Weighted LS-SVM Credit Scoring Models with AUC Maximization by Direct Search

Credit scoring models are very important tools for credit granting institutions to assess the credit risk of their customers. Most previous researches focus on improving predictive accuracy of models. In this research, a weighted LS-SVM credit scoring model with Area under ROC curve maximization is proposed and optimized by direct search. The tests on two real-world datasets show that it is effective for building the credit scoring model with good AUC performance.