| 离散变量参数UMA限中均匀数取值的研究 |
| |
| 引用本文： | 周源泉.离散变量参数UMA限中均匀数取值的研究[J].强度与环境,2003,30(3):43-49. |
| |
| 作者姓名： | 周源泉 |
| |
| 作者单位： | 北京强度环境研究所,北京,100076 |
| |
| 摘 要： | 本文用Jeffreys的无信息先验分布的Bayes方法与随机化的Fiducial方法，对UMA置信限中的均匀数u的取值问题进行了研究，结果表明，Mann建议u取1／2，对二项分布与有替换定时截尾的指数分布这是合适的，但对预定成功数或失败数的负二项分布，此值略为偏小，其相对误差与样本量n的关系为l／(2n 1).
|
| 关 键 词： | 离散变量参数 UMA限 均匀数 Bayes方法 Fiducial方法 置信限 指数分布 |
| 文章编号： | 1006-3919(2003)03-0043-07 |
| 修稿时间： | 2002年8月7日 |
On selecting the Value of uniform number in UMA limits of the parameter of discrete random Variables |
| |
| ZHOU Yuan-quan.On selecting the Value of uniform number in UMA limits of the parameter of discrete random Variables[J].Structure & Environment Engineering,2003,30(3):43-49. |
| |
| Authors: | ZHOU Yuan-quan |
| |
| Abstract: | Using Bayesian method of Jeffreys' noninformative prior and the randomized Fiducial method, the selection problem of value of uniform number in UMA limits of the parameter of discrete random variable is discussed. The results indicate, Mann' s proposition, uniform number u take 1/2, is suitable for the binomial distribution and the exponential distribution with time truncated test with replacement,but u = 1/2 is slightly small for the negative binomial distribution with predeterninate success number or fail number.The relative error is 1/(2n + 1) , where n is the sample size. |
| |
| Keywords: | Bayesian method Fiducial method UMA limit Uniform number Binomial distribution Negative binomial distribution Exponential distribution |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
