M-估计下误差密度核估计的相合性  

Convergence for Kernel Estimation of Error Density in M-Estimator

作　　者：刘东海[1] 缪柏其[2] 彭衡[2] 

机构地区：[1]中国人民武装警察部队学院指挥系,廊坊065000 [2]中国科学技术大学统计与金融系,合肥230026

出　　处：《应用概率统计》2002年第2期125-133,共9页Chinese Journal of Applied Probability and Statistics

摘　　要：线性模型Yi=x1iβ+ei,i=1,2,…,其中{ei}i∞=1 i.i．d.,有未知密度f(x)．本文讨论了在对β作一般M－估计后,基于残差做出的误差密度核估计的相合性．在比文献弱的条件下,证明了误差密度核估计的逐点弱相合、逐点强相合和一致强相合．Linear lnodel Yi= xi1 β+ ei, i = 1, 2,…, where {ei}i∞=1, i.i.d. with unknown density f(x). In this paper, weconsider some converegence of kernel estimation of error density based on the residuals of M-estimation. Underweaker assumptions than references, the article proves the weak local convergence, the strong loca1 convergence and the strong uniform convergence of kernel estimation of error density.

关 键 词：M-估计 残差 核估计 相合性 

分 类 号：O212.7[理学—概率论与数理统计]