A data mining approach to time series modelling and forecasting

This paper describes some of the relatively new techniques that can be used to create time series forecasting models. Applicability of techniques are demonstrated on a flood-forecasting study orientated towards system identification and modelling of Vltava river system. The main objective of the study was to design an easy-to-run and fast, but still accurate, model of the entire Vltava river system. The principal users of the model will be dispatchers from Vltava Water Board who use this model primarily under highly stressful conditions of flood control and protection of the Czech capital - Prague.