A hybrid approach based on saddlepoint and importance sampling methods for bootstrap tail probability estimation

被引：0
作者：
Stephen M. S. Lee
Irene O. L. Wong
机构：
来源：
Statistics and Computing | 2002年 / 12卷
关键词：
bootstrap; importance sampling; M-estimator; saddlepoint; tail probability;
D O I：
暂无
中图分类号：
学科分类号：
摘要：
We propose a simple hybrid method which makes use of both saddlepoint and importance sampling techniques to approximate the bootstrap tail probability of an M-estimator. The method does not rely on explicit formula of the Lugannani-Rice type, and is computationally more efficient than both uniform bootstrap sampling and importance resampling suggested in earlier literature. The method is also applied to construct confidence intervals for smooth functions of M-estimands.
引用
收藏
页码：209 / 217
页数：8
相关论文
共 16 条
[1]  
Chen Z.(1994)The bootstrap methods with saddlepoint approximations and importance resampling Statistica Sinica 4 407-421
[2]  
Do K.-A.(1988)Saddlepoint approximations in resampling methods Biometrika 75 417-431
[3]  
Davison A.C.(1991)On importance resampling for the bootstrap Biometrika 78 161-167
[4]  
Hinkley D.V.(1979)Bootstrap methods: Another look at the jackknife Annals of Statistics 7 1-26
[5]  
Do K.-A.(1987)Better bootstrap confidence intervals (with Discussion) Journal of the American Statistical Association 82 171-200
[6]  
Hall P.(1996)General saddlepoint approximations of marginal densities and tail probabilities Journal of the American Statistical Association 91 666-673
[7]  
Efron B.(1988)Importance sampling for bootstrap confidence intervals Journal of the American Statistical Association 83 709-714
[8]  
Efron B.(1980)Saddlepoint approximations for the distribution of the sum of independent random variables Advances in Applied Probability 12 475-490
[9]  
Gatto R.(1990)Small-sample confidence intervals Journal of the American Statistical Association 85 427-434
[10]  
Ronchetti E.(1990)Saddlepoint approximations in resampling analysis Annals of the Institute of Statistical Mathematics 42 115-131
12