scientific article; zbMATH DE number 3909577

From MaRDI portal
Publication:3685891





[For the entire collection see Zbl 0553.00008.] A description is given of a method for estimating an ARMA process, y(t), from observations for \(t=1,2,...,T\), and following this a discussion is given of the theory necessary for the validation of the method. The first stage of the method involves the fitting of an autoregression, of order \(h_ T\) determined by a criterion such as AIC. The asymptotic theory depends on the behaviour of \(h_ T^{-1}\Sigma^ T_ 1\{{\hat \epsilon}(t)-\epsilon (t)\}^ 2\) where \({\hat \epsilon}\)(t) is the residual from the autoregression and \(\epsilon\) (t) is the innovation. Theorems relating to this expression are proved.




Cited in
(69)








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 Q3685891)