A Combined Model Based on GM and SARIMA: An Example of Excavator Demand Forecasting

Accurate excavator demand forecasts not only improve the service level of the manufacturer, but also reduce the bullwhip effect throughout the supply chain. In this paper, a new combined forecasting method based on Gray Model (GM) and the Seasonal Autoregressive Integrated Moving Average (SARIMA) model is proposed. In the excavator market, noise signals usually affect prediction accuracy, which were caused by different instability factors. First, two univariate models, SARIMA model and GM (1, 1) model, are used to forecast excavator demand. The precision of the two models are in line with requirements, but the residues of the two models are opposite in a statistical sense. Then a combined model is constructed with these two models, the Differential Evolution Algorithm (DE) is used to optimize the weight coefficients of the two prediction methods of GM and SARIMA. Through comparing, it is found that the fitted values of combined model respond less to the fluctuations and its MAPE (Mean Absolute Percent Error) is smaller than SARIMA model and GM (1, 1) model. And then, China's excavators demand is forecasted by using the three models.