Presented at CSNSS'03.

SUMMARY

The better the model, the more features of the problem it explains. However, showing that the model has similarities to that of a phenomena is often less significant in applications due to lack of data. Forecasting, as special application of modelling, is neither an exception: besides statistical data one should use several types of subjective assumptions about the present and the future state of the model. In case of complex models, this fact is extremely important, because these models use often unobservable, hidden or - regarding its future evolution - uncertain variables. We developed a simple mathematical approach how these uncertainties can be managed in the model. We shall also show how these uncertainties can influence the behaviour of modelled variables, and how an approximate for time horizon of forecasts can be calculated.

KEY WORDS

complex systems, futures studies, foresight, modelling, time horizon

CLASSIFICATION