Abstract
Consumer credit prediction is considered as an important issue in the credit industry. The credit department often makes decision which depends on intuitive experience with large risk. This study proposed a new model that hybridized the support vector machine (SVM) and particle swarm optimization (PSO) to evaluate the new consumer's credit score. The hybrid model simultaneously optimizes the SVM kernel function parameters and the input feature subset in order to achieve a high accuracy. Two UCI credit data sets are selected as the experimental data to evaluate the prediction performance of the hybrid model. The experimental results are compared with other existing methods which imply that the PSO-SVM model is a promising approach for credit scoring.