在本文中,我们基于指数回归模型,在渐近最小均方误差的准则下,给出了矩估计的门限值和样本点分割的选取原理和方法,利用MC方法,对Burr(1,1,1)、Burr(1,0.5,2)、Fréchet(1)、Fréchet(2)、学生-t4、学生-t6等几种常见的极值分布进行模拟,得到了理想的结果,并运用S&P500指数和Danish火灾数据进行了实证分析。
In this paper, under the prineiple of asympetotic mean squarte error, the method how to select suitable threshold and fracture sample in moment estimator based on exponential regression model is put forward Using MC method, several extreme distribution(Burr( 1,1,1 ), Burr(1,0.5,2) Fréethet(1), Frécthet(2), Student - t4, Student- t6)are simulated and ideal results are attained. In the last of this paper, empirical studies of S&P500 index and Danish fire data are given.