RBF神经网络在个人信用组合预测中的应用

Combining forecasts of personal credit scoring based on RBF neural network

运用组合预测的思想，提出了通过RBF神经网络将多元线性回归和logistic回归组合的预测方法，并应用于某商业银行的个人信用评估中，其结果表明组合预测方法能够获得比单一方法更高的预测精度，尤其在避免“纳伪”错误方面更具优势．

Using the idea of combining forecasts, this paper presents a new approach by combining multi-linear regression and logistic regression with RBF network, and applies it in personal credit risk scoring of one commercial bank. The results indicate that the new technique is more accurate than either of the individual technology, especially in avoiding recognizing the bad application as good ones.