A method of improving credit evaluation with support vector machines

With the growth of credit business scale, credit evaluation models are becoming more and more popular for credit admission decision with minimum risk. Among many of these models, the support vector machine (SVM) draws more attention for its effectiveness. In this work, we presented a new skill to further improve the effect of SVM for credit scoring. In the learning process of SVM, we utilized the whitening transformation, which is usually adopted in signal processing. We applied our method on two real credit data sets, and found that not only the credit scoring accuracy can be enormously improved, but the learning time of SVM models can also be obviously reduced.