Dynamic support vector machines for non-stationary time series forecasting

Article type: Research Article

Authors: Cao, Lijuan^a | Gu, Qingming^b

Affiliations: [a] Institute of High Performance Computing, 89C Science Park Drive #02-11/12 118261 Singapore. E-mail: [email protected] | [b] Office of Nanjing Committee of P.R. China, Beijing East Road 210008 China. E-mail: [email protected]

Abstract: This paper proposes a modified version of support vector machines (SVMs), called dynamic support vector machines (DSVMs), to model non-stationary time series. The DSVMs are obtained by incorporating the problem domain knowledge -- non-stationarity of time series into SVMs. Unlike the standard SVMs which use fixed values of the regularization constant and the tube size in all the training data points, the DSVMs use an exponentially increasing regularization constant and an exponentially decreasing tube size to deal with structural changes in the data. The dynamic regularization constant and tube size are based on the prior knowledge that in the non-stationary time series recent data points could provide more important information than distant data points. In the experiment, the DSVMs are evaluated using both simulated and real data sets. The simulation shows that the DSVMs generalize better than the standard SVMs in forecasting non-stationary time series. Another advantage of this modification is that the DSVMs use fewer support vectors, resulting in a sparser representation of the solution.

Keywords: non-stationary time series, support vector machines

DOI: 10.3233/IDA-2002-6105

Journal: Intelligent Data Analysis, vol. 6, no. 1, pp. 67-83, 2002