统计与决策2025，Vol.41Issue(8)：53-58,6.
正态分布混合线性专家模型的稳健估计
Robust Estimation of Mixture of Linear Expert Model With Normal Distribution
摘要
Abstract
Mixture of linear expert models with error terms following normal distribution(NMoLE)are very sensitive to outli-ers or heavy-tailed data.In view of this issue,the paper proposes a robust estimation method:Firstly,M estimation is introduced into the log-likelihood function of NMoLE model,and then a robust EM algorithm is proposed.Monte Carlo simulation results show that the proposed robust estimation is robust to the data with outliers,and is better than the mixture of linear expert model with error term following t-distribution and maximum likelihood estimation.If there are no outliers in the data,the results of robust estimation are not much different from those of the mixture of linear expert model with error term following t-distribution and max-imum likelihood estimation.In addition,this paper successfully uses real data to illustrate the robustness of the proposed estima-tion method.关键词
混合线性专家模型/NMoLE模型/M估计Key words
mixture of linear expert model/NMoLE model/M estimation分类
数理科学引用本文复制引用
孙明云,李龙..正态分布混合线性专家模型的稳健估计[J].统计与决策,2025,41(8):53-58,6.基金项目
湖南省自然科学基金资助项目(2025JJ70187) （2025JJ70187）
湖南省教育厅科学研究一般项目(22C0403) （22C0403）
湖南省科技计划项目资助(2016TP1020) （2016TP1020）
衡阳师范学院智能信息处理与应用湖南省重点实验室开放基金资助项目(2022HSK-FJJ014) （2022HSK-FJJ014）
