几类数据下部分线性单指标模型的统计推断

发布时间：2018-03-25 13:48

本文选题：部分线性单指标模型　切入点：借补估计　出处：《吉林大学》2016年博士论文

【摘要】：本文主要研究几类数据下部分线性单指标模型的统计推断问题.首先,为了处理实际应用中常见的数据缺失和观测不准确的问题,考虑了响应变量缺失下部分线性单指标EV模型的借补估计,并给出了相应估计量的渐近分布.通过模拟对比研究了完全估计法和借补估计法的估计效果,并将估计方法应用于一组糖尿病患者疾病进展情况数据的实证分析当中.其次,基于纠偏的思想,给出了响应变量缺失下部分线性单指标EV模型中参数向量的经验似然推断,并通过数值模拟研究了纠偏经验似然方法的精确性和稳健性等问题.最后,基于分组经验似然方法研究了NA误差下部分线性单指标模型参数向量的经验似然推断问题,并进行了模拟研究.
[Abstract]:This paper mainly studies the statistical inference problem of partial linear single index model under several kinds of data. Firstly, in order to deal with the common problems of missing data and inaccurate observation in practical application, In this paper, the complementary estimation of partial linear single-index EV model with missing response variables is considered, and the asymptotic distribution of the corresponding estimator is given. The effects of the complete estimation method and the complementary estimation method are studied by simulation and comparison. The estimation method is applied to the empirical analysis of a group of diabetes patients' disease progression data. Secondly, based on the idea of correction, the empirical likelihood inference of parameter vectors in partial linear single index EV model with missing response variables is given. The accuracy and robustness of the empirical likelihood method are studied by numerical simulation. Finally, the empirical likelihood inference problem of parameter vector of partial linear single parameter model under na error is studied based on grouping empirical likelihood method. A simulation study was carried out.
【学位授予单位】：吉林大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：O212.1

【相似文献】