摘要
企业财务数据较为庞大,不利于数据存储与管理,因此构建基于稀疏表示的财务数据集判别分析维数压缩模型,实现财务数据的降维处理。使用扩展主元分析算法中范式约束优化稀疏表示模型,获得财务数据测试样本的稀疏数据集。引入迹比率优化主元分析方法,以稀疏数据集作为目标数据,使用优化后的迹比率主元分析法提取该稀疏数据集相对于背景财务数据的低维表示,实现财务数据集判别分析维数压缩。试验通过特征提取效果评价维数压缩效果,该模型能够实现海量财务数据的降维,针对不同类型的财务数据,均能呈现出较好的特征提取精度与贡献度,具有较高可信赖性与连续性。财务数据集大小为2000维时,该模型的压缩能力最佳。
The financial data of enterprises is relatively large,which is not conducive to data storage and management.Therefore,the dimension compression model of financial data set discriminant analysis based on sparse representation is constructed to achieve the dimensionality reduction of financial data.The sparse representation model is optimized by the normal form constraint in the extended principal component analysis algorithm,and the sparse data set of financial data test samples is obtained.The sparse data set is taken as the target data.The optimized trace ratio principal component analysis method is used to extract the low-dimensional representation of the sparse data set relative to the background financial data,so as to realize the dimension compression of the discriminant analysis of financial data set.Test by feature extraction effect evaluation dimension compression effect,this model can realize the massive financial data dimension reduction,in view of the different types of financial data,it can present a good precision and contribution,feature extraction has high trustworthiness and continuity.The financial data set size of 2000 d best compression capability of the model.
作者
刘钰祺
Liu Yuqi(Wuhan Mental Health Center,Wuhan University of Science and Technology,Wuhan,430000,China;Wuhan Hospital for Psychotherapy,Wuhan University of Science and Technology,Wuhan,430000,China;Evergrande School of Management,Wuhan University of Science and Technology,Wuhan,430000,China)
出处
《现代科学仪器》
2023年第1期204-209,共6页
Modern Scientific Instruments
关键词
稀疏表示
财务数据集
判别分析
维数压缩
迹比率
主元分析
Sparse representation
Financial data set
Discriminant analysis
Dimension compression
Trace ratio
Principal component analysis
