A New Method Evaluating Credit Risk with ES Based LS-SVM-MK

Abstract

The era of big data is here. Recent studies have revealed that emerging modern machine learning techniques are advantageous to statistical models for credit risk evaluation, such as SVM, LS-SVM. In this paper we discuss the applications of the evolution strategies based least squares support vector machine with mixture of kernel (ES based LS-SVM-MK) to design a credit evaluation system, which can discriminate good creditors from bad ones. Differing from the standard LS-SVM, the LS-SVM-MK uses the 1-norm based object function and adopts the convex combinations of single feature basic kernels. Only a linear programming problem needs to be resolved and it greatly reduces the computational costs. A real life credit dataset from a US commercial bank is used to demonstrate the good performance of the ES based LS-SVM-MK.