摘要
企业信用风险评估是金融领域的重要课题。本文针对单独运用BP神经网络评估信用风险时存在的不足,提出了一种基于PSO-BP神经网络的企业信用风险评估模型。该模型首先应用主成分分析方法降低输入BP网络的信用评估指标维数,并且采用粒子群优化算法优化BP神经网络的权值。实验表明,新模型采用的算法具有收敛速度快,预测精度高的优点,是一种有效可靠的企业信用风险评估模型。
Credit risk assessment of enterprise is a very important problem in the financial field. Its deficiency is revealed because of singly using BP neural network. In this paper, a new assessment model is proposed which is based on an optimized BP neural network. In this model, first, dimensions of data are reduced by Principal Component Analysis （PCA） ; Second, Partiele Swarm Optimization （PSO） is used for optimizing parameters of neural network. The experiment results show that PSO-BP algorithm works with quicker convergence rate and the higher forecast precision, it can content with the demand of credit risk assessment of enterprise.
出处
《计算机与现代化》
2009年第4期123-126,129,共5页
Computer and Modernization
基金
福建省自然科学基金资助项目(20511006)
关键词
信用风险评估
主成分分析
粒子群优化算法
BP神经网络
credit risk assessment
Particle Component Analysis （PCA）
Particle Swarm Optimization（PSO）
BP neural network
