Research on Personal Credit Risk Assessment Model Based on Machine Learning

As the personal credit business continues to grow and expand, credit assessment has become a key task in risk management for the entire credit industry. In order to reduce the number of non-performing loan cases and their probability of occurrence, financial institutions need to fully understand the creditworthiness of their customers before lending, assess the quality of their customers, and reduce non-performing loans by reducing the potential risk borne by the financial institutions. In this paper, based on default data, the SMOTE algorithm was first applied to solve the problem of unbalanced classification samples, and then logistic regression, random forest and SVM algorithms were selected to build the credit evaluation model, and the logistic regression model was selected as the optimal model using the AUC value as the judgment criterion.