摘要
用Bootstrap法估计随机变量的概率分布广泛适用于样本为独立同分布情形.立足于考虑随机变量序列{Xn,n≥1}为NA相协样本条件下均值X-n的Bootstrap逼近问题,首先定义了强平稳组间独立的负相协样本,然后对样本分成k组情况下X-n的k个刀切虚拟值Yi(i=1…k)赋予质量1k,得到“经验分布函数Fk*,从F*k中抽取k个独立样本Y1*,…,Yk*,用n(Yk*-Xn)的条件分布去模拟n(Xn-μ)的分布,最后证明其相合性成立.
Bootstrap method of estimating probability distribution of random variables is available for sample of I. I. D. Bootstrap approximate in valid of NA sample mean - value is studied. In the case that sample is divided into k classes, negative association sample between strongly stationary and independent classes is first defined, then the 1 jackknife virtual value Yi（ i = 1…k） ofXn is given the mass 1/k. From which, the empirical distribution function Fk^＊ are obtained. The independent sample Yi^＊ ,…, Yk^＊ is extracted from Fk^＊. It is shown that if the distribution √n（Yk^＊-μ） is simulated by the conditional distribution of √n（Yk^＊-Xn）,then the consistency holds.
出处
《黑龙江大学自然科学学报》
CAS
北大核心
2006年第2期257-259,264,共4页
Journal of Natural Science of Heilongjiang University
