简单子抽样多元双样本检验的改进方法

Improved Simple Subsampling Based on the Nearest Neighbor Method

基于简单子抽样多元双样本检验方法,提出一种改进的检验方法。改进的方法一方面对混合样本进行集成子抽样,既达到平衡样本容量的目的,又尽可能地保留所有样本点的信息;另一方面,在检验统计量的构造中根据样本的非平衡度,采用加权调整的策略,进一步减小样本非平衡度对检验结果的影响。

In this article,we propose an improved simple subsampling based on the nearest neighbor method to solve this problem. The new testing procedure achieves the goal of balancing data as well as to retain the information by using ensemble subsampling scheme. On the other hand,the new method has further increasing testing power by combining the weighting adjustment.