Bayesian prediction of matrix elliptical multivariate models with conjugate prior

Arashi, Mohamed and Khan, Shahjahan (2012) Bayesian prediction of matrix elliptical multivariate models with conjugate prior. In: 11th Islamic Countries Conference on Statistical Sciences (ICCSS 2011), 19-22 Dec 2011, Lahore, Pakistan.


This paper considers a multivariate regression model with a matrix variate elliptically contoured (MEC) distribution for the responses. The MEC is defined as a mixture distribution of inverse Laplace transformation and matrix variate normal distribution. The prediction distribution of a set of matrix future responses from the same model with common indexing parameters is obtained under Bayesian framework with the conjugate prior of the parameters. The prediction distribution is found to be a matrix-T distribution. The results of the paper is a generalization of earlier results for linear models in terms of the generalization of the (i) multiple regression model, (ii) normal or Student-t distribution, and (iii) joint conjugate prior distribution.

Statistics for USQ ePrint 21272
Statistics for this ePrint Item
Item Type: Conference or Workshop Item (Commonwealth Reporting Category E) (Paper)
Refereed: Yes
Item Status: Live Archive
Additional Information: © 2012 Islamic Countries Society of Statistical Sciences. Permanent restricted access to published version due to publisher copyright policy.
Faculty / Department / School: Historic - Faculty of Sciences - Department of Maths and Computing
Date Deposited: 27 Apr 2013 02:45
Last Modified: 02 May 2017 23:34
Uncontrolled Keywords: multivariate regression model; elliptically contoured distribution; inverse Laplace transform; conjugate prior; matrix-T distribution; prediction
Fields of Research : 01 Mathematical Sciences > 0104 Statistics > 010405 Statistical Theory
01 Mathematical Sciences > 0104 Statistics > 010404 Probability Theory
Socio-Economic Objective: E Expanding Knowledge > 97 Expanding Knowledge > 970101 Expanding Knowledge in the Mathematical Sciences

Actions (login required)

View Item Archive Repository Staff Only