摘要
小样本情形下,样本均值易受异常点的影响,导致经典方法得到的置信区间不具有稳健性.文章合理改进了PWM方法,并借助一般的和加权处理的Bootstrap方法进行区间估计比较.结果显示改进的PWM方法得到的置信区间较短,而真值覆盖率更高,并且具有更好的稳健性.
The sample mean is vulnerable to the value of abnormal points in the case of the small samples,that makes the confidence interval got by the classical method has non-robustness.This paper improves the method of PWM reasonably,and adopts the general Bootstrap sampling methods and the weighted one to compare the interval estimations.The result shows that the method of improved PWM has shorter confidence interval with higher accuracy and much more robustness.
出处
《曲阜师范大学学报（自然科学版）》
CAS
2012年第4期25-27,31,共4页
Journal of Qufu Normal University(Natural Science)
基金
江苏省水利科技创新基金(2011059)
河海大学自然科学基金(2009426311)
