Utility Package

The CLIMITS object provides a mechanism for computing both the prediction standard errors and confidence limits of an external model (that is, a user-defined model) forecast, which is described by a pair of actual and predicted time series. The first stage of the computational process involves validating both input series. This validation is accomplished by ascertaining that the actual and predicted series have nonmissing observation values under at least one matching index. In addition, the predicted series is checked for the presence of extreme values. Next, the prediction standard errors are computed from the prediction errors (that is, the model residuals). Finally, the confidence limits are computed from the prediction standard errors. You can optionally supply the value of the confidence level that is used to compute the confidence limits. Note that the CLIMITS object retains all computed results in its internal memory. Individual forecast series can be queried via the GetForecast method, which is a method in the CLIMITS object.

Table 4 summarizes the methods that are associated with the CLIMITS object.