ARTICLE
Optimization of support vector regression under uncertainty using robustness approach
- International Journal of Nonlinear Science , 39 (02) : 1-20
Discipline :
Mathématiques, Physique, Chimie et Informatique
Auteur(s) :
Mohamed ZONGO, Satafa SANOGO, Moussa BARRO, Sado TRAORE
Auteur(s) tagués :
ZONGO Mohamed
Renseignée par : ZONGO Mohamed
Résumé
In this paper, we propose a Robust Support Vector Regression (RSVR) model for forecasting using a robustness-optimization approach. We use Charnes-Cooper variable
transformation to show that the robust counterpart of uncertain Support Vector Regression(SVR) is equivalent to a quadratic optimization problem. Two numerical examples
are presented to illustrate the effectiveness of our approach: one on a theoretical function
and another using real rainfall data from Burkina Faso.
Mots-clés
Robust Support Vector Regression ; Optimization of robustness ; Uncertain data ; Robust counterpart