Title: Bayesian matrix regression
Authors: Matteo Iacopini - Ca Foscari University of Venice (Italy) [presenting]
Roberto Casarin - University Ca' Foscari of Venice (Italy)
Monica Billio - University of Venice (Italy)
Abstract: A general model is proposed for linear regression with matrix variate response, which encompasses univariate and multivariate regression as special cases. For dealing with the issue of dimensionality, we exploit a suitable decomposition which enables to achieve both parsimony and to incorporate time-varying sparsity on the coefficients with the additional aim of capturing the change over time of the relevance of covariates. Inference is carried out in the Bayesian framework via Gibbs sampler.