Multi-Step-Ahead Time Series Prediction using Multiple-Output Support Vector Regression

摘要

Accurate time series prediction over long future horizons is challenging andof great interest to both practitioners and academics. As a well-knownintelligent algorithm, the standard formulation of Support Vector Regression(SVR) could be taken for multi-step-ahead time series prediction, only relyingeither on iterated strategy or direct strategy. This study proposes a novelmultiple-step-ahead time series prediction approach which employsmultiple-output support vector regression (M-SVR) with multiple-inputmultiple-output (MIMO) prediction strategy. In addition, the rank of threeleading prediction strategies with SVR is comparatively examined, providingpractical implications on the selection of the prediction strategy formulti-step-ahead forecasting while taking SVR as modeling technique. Theproposed approach is validated with the simulated and real datasets. Thequantitative and comprehensive assessments are performed on the basis of theprediction accuracy and computational cost. The results indicate that: 1) theM-SVR using MIMO strategy achieves the best accurate forecasts with accreditedcomputational load, 2) the standard SVR using direct strategy achieves thesecond best accurate forecasts, but with the most expensive computational cost,and 3) the standard SVR using iterated strategy is the worst in terms ofprediction accuracy, but with the least computational cost.