A Decomposition-And-Ensemble Forecasting Method Based on EMEMD and FWNN Training by PSOSSO

Time series forecasting is an hot research area in signal processing and machine learning. In this area, multi-layer network structure technique is a widely used approach. In this paper, based on an adaptive decomposition method (extrema mean empirical mode decomposition) and fuzzy wavelet neural network, we proposed a decomposition-forecasting-ensemble approach for time series forecasting. Furthermore, the proposed forecasting model is trained by the PSOSSO optimal method. Experimental result shows the forecast result by proposed method is better than other ensemble methods.