基于改进萤火虫算法的Probit模型在数字金融风险预测中的性能分析

Performance Analysis of Probit Model Based on Improved Firefly Algorithm in Digital Finance Risk Prediction

为应对数字金融的风险预警,需对其背后的规律进行深度挖掘,从而实现对其风险精准、高效地预测.在数据不平衡的情况下,采用常规单一类别模型进行数据分析时存在性能不稳定的问题,研究面向数字金融欺诈的浅层特征选择方法.为了解决数字化金融风险数据中存在的冗余和无关特征,采用了一种基于二元萤火虫算法的Probit模型,以更有效的方式来筛选数据中的初始特征.实验结果显示,融合烟花协同进化的二元萤火虫算法收敛速度和精度均显著优于其他算法.在初步特征子集中最佳分类精率为93.5358%,平均分类精度为86.2543%,运行时间为5.4883 s;运行时间得到了优化,分类精度同时也得到了一定的提高.研究提出的改进算法模型可以用于指导金融风险管理实践,帮助金融机构更好地识别和预防金融欺诈风险.

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 exists 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 characteristics of the data.The experimental results showed that the convergence speed and accuracy of the binary firefly algorithm fused with fireworks collaborative evolution were significantly better than other algorithms.The optimal classification accuracy in the preliminary feature subset was 93.5358%,the average classification accuracy was 86.2543%,and the running time was 5.4883 seconds;The running time was optimized,and the classification accuracy also improved.The improved algorithm model proposed in the study can be used to guide financial risk management practices and aid financial institutions in identifying and preventing financial fraud risks.