竞争风险混合模型的参数估计与检验  

Estimation and Testing for Competing Risks Mixture Model

作　　者：胡涛[1] 崔恒建[1] 宋立新[2] 

机构地区：[1]北京师范大学统计与金融数学系,北京100875 [2]大连理工大学应用数学系,大连116024

出　　处：《应用数学学报》2007年第4期577-591,共15页Acta Mathematicae Applicatae Sinica

基　　金：国家自然科学基金重点项目(10231030号)和面上项目(10271049号)资助.

摘　　要：本文在独立同分布I型区间删失情形下,研究了竞争风险混合模型中当参数真值是内点时,参数极大似然估计的性质,获得了其强相合性和渐近正态性.在较为宽松的条件下,给出了竞争风险混合模型参数序关系假设检验的检验方法,同时得到了似然比检验统计量及其在零假设下的渐近分布为加权x^2分布,并给出了—个例子并进行了功效比较.In this paper, we study some sis testing in the competing risks mixture asymptotics of MLE for parameters and hypothemodel with case 1 interval censoring data. Under some proper conditions, we show that MLE is strong consistent and asymptotically normal. Furthermore, for the problem of testing homogeneity of parameters against the simple order restriction, the likelihood ratio test statistic for the testing problem is given and its asymptotic null distribution is derived that is weighted x^2-distribution. Some power comparisons are presented for an example.

关 键 词：竞争风险 混合模型 Ⅰ型区间删失 极大似然估计 渐近性质 

分 类 号：O212.7[理学—概率论与数理统计]