Bayesian Inference on the Fréchet-type Extreme Value Distribution based on the Characteristic Prior

Abstract

Abstract In this paper, a new method has been used for constructing a prior density function based on the characteristic function, which frees the Bayesian inference from subjectivity to a personal choice that has worried some statisticians. For comparing the characteristic prior with the informative prior, Bayesian inference on the Fréchet-type extreme value distribution parameters has been studied based on symmetric and asymmetric loss functions, via Monte Carlo simulations. The simulation results indicated that the characteristic prior outperformed the informative prior for different sample sizes and several values of the parameters. Finally, a numerical example is given to demonstrate the efficiency of the proposed priors.