摘要
贷款业务是银行极为重要的资产业务,构建一个适用的客户信用评估模型十分重要。由于近年来在智能学习系统领域发展起来的新理论,并引入小样本学习的通用学习算法---支持向量机(Support Vector Machines,简称SVM),建立银行客户信用评估模型。由于在统计学习理论中的结构风险最小化的SVM算法,克服了传统信用评估模型中的过拟合和局部最优的缺点。同时,通过在模型中采用核函数,有效地解决了线性不可分问题。因此,使得基于这种技术的评估模型具有较强的实用性。通过与神经网络模型的比较,证实了该方法用于风险评估的有效性及优越性。
loan is the key capital business in banks. Its very important to build a suitable model for credit scoring. Based on the new developing theory in the intelligent learning system domain, the support vector machine （SVM） technique is introduced. On the basis of SLT （statistical learning theory）, this approach with methodology of SRM （structural risk minimization） will overcome the shortcomings of traditional credit assessment models, such as overfitting and local optimization, meanwhile, by using kernel functions in model , it will effectively solve the problems of linear inseparability. The practical result has indicated that SVM is effective and more advantageous than neural network model.
出处
《科学技术与工程》
2007年第8期1624-1627,共4页
Science Technology and Engineering
关键词
银行客户信用评估
支持向量机
分类
customer＇s credit scoring in banks SVM classification
