Minimax estimation of a bivariate cumulative distribution function (Q2189751)

The authors consider the problem of estimating a bivariate cumulative distribution function \(F\) under the weighted squared error loss and the weighted Cramer-von Mises loss. Estimators, which are minimax among procedures being affine transformation of the bivariate empirical distribution function, are found. A decision rule \(\hat{F}_A\in D_A\) is a minimax affine estimator of \(F\) if \(\sup_F R(\hat{F}_A,F)=\inf_{\hat{F}\in D_A}\) \(\sup_F R(\hat{F},F):=\rho_A\) where \(D_A\) is a set of affine estimators, \(R\) is a risk function, i. e. expected loss function, and \(\rho_A\) is called a minimax affine risk. Extensions to higher dimensions are briefly discussed.