Incorporating Grey Relational Analysis into Grey Prediction Models to Forecast the Demand for Magnesium Materials

Magnesium is a promising light metal that has been widely used to manufacture components for automobiles, bicycles and electronics. By forecasting the demand for magnesium materials, we can recognize its prospects in these industries. Therefore, this study applies the GM(1,1) power model, the most frequently used gray prediction model, to forecast the demand for magnesium materials. Gray prediction is appropriate for this study, because there is little available data on magnesium material demand and it does not coincide with statistical assumptions. In contrast to the original GM(1,1) power model, which simply treats each sample with equal importance, this study uses gray relational analysis to estimate the weight of each sample to improve the prediction accuracy. The forecasting ability of the proposed gray residual modification models was verified using real data regarding magnesium material demand. The results showed that the proposed variant of the GM(1,1) power model offers performance that is comparable to other gray prediction models.