Research on Dynamic Prediction Model of Consumer Credit Risk under Fintech Innovation

Consumer credit has become an increasingly popular form of credit in recent years, and its risk prediction is one of the elements that need to be paid attention to in the development of financial technology innovation. The study constructs a consumer credit risk prediction model based on survival analysis and introduces the concept of survival time into the field of consumer credit risk prediction. The survival analysis method and Cox proportional risk model are used to construct a dynamic prediction model of consumer credit risk. Compare the ROC curve (AUC value), KS value, and probability value of this model with other risk prediction models in order to analyze the prediction performance of the Cox model. The Cox model is used in an example analysis to predict whether a borrower is overdue and its overdue date and the predicted results are compared with the actual results to further test the predictive effect of the Cox model. The predictive performance of the Cox proportional risk model is significantly better than other risk prediction models. The Cox model predicts that borrowers 3 and 9 out of 10 will default on their loans, and the default date will be the 286th day and 357th day, respectively. This prediction result overlaps with the actual situation, and Cox’s prediction performance is excellent.