回归系数的根方型主成分估计

在线性回归模型中，当自变量间存在复共线性时，回归系数的最小二乘估计就失去了它的优良性，而主成分估计和根方估计都具有抗复共线性的特性，本文将二者有机结台，保留它们各自的优点，提出了根方型主成分估计，并证明了当复共线性存在时，根方型主成分估计优于根方估计、主成分估计和最小二乘估计，通过实例分析，说明它具有一定的实用价值。

Combining Root Root and Principal Components Estimates of Regression Coefficient

In linear regression model, the LS estimate of the regression coefficient will lost its optimal property as the predictor varibles have multicollinearity. Both Principal Component estimate and Root Root estimate can resist the multicollinearty. In this paper, we combine the two estimates together and propese a new estimate Root Root Principal Component (RRPC) estimate which retained the optimal properties of Principal Component estimate and Root Root estimate, moreover, we prove that RRPC estimate improves Principal Components estimate, Root Root estimate and LS estimate. Also we give a practical example which can explain the RRPC estimate has the great value of utility.