Weibull 分布における未来標本の最尤予測量について

概要

論文の詳細を見る
This paper describes the results of a study of the maximum likelihood predictor of future order statics for two sample problems, on the basis of Type II censored samples when the two-parameter Weibull distribution is assumed. The maximum likelihood predictors obtained are compared with the best linear invariant predictors and the statistics which are given by transforming exponentially the best linear invariant predictors for the extreme-value distribution. The mean square errors of those statistics are computed by Monte Carlo simulations for some sample sizes. The explicit expression of the maximum likelihood predictor on the basis of type II censored samples is not given, but the advantage of this predictor is that it does not involve the moments of the Weibull or extreme-value order statitics.
湘南工科大学の論文
1991-03-31