A hybrid credit scoring model based on clustering and support vector machines

We propose a hybrid credit scoring model which employs clustering techniques to pre-process samples, and support vector machines to build classifiers. Firstly, we use self-organizing map to determine the number of clusters and the starting point of each cluster, and then use K-means clustering algorithm to generate clusters of samples. Secondly, we replace the class label of samples with the corresponding the cluster identifier. Thirdly, the pre-processed samples are used for training. We demonstrate the effectiveness of the proposed hybrid credit scoring model by testing it on two real world credit datasets.