The Applied Research Of Credit Scoring Combination Models Based On SA-GA Algorithm

In recent years, with the rapid growth of Chinese economy and expanding domestic consumption, driven by a variety of policy just like urban Housing System Reform, the development of personal credit demand in China is strong and the scale is expanded gradually. However, in the process of continuous development, there are many problems in domestic commercial banks credit risk management business. Particularly lack personal credit scoring method which match with the development of personal credit. To some extent this problem has constrained and impeded the sound development of consumer credit business. Therefore, develop such a set of personal credit scoring methods can reduce the risk, not only has high academic value, but also has a strong use of significance.This issue of personal credit score, based on domestic and foreign personal credit score models and the principle of combination forecast, propose optimization in the weight of a single model in combination model through using SA-GA. Construction individual credit score combination forecast mode based on SA-GA. Firstly, start from the applicability of the question that the principle of the genetic algorithm (GA) and simulated annealing (SA) on the GA to optimize the weight of combination forecasting model, and detailed analyze feasibility in SA algorithm which has strong local search ability into GA algorithmic which has strong global search ability. And then, determine the BP network model and RBF network model as a single model. After that, this paper highlights the personal credit forecasting Model theory and process based on the SA-GA algorithm. And false positive rate of combination forecasting model as the fitness function of SA-GA. Use the strong local search ability of GA algorithm to improve the overall predictive. Finally, comparative analysis the accuracy and soundness of different sample application in two single model, GA model and SA-GA combined model. It show that combined model can be effectively integrated the advantages of a single model. And the improved GA which based on SA algorithm has more advantages in prediction accuracy and soundness. The advantage of it is interpretable, can help to reach the desired results of model.