scientific article; zbMATH DE number 4020269

From MaRDI portal
Publication:3763443





A new method based on spectral estimation is proposed for the identification and estimation of univariate ARMA-models. The method maximizes the entropy of the estimated innovation spectrum. On the other hand, it is possible to estimate the spectral density of the disturbance of a linear transfer function model without making any assumptions about the transfer function forms. This enables one to start the identification of a transfer function model by first estimating an ARMA-model for the disturbance. At the next stage, one can derive asymptotically efficient estimates for all the linear parameters in the model. On the basis of these efficient estimates, the specification of suitable transfer function forms is relatively straightforward.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3763443)