Minimax properties for linear estimators of the location parameter of a linear model
Pedro Alson · Statistics · 1988
Minimax properties of the quadratic risk of linear estimators of the location parameter of a linear model are studied. Minimax linear estimators are found for two kinds of sets: Tcubes and ellipsoids centred at the origin. It is seen, for linear estimators, the equivalence between linear admissibility and minimaxity. The results are valid for arbitrary quadratic risk and design matrix