摘要
为提高医院财务信息管理能力,构建了一种结合主成分分析(Principal Component Analysis,PCA)和变分自编码器(Variational Auto-Encoders,VAE)的异常检测模型。基于收集和预处理财务数据,通过PCA进行特征提取,利用VAE学习数据的潜在分布,并通过k折交叉验证提高模型的预测性能。实验结果显示,在训练集与测试集比例为9∶1的情况下,PCA-VAE在异常检测任务中表现出了优秀的性能,其精度、召回率和F1得分分别为0.9467、0.9421和0.9444,显著优于传统机器学习算法和结合PCA方法的分类模型。
In order to improve the hospital financial information management ability,an anomaly detection model combining Principal Component Analysis(PCA)and Variational Auto⁃Encoders(VAE)is constructed.Based on collection and preprocessing financial data,feature extraction by PCA,learning the potential distribution of data using VAE and improving the predictive performance of the model by k⁃fold cross⁃validation.The experimental results showed that PCA⁃VAE showed excellent performance in the anomaly detection task with the training set ratio of 9∶1,with its accuracy,recall and F1 scores of 0.9467,0.9421 and 0.9444,respectively,significantly outperforming traditional machine learning algorithms and classification models combining PCA methods.
作者
竺三子
孙训
马哲文
ZHU Sanzi;SUN Xun;MA Zhewen(Department of Finance,Xuancheng People’s Hospital,Xuancheng 242000,China;Department of Administration,Xuancheng People’s Hospital,Xuancheng 242000,China)
出处
《电子设计工程》
2025年第1期17-20,26,共5页
Electronic Design Engineering
基金
安徽省宣城市科技计划项目(2057)。
关键词
财务管理
主成分分析
变分自编码器
异常检测
financial management
Principal Component Analysis
Variational Auto⁃Encoders
anomaly detection