修正估計函數

Abstract

Firth 在 1993 年提出一個修正的計分函數藉以移除最大概似估計的一階漸近偏誤。在本篇論文中，我們利用他的方法求出一個修正的估計函數以降低原始估計的均方差。然後，再以廣義最小平方估計法為例來說明修正估計函數的用途。

Firth (1993) proposed a modified score function to remove the first-order term of the asymptotic bias of the maximum likelihood estimator. In this paper, we utilize his method to propose a modified estimating function to reduce the asymptotic mean squared error of the original estimator. Then we give an example from the generalized least squares method to illustrate the usefulness of this theory.