A Combined Credit Scoring Model Based on Posterior Probability SVM and Logistic Regression for Residential Loans

摘要

To address the deficiency of low accuracy of single models in credit scoring, this paper presents a combined credit scoring model for residential loan’s credit scoring. Based on two single models of logistic regression and posterior probability support vector machine, this paper constructed a linear combined credit scoring model with non-negative weights of each single model. The combined credit scoring model was applied to the credit scoring of a residential loan. Taking the maximum posterior probability as the classification criteria, the experimental results indicate that the combined scoring model is superior to single models in terms of accuracy and stability, which makes it a reasonable option for the credit scoring in assessing individual residential loans.