ABSTRACT
Within current forecasting practice, the computing facilities used to run macro- and micro-economic models which handle large quantities of statistical data, are pro-grammed using more or less powerful procedural languages. These programs carry out only the calculations involved in the resolution of the models and in some cases provide a preprogrammed presentation of the results. As a consequence, numerous stages within the complex process of forecasting rely on human expertise and have to be carried out manually by users. These stages include the writing of the model itself and all the related operations; the definition of scenarios, necessary for carrying out the various simulations of future events; the analysis of the simulations results; and the comparison of these results with the conclusions which can be drawn from them.
