An Ensemble Model with Grey Clustering for Hog Price Prediction

摘要

An accurate hog price prediction can be quite helpful for agricultural participants and administrative authorities. A novel hog price ensemble forecasting model is proposed and demonstrated with a practical case study in this paper First, we use the ensemble empirical mode decomposition (EEMD) algorithm to decompose the original hog price series into several sub-series and one residual. Second, the grey clustering approach is introduced to reconstruct the sub-series, in order to obtain more definite economic implications of the series as well as to reduce the difficulties of forecast modeling. Third, the support vector machine (SVM) is employed to generate individual forecasts of all the reconstructed series and the final prediction output is the sum of the individual results. Using the hog price of the wholesale market in China as sample data, the empirical results show an encouraging finding that the proposed model has a favorable forecast performance compared with several benchmark models, in terms of mean absolute error (MAE), root mean squared error (RMSE) and mean absolute percent error (MAPE) criteria. Furthermore, with the high-, middle- and low-frequency series generated by the grey clustering approach, the pattern and mechanism of price volatility can be better understood. Among all the factors which influence hog price movement, macroeconomic development is the most vital one.