基于PSO-SVM的上市公司财务危机预警  

Financial Distress Early-warning of Listed Companies based on PSO-SVM

作　　者：周红晓[1] 田伟福[2] 

机构地区：[1]金华职业技术学院,浙江金华321007 [2]浙江师范大学,浙江金华321004

出　　处：《金华职业技术学院学报》2014年第6期61-66,共6页Journal of Jinhua Polytechnic

摘　　要：为实现上市公司的财务危机预警,根据我国的实际情况,将因财务状况异常而被特别处理(ST)作为沪深证券交易所上市公司陷入财务危机的标志,建立财务危机预警模型。该模型采用粒子群优化算法对支持向量机参数进行优选。仿真结果表明,该模型能有效地避免过拟合和欠拟合现象,与神经网络模型相比泛化能力强,预测精确度高。In order to realize the financial crisis early-warning of the listed corporation, according to the actual situation in our country, the special treatment （ST）is usually used as a symbol of the financial crisis in the companies which have been listed on Shanghai and Shenzhen Stock Exchange. A financial distress early-warning model of listed companies is constructed. Particle swarm optimization （PSO）is used to optimize the parameters of support vector machine （SVM）in this model. Empirical studies show that the model can effectively avoid the overfitting and underfitting phenomenon. Compared with neural network model, the model is of high precision and strong generalization ability.

关 键 词：支持向量机 粒子群优化算法 财务危机预警 libsvm工具箱 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] F275[自动化与计算机技术—控制科学与工程]