基于SVM主动学习的制革企业财务管理数据分类方法  

Classification Method for Financial Management Data of Tanning Enterprises Based on SVM Active Learning

作　　者：高溪[1,2] 王良浩 GAO Xi;WANG Lianghao(Yantai Institute Of China Agricultural University,Yantai 264000,China;Guangxi Vocational&Technical College,Nanning 530001,China)

机构地区：[1]中国农业大学烟台研究院,山东烟台264000 [2]广西职业技术学院,广西南宁530001

出　　处：《皮革与化工》2023年第6期17-22,共6页Leather And Chemicals

摘　　要：制革企业财务管理数据具备不平衡性和非线性特征,导致难以对于数据属性进行标注,因此提出了一种基于SVM主动学习的制革企业财务管理数据分类方法。依靠向量空间模型对制革企业财务管理数据特征项进行提取并降维,将降维后的数据映射到特征空间。在特征空间利用SVM主动学习机进行数据属性标注,结合分类决策概率实现制革企业财务管理数据分类。仿真实验结果表明,所提方法的制革企业财务管理数据分类效果好,容错率较低,Kappa系数较高,制革企业财务管理数据分类效果更加理想。The financial management data of tannery enterprises have imbalanced and nonlinear characteristics,making it difficult to annotate data attributes.Therefore,a classification method for financial management data of tannery enterprises based on SVM active learning is proposed.Financial management data feature items in tannery enterprises are extracted and its dimensionality is reduced by vector space models,and the reduced data is mapped to the feature space.SVM active learning machine in feature space is utilized for data attribute annotation,and classification decision probability is combined to achieve financial management data classification for leather enterprises.The simulation experiment results show that the proposed method has good classification performance for financial management data of tannery enterprises,low fault tolerance,and high Kappa coefficient.The classification effect of financial management data of tannery enterprises is more ideal.

关 键 词：SVM主动学习 制革企业 财务管理数据 向量空间模型 特征降维 决策概率 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]