基于主成分分析和支持向量回归机组合模型的电子商务信用风险度预测研究

E-commerce Credit Risk Prediction Base on PCA-SVR

随着电子商务的快速膨胀,信用风险对电子商务发展的影响越来越突出。信用风险已成为电子商务企业所面临的最主要风险之一。文章结合企业的财会指标和电子商务运营能力构建企业电子商务信用风险度预测指标,并利用主成分分析对指标进行筛选,在此基础上通过支持向量回归机对电子商务信用风险度进行预测,并进行实证检验,实证结果表明,此方法与标准支持向量回归机和神经网络相比具有更高的分类精度,证实了该方法的可行性和优越性,为电子商务建立可靠的信用风险度预测系统提供依据。

With the rapid expansion of e - commerce, the impact of the credit risk about e - commerce development has ^- come increasingly prominent. Credit risk has become one of the major risks of e - commerce businesses. This paper considerd corpo- rate finance and accounting indicators and capabilities of e - commerce operational to build e - commerce credit risk pJ^xtiction index- es, using principal component analysis （PCA） to screened the indexes. Then it predicted the degree of e - commerce credit risk by support vector regression （SVR）, and by empirical testing. The empirical results show that this method with a single support vector regression and neural network methods were compared with higher class/flcation accuracy, confirming the feasibility of this method and superiority, provide the basis for e - commerce enterprises to establish a reliable credit risk prediction system.