Performance Analysis of Probit Model Based on Improved Firefly Algorithm in Digital Finance Risk Prediction
To cope with the risk warning of digital finance,it is necessary to explore the underlying patterns to achieve accurate and efficient prediction of its risks.In the case of imbalanced data,unstable performance ex-ists by means of conventional single category models for data analysis.Thus,a shallow feature selection method for digital financial fraud was studied.In order to remove the redundant and irrelevant data in digital financial risk data,a Probit model based on the binary firefly algorithm was adopted to effectively filter the initial charac-teristics of the data.The experimental results showed that the convergence speed and accuracy of the binary fire-fly algorithm fused with fireworks collaborative evolution were significantly better than other algorithms.The opti-mal classification accuracy in the preliminary feature subset was 93.535 8%,the average classification accuracy was 86.254 3%,and the running time was 5.488 3 seconds;The running time was optimized,and the classifi-cation accuracy also improved.The improved algorithm model proposed in the study can be used to guide finan-cial risk management practices and aid financial institutions in identifying and preventing financial fraud risks.
NETL
NSTL
万方数据
维普
