This study embedded a recursive feature selection scheme in support vector machines (SVM) for credit rating forecasting. SVMs have been successfully applied in numerous areas, and have demonstrated excellent performance. However, due to the high dimensionality of our input variables, this study employed a fast recursive feature selection algorithm to eliminate irrelevant features and enhance the performance of SVM classifiers. Empirical results have indicated that one-vs-one SVM with embedded recursive feature selection outperforms other multi-class SVMs. Compared to traditional multi-class classifiers, the performance improvement owing to embedded recursive feature selections is significant.
關聯:
Journal of Statistics & Management Systems, 13(1): 165-177
