Foreign Exchange Rates Forecasting with a C-Ascending Least Squares Support Vector Regression Model

摘要

In this paper, a modified least squares support vector regression (LSSVR) model, called C-ascending least squares support vector regression (C-ALSSVR), is proposed for foreign exchange rates forecasting. The generic idea of the proposed C-ALSSVR model is based on the prior knowledge that different data points often provide different information for modeling and more weights should be given to those data points containing more information. The C-ALSSVR can be obtained by a simple modification of the regularization parameter in LSSVR, whereby more weights are given to the recent least squares errors than the distant least squares errors while keeping the regularized terms in its original form. For verification purpose, the performance of the C-ALSSVR model is evaluated using three typical foreign exchange rates. Experimental results obtained demonstrated that the C-ALSSVR model is very promising tool in foreign exchange rates forecasting.