商业银行的个人信用评估研究:基于PCA-GA-BP算法

随着个人消费贷款的普及,贷款人的个人信用评估变得尤为重要。本文选取德国和澳大利亚某商业银行的个人信贷数据为样本数据,采用主成分分析提取样本数据的主成分,通过遗传算法优化神经网络的网络结构、初始连接权值和阀值,然后将优化的神经网络算法用于个人信用评估。与其他算法的准确率比较的结果表明,基于主成分分析—遗传算法—神经网络算法的个人信用评估准确率要高,而且模型的网络结构得到优化,运算时间也有缩短。

A Study on Individual Credit Evaluation for Commercial Bank Based on PCA-GA-BP

This paper selects the personal credit data of a commercial bank in Germany and Australia, extracts the main component by principal component analysis, and then optimizes network structure, initial connection weights and thresholds of neural network by genetic algorithm, finally, according to this improved neural network technology, evaluates personal credit. Finally, it compares the accuracy based on this algorithm to these accuracies based on other algorithms. According to the results, the accuracy based on this algorithm is better than these accuracies based on other algorithms.