删失回归模型的加权Bootstrap逼近  

Weighted Bootstrap Inference on Censored Regression Model

作　　者：赵树然[1] 徐兴忠[1] 任培民[2] 

机构地区：[1]北京理工大学理学院,北京100081 [2]青岛大学经济学院,山东青岛266071

出　　处：《北京理工大学学报》2008年第7期648-651,共4页Transactions of Beijing Institute of Technology

基　　金：国家自然科学基金资助项目(10771015)

摘　　要：研究带删失数据的回归模型.基于合成数据,运用加权Bootstrap方法获得回归系数样本分布的近似估计,并对一般权序列,证明了这种分布近似的有效性及回归系数的加权Bootstrap估计的相合性.据此构造回归系数的加权Bootstrap置信域.数值模拟结果表明,就覆盖率而言,加权Bootstrap置信域优于传统的渐近正态置信域.Regression model with censored data is studied. Based on synthetic data, weighted bootstrap is used to approximate the sample distribution of regression coefficient. For general choices of random weights, the validity of such a distribution approximation and the weak consistency of weighted bootstrap estimate to regression coefficient are established. A weighted bootstrap confidence interval for regression coefficient is then constructed. Simulation studies were conducted to compare the weighted bootstrap interval with the interval based on normal approximation and showed that the former is superior to the latter in terms of coverage probability.

关 键 词：线性回归模型 置信区间 加权Bootstrap推断 合成数据 

分 类 号：O212.1[理学—概率论与数理统计]