【作者】 黄章树； 乔昕；

【Author】 HUANG Zhang-shu,QIAO Xin(School of Management,Fuzhou University,Fuzhou 350108,China)

【机构】 福州大学管理学院；

【摘要】 本文在确立了信用评价指标体系的基础上,分别采用了PSO-LSSVM方法、BP神经网络方法和C5.0决策树方法,构建了机械制造业上市公司的信用评价模型,用来对2001-2010年50家机械制造业上市公司进行两类模式分类。以t-3、t-4年的财务和非财务数据组合的面板数据作为研究样本,分别对机械制造业上市公司进行信用评价的实证研究。实证结果表明:t-4年PSO-LSSVM预测精度最高,t-3年BP神经网络模型的预测准确率最高。这三种方法的预测精度都较高,说明这三个数据挖掘模型可以用于中长期预测,并且都适合对机械制造业上市公司进行信用评价。更多还原

【Abstract】 Using PSO-LSSVM method、BP neural network method and C5.0 decision tree method,this paper has established credit evaluation index system and constructed a credit evaluation model of the machinery manufacturing industry’s listed company,to classify the two kinds of mechanical manufacturing listed companies.There were 50 samples between 2001 and 2010.The t-3,t-4 years’financial and non-financial panel data was viewed as research sample,to do the credit evaluation empirical research of machinery manufacturing listed companies.The empirical results show that: t-3 years of BP neural network model was the best,and t-4 years of PSO-LSSVM models’effect was the best.The prediction precision of the three models was high,demonstrated that the two models can be used for long-term prediction,both models are suitable for machinery manufacturing companies’credit ratings.更多还原