The predictive accuracy A is a stronger criterion than R since it requires that not only the relative ordering of the target field values but also their absolute size is correctly modeled. A typical use scenario in which this quality measure is important is again a direct marketing campaign. This time, the aim of the campaign is to generate a given amount X of additional income. The task is to select a sufficient number of the most promising customers so that the planned income will (probably) be reached. Here, it is inevitable that the model predicts not only the ordering of the predicted fields but also their absolute values.
You can retrieve the predictive accuracy A of a Regression model by using the function DM_getPredAccuracy. The quality is indicated by a number between 0 and 1.