摘要
针对单一模型存在的分类精度有限的不足,提出了应用组合预测模型进行信用评估的方法。选择Logistic回归和后验概率SVM模型作为单一模型,构建了基于二者的非负权重线性组合预测模型,并将模型应用于住房信贷评估。采用基于最大后验概率的分类准则,应用结果表明,组合评估模型的分类精度高于单一模型,并且获得了较好的稳健性,对于构建住房信贷评估模型是一个很好的选择。
Aiming at the deficiency of low accuracy of single models, a combining forecast model for credit evaluation is presented. Based on two single models of Logistic regression and posterior probability support vector machine, a linear combining forecast model with non - negative weights of each single model is constructed. The combining forecast model was applied into credit evaluation of individual residential loan. Based on classification criteria of maximum posterior probability, the experimental results indicate that combining evaluation model is superior to single model in terms of accuracy and stability, which makes it a reasonable option for the credit evaluation in individual residential loan.
出处
《黑龙江大学自然科学学报》
CAS
北大核心
2008年第3期281-286,共6页
Journal of Natural Science of Heilongjiang University
基金
国家哲学社会科学创新基地资助项目(htcsr06t06)
哈尔滨工业大学技术政策管理项目(TPM)
