Due to dynamics and complexity within production and delivery networks, customer demands are often highly volatile. In order to achieve a well-founded production planning and control, future customer demands have to be predicted precisely. Classical statistical forecasting methods are often easy to apply but are not able to react on dynamic effects within the data. Methods of nonlinear dynamics consider qualitative in addition to quantitative information within past order data to find possible deterministic structures and, as a result, to achieve better forecasts of the future. This article deals with the development of a data base containing recommendations to choose suitable prediction methods in different situations.
