Linear Models, Time Series and Outliers. 3. Stochastic Difference Equation Models.

摘要

There are clear cut cases in which models involving time are deterministic since they describe some known physical phenomenon which is a function of time. In other instances, there is properly speaking no fixed deterministic component but time series can be modeled by stochastic difference models. In this report it is shown how the need for a deterministic component may be exposed by the analysis itself. A suitably general framework is provided by stochastic difference equation models so that deterministic components are automatically accommodated.