A Comparison of Standard Methods with a Neural Network Approach to Forecast Univariate Time Series
Karl A. Krycha · Institutional Repository (IHS Vienna) · 1992
Abstract: The following paper tries to develop a simple neural network approach to analyse time series data. After an introductional part some standard approaches are presented. A neural network is developed and results on some different data sets are compared. It will be shown that even very simple and rather small network topologies may capture in a very efficient way time series characteristics.;