Statistics Seminar 27th October 2004

VARIABLE SELECTION AND COVARIANCE SELECTION IN MULTIVARIATE REGRESSION MODELS

Speaker: Ed Cripps
Department of Statistics
University of New South Wales

Time:4:00p.m. Wednesday 27th October 2004
Venue: Red Centre Building Room RC-3084
near Barker Street Gate 14

This talk outlines a general framework for Bayesian variable selection and covariance selection in a multivariate regression model with Gaussia errors. By variable selection we mean allowing certain regression coefficients to be zero. By covariance selection we mean allowing certain elements of the inverse covariance
matrix to be zero. All the model parameters are estimated by model averaging

using a Markov chain Monte Carlo simulation method. The effectiveness of variable selection and covariance selection in estimating the multivariate regression model is assessed by using four loss and four simulated data sets. Each of the simulated data sets is based on parameter estimates obtained from a corresponding real data set.

Y. Fan
E-mail yanan@maths.unsw.edu.au
You are invited to join us for an afternoon tea in the staff common, Red Centre Building RC-3082 at 3:45pm.