其渐近正态性并研究了其渐近功效.我们通过模拟和实例分析展示了该检验在有限样本下相比现有高维检验方法的优良性.
The paper is concerned with the two-sample mean testing problem in high-dimensionsettings. We propose a composite Hotelling,s T-square test, establish its asymptotical normality and study its local power. The finite-sample priority of the proposed test over existing high-dimensional tests is shown by simulations and illustrated by a real data-set analysis.
出处
《应用概率统计》
CSCD
北大核心
2017年第4期349-368,共20页
Chinese Journal of Applied Probability and Statistics
基金
supported in part by the National Natural Science Foundation of China(Grant No.11371142)