Abstract：In this paper, a new biased estimate is proposed for the regression coefficients of multivariate linear models, combining the technique of least squares estimate, principal components estimate and combinatorial principal components estimate, in order to improve least squares estimate. It is shown that the new estimate superior to least squares estimate, principal components estimate and combinatorial principal components estimate under suitable conditions, and is admissible estimate.
收稿日期: 2003-02-25
引用本文:
蔡新民,黄养新. 回归系数的一种有偏估计[J]. , 2003, 42(2): 0-0.
蔡新民,黄养新. A biased estimate for regression coefficients. , 2003, 42(2): 0-0.
