两个一维总体分布相等性的permutation检验  

Permutation tests for equality of two distributions in one-dimension

作　　者：黄群[1][1] 林洁梅[1][1] 赵佳因[1][1] 

机构地区：[1]北京城市学院 公共基础教学部,北京100083

出　　处：《辽宁工程技术大学学报（自然科学版）》2012年第6期929-932,共4页Journal of Liaoning Technical University (Natural Science)

基　　金：国家自然科学基金资助项目(11001267)

摘　　要：检验两个一维总体分布的相等性,采用理论证明和数值模拟的方法,在Prastgaard提出的Kolmogorov-Smirnov统计量T1及相应的permutation统计量T 1基础上,提出了基于K-S统计量的新统计量T2及相应的permutation统计量T 2,研究了在一维的情形下T2与T 2的分布,并进行了数值模拟.结果表明:新统计量在原假设下有相同的极限分布,数值模拟结果优于原统计量.该方法对检验两总体分布的相等性有一定的参考价值.In order to test the equality of two distributions in one-dimension,the method of theoretical proofing and numerical simulation are used in this study.Based on Kolmogorov-Smirnov tests T1 and corresponding permutation test T1 proposed by Prastgaard,the permutation test for another type of statistics conforming K-S test T2 and corresponding permutation test T2 are proposed in this paper.The results show that they have the same limiting distribution in one-dimension.The numerical simulation results using newly proposed test method are slightly better.The method proposed has certain reference value for testing the equality of two distributions.

关 键 词：permutation方法 K-S统计量 permutation统计量 假设检验 统计推断 经验分布函数 极限分布 数值模拟 

分 类 号：O212.7[理学—概率论与数理统计]