摘要
Hoed和Kennard提出了广义岭估计[1].本文提出了另一类广义岭估计:β(K)=(X’X+K)-1X’Y,K=di-ag(K1,K2,…Kp),它是岭估计的自然推广,但和前者的广义岭估计不同.证明了存在K>0使β(K)的均方误差小于最小二乘估计β(X’X)-1X’Y的均方误差,并且β(K)是β的可容许估计.
The new class estimators of β are proposed: β(K) = (X'X + K)-1 X'Y, K = diag( K1, K2,… Kp) ,and it is natural generalization of kidge estimutor. But it is different from the generalized kidge estimute by Hoerl and Kennard putting forward. It is proved that there is a K > 0 to make less MSE ofβ( K) than MSE of J3 . Besides β(K) is an admissible estimate of β.
出处
《河南教育学院学报（自然科学版）》
2003年第1期1-2,共2页
Journal of Henan Institute of Education(Natural Science Edition)
