摘要
以114家中小上市公司为研究对象,运用BP神经网络模型和径向基网络模型对训练样本和测试样本中一定比例的"非ST"和"被ST"进行了信用评估。按照各上市公司财务状况把公司划分为"好"和"差"两类。仿真结果表明:BP神经网络模型对测试样本的预测准确率高达88.9%,而径向基网络模型对测试样本的预测准确率只有77.8%,比BP神经网络模型的准确预测率低了11个百分点。
In this paper,114 small and medium-sized listed companies as the object of study,using the BP neural network model and radial basis network model of the training samples and test samples of a certain proportion of"non-St"and"St"credit evaluation.According to the financial situation of the listed companies to classify the company as"good"and"bad"two categories.The simulation results show that the BP neural network model can predict the accuracy rate of the test samples as high as 88.9%,and the radial basis network model is only77.8%to the test samples,and the accurate prediction rate of BP neural network model is 11%lower.
作者
何欣
张红梅
HE Xin;ZHANG Hongmei(Guizhou University of Finance and Economic s School of Finance;Guizhou Institution for Technology Innovation& Entrepren eurship Investmen;Guizhou Institution for Urban Economy and Development,Guiyang 550025,China)
出处
《科技创业月刊》
2018年第5期133-137,共5页
Journal of Entrepreneurship in Science & Technology
基金
贵州财经大学校级项目"科技金融支撑贵州精准扶贫的路径与对策研究"(2017XYB07)