摘要
研究一类方差分量模型中的方差分量的估计改进问题,首先在含两个方差分量模型中给出σ21二次型估计类,并且此估计类还具有无偏性和不变性.考虑二次损失(δ-θ)2,在此估计类基础上放弃无偏性进行非负改进,不仅得到优于二次不变无偏估计类的σ21的非负二次不变估计类,而且还说明了它优于方差分析估计和最小均方误差估计,文献[5]中给出s>2时的非负改进,但是非负改进存在是有条件的,本文克服了这个缺陷.最后给出了非负改进存在的充分必要条件.
We mainly discuss the improvement of estimators of variance components in a class of variance components model. First, we propose a class of quadratic estimator(QE) of σ1^2 with two variance components and derive QE having some properties, such as unbiased and invariant. Furthermore, we improve the estimators to nonegative estimate by giving up unbiased property, we derived the conditions under which there will exist improved nonnegative estimators of the σ1^2, uniformly dominating the quadratic invariant unbiased estimator, the analysis of variance estimator (ANOVAE) and the least square error estimator (LSE). Finally, we derive the sufficient and necessary conditions for nonegative estimator existing.
出处
《大学数学》
北大核心
2008年第4期83-87,共5页
College Mathematics
基金
安徽省高校省级自然科学研究重点项目(2007A012)
安徽师范大学青年基金
关键词
方差分量
二次不变估计
非负估计
variance components
quadratic invariant estimator
nonnegative estimator