Application Of Personal Credit Evaluation Based On Logistic Regression And Neural Network

Along with the rapid development of China’s economy, the personal credit consumption also gradually becomes the main way of personal consumption. The rapid growth of the individual credit consumption demand has a complete set of commercial bank credit risk management system. But the lack of scientific personal credit evaluation system is China’s commercial bank credit risk management in one of the most serious problems. Presently although proposed including mathematical statistics, artificial intelligence and so on many kinds of credit evaluation model just some credit evaluation model is can’t be widely, the main reason is the many model accuracy improved at the same time, the conservatism is not very ideal, model generalization ability is not strong, to solve these problems, only the improved model is not enough, the choice of the index and dealing with also is very important. So, this article will use the feature selection and combination forecast model ideas to solve these problems.In this paper based on the study of the scholars both at home and abroad, we analyzed the principle and modeling thought of single model and combination model. In the statistical learning theory, logistic regression was widely believed to be a good model, that it has not too many requirements in data and assumptions. Second, the robustness of Logistic regression is good, can apply to different inspection samples. In addition, Logistic regression can predict which variable is remarkable to the personal credit status. Today, neural network is a widely used classification algorithm of intelligence. It has many characteristics, for example nonlinear, high tolerance, classification precision higher. But they also have their own defects; we use the combination model in order to achieve better classification results.