标题:
基于改进的中位数绝对偏差稳健尺度估计Robust Scale Estimation Based on the Improved Median Absolute Deviations
作者:
杨苹莉
关键字:
尺度估计, 稳健性, 得分函数Scale Estimation, Robustness, Score Function
期刊名称:
《Statistics and Application》, Vol.4 No.2, 2015-06-29
摘要:
本文基于Smirnor-Shevlyakov在2014年针对位置参数已知为0的稳健尺度估计(即改进的中位数绝对偏差FQn)，提出了位置参数未知时的稳健尺度估计(称之为广义中位数绝对偏差GMAD)。数据分析表明：FQn在位置参数未知时不稳健，但GMAD估计在位置参数为0以及未知时均稳健。Robust scale estimation with unknown location parameters which is called general median absolute deviations (GMAD) was proposed based on a robust scale estimation with location parameters of 0 (improved median absolute deviations FQn) given by Smirnor-Shevlyakov in 2014. The data analysis showed that FQn loses robustness when location parameters are unknown, but GMAD is robust when location parameters are zero or unknown.