The Minimax Estimation of Corresponding Risk of Statistical Model
Xiaonan Xiao · Journal of Xiamen University · 2008
Concerning the risk estimation,Bunke discussed the minimax of the best linear unbiassed estimator of a multiparameter controled statistical model which has the quadratic loss function with the positive definite matrix.But to the unjustifiable weighting that has large estimation errors on the corresponding risk shis quadratic loss function is not applicable.So the present article proposed a more ideal quadratic loss function,and made the minimax estimators and comparisons on the correlated risk.The discussion shows that the proposed quadratie loss function is rational and applicable.