线性模型的Bootstrap方法

THE BOOTSTRAP METHOD IN THE LINEAR MODEL

本文通过对加权中心化残差重新抽样,建立了异方差线性模型回归系数可估函数的最小二乘估计的Bootstrap估计,证明了由此导出的Bootstrap方差估计具有一定的稳健性,并在一定条件下,得到了Bootstrap估计的相合性。

A bootstrap method in the heteroscedastic linear modal is proposed based on the weighted centrical residuals and the robustness of bootstrap variance estimator is obtained, under the centain conditions, the consistencey of bootstrap estimator is proved.