ABSTRACT
In this chapter, the EEMD-LSSVR–based decomposition and ensemble methodology is utilized to forecast exchange rates. The EEMD-LSSVR decomposition and integrated approach mainly consists of three stages: First, data decomposition. The EEMD algorithm is used to decompose the exchange rate to several different intrinsic mode functions (IMFs) and a residual sequence. Secondly, single forecasting. LSSVR is employed to forecast those different IMFs and the residual sequence respectively. Thirdly, ensemble forecasting. Based on the simple addition ensemble method, the results are aggregated into final integrated results.
