Akaike 資訊準則的有限樣本修正

Abstract

在這篇論文裡，我們考慮 Akaike 資訊準則的有限樣本修正。我們 把最大概似函數的 漸近誤差計算到o(1/n)的收斂速度用來做 Akaike 資 訊準則的有限樣本修正。另外我們也 導出了收斂速度為o(1/n)的最大概 似估計量的漸近誤差。 In this paper we consider the finite correction of Akaike's information crit erion and refine it by considering the asymptotic bias of the maximumlikelihoo d with the rate of convergence is o(1/n). The asymptotic bias of the ML estima tors whose rate of convergenceis o(1/n) are also derived.