摘要
研究了半参数变系数部分线性测量误差模型的惩罚经验似然问题.首先基于无偏辅助函数,构造了惩罚经验对数似然函数,证明了所得惩罚经验似然估计具有Oracle性质.其次,考虑了关于参数的假设检验问题,给出了检验统计量及其渐近分布.最后通过实验模拟和实例分析验证了所提方法的有限样本性质.
Penalized empirical likelihood for parameter estimation and variable selection in semiparametric varying-coefficient partially linear errors-in-variables models is investigated.It is shown that the penalized empirical likelihood estimators have the oracle property.Also,we conclude that the asymptotic distribution of penalized empirical likelihood ratio test statistic is a chi-square distribution under the null hypothesis.The performance of the proposed method is evaluated by some simulations and a real data example.
作者
陈夏
郭韫华
何军伟
Chen Xia;Guo Yunhua;He Junwei(School of Mathematics and Information Science,Shaanxi Normal University,Xi'an 710119,China)
出处
《纯粹数学与应用数学》
2020年第3期253-269,共17页
Pure and Applied Mathematics
基金
教育部人文社会科学研究青年基金(18YJC910003)
陕西省自然科学基础研究计划项目(2020JM-276).
关键词
半参数变系数部分线性模型
测量误差
变量选择
惩罚经验似然
semiparametric varying-coefficient partially linear model
measurement error
variable selection
penalized empirical likelihood