Online Loan Default Prediction Model Based on Deep Learning Neural Network (NN)

With the rapid development of China's economy and the emergence of the concept of advanced consumption, credit has become an important way for enterprises and individuals to solve economic difficulties. Lending has promoted the rapid rise of loan financial business, spawned many credit institutions, followed by increasingly fierce market competition and more diversified loan methods. Compared with other credit institutions, the income that credit can bring to financial enterprises is much higher than that of ordinary loan business, but the credit risk is very high due to the insolvency and other defaults in the loan. Therefore, this paper puts forward the online loan default prediction model based on the deep learning NN. Firstly, it analyzes the main reasons for the online loan default risk, then explains the theoretical basis for the selection of the deep learning NN model, and finally puts forward the planning scheme for the loan default prediction based on the deep learning NN. Through the experimental analysis of online loan default prediction model, the statistical results show that there is no obvious correlation between loan grade and customer default. The number of defaults of customers with higher loan grade is small, but their average loan amount is high and the corresponding loan interest rate is high; In the application of deep learning NN to online loan default prediction, the number of customer defaults in the balanced training set reached 1025038, but the category ratio decreased to 512348:512654, and the test set also decreased, almost reaching the ratio of 1:1. This has little impact on the subsequent model prediction. The online loan default prediction model based on deep learning NN proposed in this paper is of great significance to reduce loan default.