In order to ensure a well-founded production planning and scheduling, a company has to predict its future sales precisely. However, customer demands are often highly volatile. Classical statistical prediction methods are often easy to apply but they are not able to react on dynamic structures within order data. In contrast, methods of nonlinear dynamics consider qualitative information in addition to quantitative, in order to identify possible deterministic structures within the time series and thus to achieve better forecasts into the future. In this project, methods of nonlinear dynamics were adapted to forecast regular as well as intermittent time series of customer demands. Evaluation studies showed particular potential of these methods to forecast complex, dynamic time series.
Contact persons:
S. Oelker
Funded by:
DFG
Duration:
01.04.2011 - 30.09.2012
See project's publications
List all projects
