Base a EMD-Grey Model for Textile Export Time Series Prediction

Abstract

The grey forecasting model has been used widely. This paper proposes a hybrid method that combines the GM (1,1) grey forecasting model and the EMD method to forecast textile export time series with seasonality characteristics. Empirical mode decomposition is good at dealing with nonlinear, non-stationary signal and GM (1,1) model can be used to prediction time series. Through the empirical mode decomposition, the textile export series is decomposed into several intrinsic mode function component and a trend component, then using GM (1,1) model to predict these components, and reconstruct the final prediction results. The time series data of the total number of textile export (2003 to 2011) are used as test data sets. Results show that the method has better prediction effect compared with the direct prediction method.