摘要
本文考虑了线性模型中回归系数β＝（β１，…，βｐ）＇和误差方差σ＾２的联立经验Ｒａｙｅｓ（ＥＢ）估计。在二次损失下，利用密度函数及其偏导数的核估计构造出参数θ＝（β１，…，βｐ，σ＾２）的联立ＥＢ估计，在一定条件下证明了θ的联立ＥＢ估计的收敛速度任意接近于１．最后，给出了一个实例。
In this paper, the simultaneous empirical Bayes (EB) estimator for regression cofficient β= (β1,β2,···,βp)' and error variance σ2 is considered in linear models. Suppose that the loss function is square, the convergence rate of simultaneous EB estimator about the parameter θ = (β1,···βp,σ2) is obtained and this convergence rate can be arbitrarily close to 1 under suitable conditions. At last, an example is given.
