Combining Genetic Algorithms, Neural Networks and Data Filtering for Time Series Forecasting
José Neves, Paulo Cortez · Portuguese National Funding Agency for Science, Research and Technology (RCAAP Project by FCT) · 1998
In the last few decades an increasing focus as been put over the field of Time Series Forecasting (TSF), the forecast of a time ordered variable. Contributions from the arenas of Operational Research, Statistics, and Computer Science as lead to solid TSF methods (eg. Exponential Smoothing or Regression) that replaced the old fashion ones, which were primary based on intuition. Although these methods give accurate forecasts on linear Time Series (TS), their handicap is with noise or nonlinear components, which is a commum situation (eg. in financial daily TS). An alternative approach for TSF as recently emerged from the field of Artificial Intelligence, where new optimization algorithms, such as Genetic Algorithms and Artificial Neural Networks have became popular. Following this trend, the present work reports on a Genetic Algoritm Neural Network system, and in its use for TSF.