Combination of forecasting using modified GMDH and genetic algorithm

Abstract

Many studies have demonstrated that combining forecasts improves accuracy relative to individual forecasts. In this paper, the combing forecasts is used to improve on individual forecasts is investigated. A combining approach based on the modified Group Method Data Handling (GMDH) method and genetic algorithm (GA), is called as the GAGMDH model is proposed. Four time series forecasting techniques are used as individual forecast, namely linear regression, quadratic regression, exponential smoothing and ARIMA models. The forecasted results of individual forecasting models are used as the input of combining forecasting, and the outputs are the results of combination forecasting. To assess the effectiveness of the GAGMDH model, we used the time series yearly cancer death rate in Pennsylvania. The empirical results with a real data set clearly suggest that the GAGMDH model can improve the forecasting capability of the model compared with optimal simple combining forecasting methods and neural networks combining forecasting methods