Construction of a Fusion Model for Corporate Financial Fraud Detection with XGBoost
This study builds a fused model based on the XGBoost algorithm and explores parameter settings,performance evaluation,and optimization strategies to enhance the accuracy of financial fraud detection.The results demonstrate that the optimized SVM model achieves an AUC area of 0.77,while the random forest algorithm achieves 0.83.In contrast,the proposed XGBoost model achieves an AUC area of 0.85,indicating high accuracy in detecting financial fraud,with an AUC area close to the optimal range.Compared to traditional and deep learning algorithms,XGBoost exhibits significant advantages in both model performance and efficiency.Therefore,the proposed XGBoost-based corporate financial fraud detection model holds significant value for practical applications.Future research can further explore the application of XGBoost algorithm in other domains.
万方数据
维普
