Research on application of personal credit scoring based on BP-logistic hybrid algorithm

In order to improve the robustness and accuracy of the credit evaluation model, we study on individual credit risk, select a statistical method of Logistic regression and a non-statistical method of neural network BP algorithm, which are most frequently used methods by domestic and foreign banks. Furthermore, we separately improve these two methods to some degrees, using Clementine tools to build Personal Credit evaluation model based on BP-Logistic mixed strategy, which improves the accuracy and robustness of the assessment model.