The time series data of material demand of manufacturing companies are often non-stationary. Paper uses empirical mode decomposition (EMD) to convert non-stationary time series into a series of intrinsic mode function (IMF) and a residual term (RES), and then digged out more information combined with least squares support vector machine regression (LSSVR) to forecast the model. Finally, the empirical results show that the EMD-LSSVR combination forecast can effectively predict non-stationary material demand time series, and the prediction accuracy is high. It has a certain degree of promotion and practical value.
展开▼
