ABSTRACT
Example: AR(1) For the first order autoregressive model, p = 1 and q — 0 so that m = 1 and the DLM representation is
which is precisely the infinite moving average form of the AR(1) model.
Autoregressive models may be expressed in an alternative (perhaps more intuitive) form, namely
In this form it is easier to observe that forecasting more than one step ahead requires future values of the series being forecasted. The need for knowledge of future values of series is a feature of all regression component DLMs when predicting step ahead. Practical approaches to the problem are discussed in Chapter 5.
