Parameter Estimation for Extreme-value Distribution Based on the p_i-th Quantiles of Samples

The asymptotic normal and asymptotic unbiased estimation of distribution parameter and scale parameter is proposed by linear regression model, based on k P_i-th quantiles of extreme-value distribution simple sample. The asymptotic confidence estimation of the parameter is given for extreme-value distribution. The Weibull distribution of shape parameter and scale parameter is turned into an extreme-value distribution. And thus the asymptotic confidence estimation for the Weibull distribution of this kind of parameters is also obtaind.