One of the primary objectives of building a model for a time series is to be able to forecast the values for that series at future times. Of equal importance is the assessment of the precision of those forecasts. In this chapter, we shall consider the calculation of forecasts and their properties for both deterministic trend models and ARIMA models. Forecasts for models that combine deterministic trends with ARIMA stochastic components are considered also.
For the most part, we shall assume that the model is known exactly, including specific values for all the parameters. Although this is never true in practice, the use of estimated parameters for large sample sizes does not seriously affect the results.
KeywordsLead Time Forecast Error Prediction Interval Exponentially Weighted Move Average ARIMA Model
Unable to display preview. Download preview PDF.