Empirical mode decomposition based support vector regression for agricultural price forecasting

Price information is a crucial market information for a farmer. The price instability and uncertainty pose a significant challenge to decision makers in making proper production and marketing plans to minimize risk. Agricultural price series cannot be modelled and predicted accurately by traditional econometric models owing to its nonlinearity and nonstationary behaviour. In the present study an attempt has been made to model and predict price series using Empirical Mode Decomposition (EMD) based Support Vector Regression (SVR) model. EMD decomposes the original nonlinear and nonstationary dataset into a finite and small number of sub-signals. Then each sub-signal was modelled and forecasted by SVR method. Finally, all the forecasted values of sub-signal were aggregated to make final ensemble forecast. The effectiveness and predictability of the proposed methodology was verified using Chilli wholesale price index (WPI) dataset as sample. The results indicated that the performance of the proposed model was substantially superior as compared to the standard SVR.