Linex损失下两种信用评分模型的比较

研究和比较了两种信用评分的学习方法,一个是Linex损失的投影寻踪学习网络,一个是Linex损失的BP神经网络。分别分析两种算法的优缺点,并给出了UCI机器学习数据库中德国信用评分数据集上的测试结果,实验结果表明,Linex损失下投影寻踪学习网络优于BP神经网络。

Comparison of the Two Credit Scoring Models under Linex Loss Function

This paper studies and compares two types of learning methods for credit scoring,one is Linex loss function-based projection pursuit learning network,and the other is Linex loss function-based neural network. It respectively analyzes the advantages and disadvantages of the two algorithms and provides the test results on the Germany credit scoring data sets in UCI. The results show that the projection pursuit learning network is better than BP neural network under Linex loss function.