Computing prediction intervals with CAATTs: Implementation in CaseWare IDEA and complemented in R project software

In Auditing, an expectation can be formed applying methodologically sound analytical procedures to produce a forecast based on past records. Computer-Assisted Audit Tools and Techniques (CAATTs) have an important part in this process, since those methodologies can involve sophisticated statistical analysis and can be time consuming. For audit time series, CaseWare IDEA can be used to compute point forecasts, meaning that a single value is produced for each forecast. Even though the practical usefulness of this type of forecast is undisputed, the uncertainty in the forecasts is not expressed. To overcome this issue, a point forecast should be complemented with its prediction intervals. The R Project software is an interesting statistical tool to consider for this purpose, as it provides quick accurate analysis in a reproducible way. This paper describes how prediction intervals based on regression analysis can be computed in R Project software, for audit series with trend and/or seasonality.