摘要
目前,对农户信贷风险的评估方法很多,但大多采用专家主观判断法、统计判别分析法,主观性强,受样本数量限制。随着引入人工智能技术,出现了遗传算法、神经网络模型、支持向量机、随机森林等方法。在诸方法中,SVM是一种适用于少量样本的学习方法,可用于处理线性和非线性分类问题,尤其适用于农户信贷信息获取少而难的评估。文章运用农户信贷理论,以黑龙江省某农商行为主要研究对象,分析了农户贷款信用风险管理和评价体系现状,运用SVM模型和主成分分析构建了农户信用风险评价指标体系,并运用某农商行的数据进行了实证分析研究。得出结论:SVM模型在所有数据集上表现最好,其提取规则的准确性超越了传统的分类方法,可用作特征选择法的基础来确定违约风险重要的特征。
At present,there are many methods for evaluating the credit risk of farmers,but most of them still use expert subjective judgment and statistical discriminant analysis,which have strong subjectivity and are limited by the number of samples.With the introduction of artificial intelligence technology,methods such as genetic algorithms,neural network models,support vector machines and random forests have emerged.Among various methods,SVM is a learning method suitable for a small number of samples,which can be used to handle linear and nonlinear classification problems,especially suitable for evaluating farmers with limited and difficult access to credit information.This article applies the theory of farmers'credit and focuses on the behavior of a certain agricultural and commercial enterprise in Heilongjiang Province.It analyzes the current situation of credit risk management and evaluation system for farmers'loans,constructs a credit risk evaluation index system for farmers using SVM model and principal component analysis,and conducts empirical analysis using data from a certain agricultural and commercial enterprise.The conclusion is that SVM classifier performs best on all datasets,The accuracy of its extraction rules surpasses traditional classification methods and can be used as the basis for feature selection to determine the most important features of default risk.
作者
刘香
Liu Xiang(Harbin Finance University,Harbin,Heilongjiang,150030)
出处
《市场周刊》
2024年第7期19-24,共6页
Market Weekly
基金
黑龙江省金融学会“基于SVM模型的黑龙江新型农村金融机构农户信用风险评估研究”(项目编号:190105)。
关键词
农户信用风险
评价指标体系
主成分分析
SVM模型
农村金融机构
farmers'credit risk
evaluation index system
principal component analysis
SVM model
rural financial institutions
