| 复共线性与广义岭型估计 |
| |
| 引用本文： | 赵海清,詹环.复共线性与广义岭型估计[J].大学数学,2009,25(3). |
| |
| 作者姓名： | 赵海清 詹环 |
| |
| 作者单位： | 1. 湛江师范学院,数学与计算科学学院,广东,湛江,524088
2. 装甲兵工程学院,数学教研室,北京,100072 |
| |
| 摘 要： | 针对线性回归模型Y=Xβ+l的典则形式Y=a01+Z+l,l~(0,σ2I)在设计阵X呈病态时,提出了一类新估计■(k;q)=〔Λ1OOkIq+Λ2〕-1Z′Y,称之为广义岭型估计.优点是结合主成分估计和岭估计的思想和方法,将X′X的特征值分为不同大小属性的两部分Λ1与Λ2,并分别添加不同的常数,致使新估计类的均方误差大幅降低的同时计算量大大减少,而且便于对原变量做出解释.文中进一步讨论了该估计优于岭估计的k的存在性以及充分条件.
|
| 关 键 词： | 典则形式 岭估计 主成分估计 广义岭型估计 |
Approximate Multicollinearity and Generalized Ridge Estimation |
| |
| ZHAO Hai-qing,ZHAN Huan.Approximate Multicollinearity and Generalized Ridge Estimation[J].College Mathematics,2009,25(3). |
| |
| Authors: | ZHAO Hai-qing ZHAN Huan |
| |
| Institution: | 1.College of Maths;Zhanjiang Normal University;Zhanjiang 524088;China;2.Department of Mathematics Academy of Armored Froce Engineering;Beijing 100072;China |
| |
| Abstract: | Considering the linear regression model,Y=a01+Z+l,l～(0,σ2I),in terms of the approximate multicollinearity of the original data matrix,this paper gains the generalized ridge estimation of the parameter ,named ■(k;q),with the ideas and methods of the principal components and the ridge estimation,its advantage is lower calculation and is easy to explain the model.Then it further discusses the sufficient condition of ■(k;q) superior to ■(k). |
| |
| Keywords: | ridge estimation principal components estimation generalized ridge estimation |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
