非参数bootstrap方法及其应用  

Non-Prarmetric Bootstrap Method for Validation Stabilities of Parameters of Statistical Model

作　　者：孔丹莉[1] 丁元林[1] 

机构地区：[1]广东医学院预防医学教研室,湛江524023

出　　处：《数理医药学杂志》2006年第3期232-233,共2页Journal of Mathematical Medicine

摘　　要：目的:介绍非参数bootstrap方法在验证统计模型参数估计值稳定性方面的应用。方法:采用非参数bootstrap方法验证累积比数L og istic回归模型参数估计值的稳定性。结果:累积比数L og istic回归模型中6个因素的θ∧-β∧/S∧θ均小于25%,可认为模型参数估计值的稳定性较好。结论:采用非参数bootstrap方法验证统计模型参数估计值的稳定性值得推广应用。Objective：This paper introduced non-parametric bootstrap method for validating stabilities of parameters of statistical model. Medthod： stabilites of parameters of cumulative-odds logistic regression were validated. Result： stabilities of parameters of six factors included in cumulative-odds logistic regression were better because their values of ｜↑^θ-↑^｜/↑^Sθ were all smaller than 25% . Conclusion： non-parametric bootstrap method for validating stabilites of parameters of statistical model is good for popularization.

关 键 词：非参数bootstrap方法 累积比数Logistic回归模型 稳定性 

分 类 号：R311[医药卫生—基础医学]