Credit risk evaluation with kernel-based affine subspace nearest points learning method

Credit risk evaluation has long been an important and widely studied topic in bank lending decisions and profitability. Currently emerging data mining and machine learning techniques, such as support vector machine (SVM), have been discussed widely in credit risk evaluation. In this paper a new kernel-based learning method called kernel affine subspace nearest point (KASNP) approach is proposed for credit risk evaluation. KASNP approach is derived from the nearest point problem of SVM, which extends the areas searched for the nearest points from the convex hulls in SVM to affine subspaces. Similar to SVM, KASNP can also classify the typical nonlinear two-spiral problem well. But unlike SVM to solve the difficult convex quadratic programming problem, KASNP is an unconstrained optimal problem whose solution can be directly computed. We apply KASNP for credit evaluation, and the experiments on three credit datasets show that the proposed KASNP is more competitive for creditors classification.