样本函数条件极值中减低偏差的方法

BIAS REDUCTION METHODS FOR THE CONDITIONAL EXTREMA OF SAMPLE FUNCTIONS

对样本函数条件极值中偏差项的阶进行了分析,探讨了减低偏差项的方法,分析表明古典折刀法、减-d折刀法均不能减低偏差项;在此基础上,提出了减低偏差项的自助法,并论证了在均方误差意义下,θ_(nab)是一种较优的估计.

The order of bias term in the conditional extrema of sample functions is analyzed and some methods for bias reduction are discussed in this paper. It comes to a conclusion that it is not enough for bias reduction while using the classical jackknife or delete-d jackknife method. On the basis of that, a bootstrapping method for bias reduction is mentioned and it is proven that θnab is a better estimator in the sense of mse.