半变系数模型的几乎无偏岭估计探讨  

Almost Unbiased Ridge Estimation of Semi-varying Coefficient Model

作　　者：曹连英[1] 王蕾 张博[1] 徐文科[1] Cao Lianying;Wang Lei;Zhang Bo;Xu Wenke(School of Science,Northeast Forestry University,Harbin 150040,China)

机构地区：[1]东北林业大学理学院

出　　处：《统计与决策》2018年第17期25-28,共4页Statistics & Decision

基　　金：中央高校基本科研业务费专项资金资助项目(2572018BC20);黑龙江省自然科学基金资助项目(C201408)

摘　　要：文章研究了具有多重共线性的半变系数模型的估计问题,基于轮廓最小二乘方法给出了半变系数模型的几乎无偏岭估计。证明几乎无偏岭估计在偏与均方误差意义下优于岭估计,并用数值模拟算例说明了几乎无偏岭估计的有效性和可行性。This paper studies the estimation problem of semi-varying coefficient model with multi-collinearity. Based on profile least square method, the paper gives the almost unbiased ridge estimation of semi-varying coefficient model, and proves that the almost unbiased ridge estimation is superior to ridge estimation in the sense of deviation and mean square error. And finally the paper uses a numerical example to illustrate the validity and feasibility of almost unbiased ridge estimation.

关 键 词：半变系数模型 多重共线性 几乎无偏岭估计 轮廓最小二乘估计 

分 类 号：O212[理学—概率论与数理统计]