Multivariate estimation of Poisson parameters (Q2293381)

Let \((\xi_1,\dots, \xi_p)\) be a vector of independent Poisson random variables with mean vector \(\lambda=(\lambda_1,\dots, \lambda_p)\). The problem of estimating the vector \(\lambda\) is considered. The new loss functions are introduced which penalise bad estimates of each of the parameters and also the sum of the parameters. Under the loss functions presented, classes of minimax estimators are derived that uniformly dominate the maximum likelihood estimators.