我国上市公司被特别处理的组合预测模型研究  

Study on Nonlinear Combining Forecasting Model of ST for Listed Firms in China

作　　者：李秉祥[1] 

机构地区：[1]西安理工大学工商管理学院

出　　处：《运筹与管理》2004年第3期76-80,共5页Operations Research and Management Science

基　　金：国家自然科学基金资助项目(70372053)

摘　　要：针对组合预测比单项预测具有更高的预测精度,本文提出了一种基于模糊神经网络的上市公司被ST的非线性组合建模与预测新方法,并给出了相应的混合学习算法。通过与多元线性回归模型、Fisher模型和Lo gistic回归模型的预测结果对比表明,该方法具有预测精度高,学习与泛化能力强,适应性广的优点。As combining forecasts are more accurate than individual ones, this paper presents a new method to set up nonlinear composite and forecast the special treatment (ST) for listed firm based on fuzzy neural network, and gives the corresponding composite learning algorithm. The method has more accuracy, reinforcement learning properties and mapping capabilities than LPM, Fisher model and Logistic model. The method is better than other ones in forecasting the ST for listed firms.

关 键 词：组合预测 多元线性分析 模糊神经网络 算法 

分 类 号：F440[经济管理—产业经济]