Combining Forecast Model Based on PSO for Personal Credit Scoring

Aiming at the insufficiency of credit scoring models, this paper puts out a new approach by using combining forecast model for personal credit scoring. Based on linear regression and logistic regression models, this paper constructed a combining model and used particle swarm optimization (PSO) to search each single model’s weight. In order to control the type II error rate,the particle’s fitness function was used to achieve the goal. The application results indicate that the combining model gets higher accuracies with lower type II error rates on training samples and testing samples. The combining model based on PSO algorithm presents more applicable for commercial banks to control credit risks.