A multi-steap ahead prediction method based on local dynamic properties

The task of forecasting a time series over a long horizon is commonly tackled b y iterating one-step-ahead predictors.Despite the popularity that this approach gained in the prediction communit y, its design is still plagued by a number of important unresolved issues, the most important being the accumulation of prediction errors. We introduce a local method to learn one-step-ahead predictors with the aim of reducing the propagation of errors during the iteration. For each prediction, our method selects the structure of the local approximator using, in a local version, well-kno wn results of dynamic system theory. Experimental results on tw o time series from the San ta F ecompetition show that the technique is competitive with state-of-the-art forecasting methods.