Analysis of multivariate arma processes with non-stationary innovations |
| |
Authors: | M. Shelton Peiris |
| |
Affiliation: | The University of Sydney , N.S.W, 2006, Australia |
| |
Abstract: | The paper considers vector ARMA processes with nonstationary innovations. It is suggested that this class of models provide a very efficient framework for nonstationary problems. A generalization of the Yule-Walker equations relating the underlying process is obtained. Identification procedures are discussed. The associated prediction problem is solved using the Hilbert space approach. |
| |
Keywords: | nonstationary covariance matrix function identification prediction time series Hilbert space innovation |
|