English

On predictive density estimation with additional information

Statistics Theory 2017-09-25 v1 Methodology Statistics Theory

Abstract

Based on independently distributed X1Np(θ1,σ12Ip)X_1 \sim N_p(\theta_1, \sigma^2_1 I_p) and X2Np(θ2,σ22Ip)X_2 \sim N_p(\theta_2, \sigma^2_2 I_p), we consider the efficiency of various predictive density estimators for Y1Np(θ1,σY2Ip)Y_1 \sim N_p(\theta_1, \sigma^2_Y I_p), with the additional information θ1θ2A\theta_1 - \theta_2 \in A and known σ12,σ22,σY2\sigma^2_1, \sigma^2_2, \sigma^2_Y. We provide improvements on benchmark predictive densities such as plug-in, the maximum likelihood, and the minimum risk equivariant predictive densities. Dominance results are obtained for α\alpha-divergence losses and include Bayesian improvements for reverse Kullback-Leibler loss, and Kullback-Leibler (KL) loss in the univariate case (p=1p=1). An ensemble of techniques are exploited, including variance expansion (for KL loss), point estimation duality, and concave inequalities. Representations for Bayesian predictive densities, and in particular for q^πU,A\hat{q}_{\pi_{U,A}} associated with a uniform prior for θ=(θ1,θ2)\theta=(\theta_1, \theta_2) truncated to {θR2p:θ1θ2A}\{\theta \in \mathbb{R}^{2p}: \theta_1 - \theta_2 \in A \}, are established and are used for the Bayesian dominance findings. Finally and interestingly, these Bayesian predictive densities also relate to skew-normal distributions, as well as new forms of such distributions.

Keywords

Cite

@article{arxiv.1709.07778,
  title  = {On predictive density estimation with additional information},
  author = {Éric Marchand and Abdolnasser Sadeghkhani},
  journal= {arXiv preprint arXiv:1709.07778},
  year   = {2017}
}

Comments

30 pages, 4 Figures