当设计矩阵X复共线时,对齐次线性约束回归模型参数的约束最小二乘估计进行改进,提出参数的主成分压缩估计,并对新参数估计的性质进行了讨论,最后进行了数值模拟,验证了算法的参数估计优于约束最小二乘估计.
If design matrix Xis multicollinear in homogeneous linearly constrained regressionmodel, proposes the principal component shrinkage estimates of the parameters to improvethe constrained least squares estimates of parameters. The excellent statistical properties ofnew parameters estimators are discussed. Finally, Numerical simulation is done. It verifiesthat the principal component shrinkage estimate is more excellent than the constrained leastsquares estimate.