线性模型的岭型广义逆估计

Ridge type-generalized inverse regression estimator in linear model

从LS估计出发 ,基于设计阵的奇异值分解式 ,应用岭估计方法 ,构造了回归系数的一个新的有偏估计———岭型广义逆估计 .计算结果表明 ,在设计阵呈病态时 ,岭型广义逆估计确能明显改善LS估计 .

Starting from the LS regression estimation,a new biased estimator of regression coefficients called ridge type generalized inverse regression estimator is generated combining the singular value decomposition expression of the design matrix and the ordinary ridge regression estimation methods.The computational results demonstrate that the RGIR estimator can improve the properties of LS regression estimator when the design matrix is ill conditioned.