摘要
介绍对银行贷款利润率的主要影响因素,基于改进型人工神经网络建立了贷款利润率的神经网络预测模型。在贷款利润率的神经网络预测系统中采用信息传输Tanh函数。取代Sigmoid函数,通过验证证实Tanh函数提高了贷款利润率的预测精确度。测试表明:采用Tanh信息传输函数,预测误差率能降低0.045。
The main influence factors of Loan Profit rates were discussed in this paper. A prediction model for loan profit rates based on improved BP artificial neural network was developed. A neural network test system for loan profit rates was established by using the information transfer Tanh function, which replaced the Sigmoid function in this paper. The information transfer function Tanh was more excellent than the function Sigmoid. It was proven that a higher forecasting precision for loan profit rates was obtained utilizing the Tanh function. The results showed that, the ararag prediction error of loan profit rates using Tanh information transfer function was low than 0. 045.
出处
《武汉理工大学学报》
CAS
CSCD
北大核心
2006年第6期133-136,共4页
Journal of Wuhan University of Technology
关键词
神经网络
信息传输函数
贷款利润率
neural network
information transfer function
loan profit rates