BACKPROPAGATION IN TIME-SERIES FORECASTING

One of the major constraints on the use of backpropagation neural netw orks as a practical forecasting tool is the number of training pattern s needed. We propose a methodology that reduces the data requirements. The general idea is to use the Box-Jenkins model in an exploratory ph ase to identify the 'lag components' of the series, to determine a com pact network structure with one input unit for each lag, and then appl y the validation procedure. This process minimizes the size of the net work and consequently the data required to train the network. The resu lts obtained in eight studies show the potential of the new methodolog y as an alternative to the traditional time-series models.