Forecasting Non-normal Demand by Support Vector Machines with Ensemble Empirical Mode Decomposition

• DOI: 10.4156/aiss.vol3.issue3.11
• Publication Date: 2011-06-14T23:10:42Z
• Authors: Yukun Bao -; Rui Zhang -; Tao Xiong -; Zhongyi Hu -
• Keywords: Mode (computer interface)

Non-Normal demand is the with infrequent occurrences or irregular sizes, which very difficult to forecast. In this study, an ensemble empirical mode decomposition (EEMD) based support vector machines (SVMs) learning approach proposed forecast in these two cases. This under a “decomposition-and-ensemble” principal decompose original non-normal series into several independent “smooth” and “continuous” subseries including small number of intrinsic functions (IMFs) residue by EEMD technique. Then SVMs are used model each so as achieve more accurate respectively. Finally, forecasts all aggregated formulate for series. Two data sets has four artificial were test effectiveness approach. Empirical results demonstrate that outperforms other forecasting methods such , ARIMA Croston method terms RMSE, MAPE, MdRAE MASE.