DefinePK

Abstract

Hybridization of existing competitive modeling methodologies is now an active area of research. The GMDH algorithm is a heuristic and computer oriented method which provides the foundation for the construction of high order regression models of complex system. The research for improving the effectiveness of forecasting models has never been stopped. Currently it was reported that a hybrid system in prediction and classification achieved a higher performance level against the traditional system. The selection of the forecasting model is the important criteria that will influence to the forecasting accuracy. So the enhancement of conventional GMDH model through hybridization will improve the prediction accuracy of the traditional GMDH for time series forecasting. This paper presents a short overview of Group Method of Data Handling (GMDH), its modification and hybridization for time series forecasting. The overview will aim to provide further investigation on the hybrid Group Method of Data Handling (GMDH) with the Wavelet Decomposition for Time Series Forecasting. This modification and hybridization will be proposed a new hybrid GMDH that will be helped for the computational constrain to become more flexible as well as robust/efficient than the conventional GMDH.