Forecasting financial condition of Chinese listed companies based on support vector machine
Forecasting financial condition of Chinese listed companies based on support vector machine
复制标题
DOI：
10.1016/j.eswa.2007.06.037
复制
发表时间：
2008-05-04
影响因子：
8.5
通讯作者：
Zen, Yueming
中科院分区：
文献类型：
--
作者：
Ding, Yongsheng;Song, Xinping;Zen, Yueming
由于中国资本市场的剧烈变化和特殊性，建立一个强有力的财务危机预警模型具有挑战性。本文首先运用统计学方法分析了中国特殊处理公司作为困境样本的可行性。在此基础上，建立了基于支持向量机（SVM）的中国高科技制造企业非匹配样本预测模型。采用10折交叉验证的网格搜索技术来寻找支持向量机核函数的最佳参数值。实验结果表明，所提出的支持向量机模型优于传统的统计方法和反向传播神经网络。总的来说，支持向量机为中国上市公司财务困境的预测提供了一个稳健的模型，具有较高的预测精度。采用中国特殊事件作为分界线对模型的预测精度也有一定影响。(c)2007爱思唯尔有限公司保留所有权利。
Due to the radical changing and specialty of Chinese capital market, it is challenging to develop a powerful financial distress prediction model. In this paper, we first analyzed the feasibility of Chinese special-treated companies as distressed sample by using statistical methods. Then we developed a prediction model based on support vector machines (SVM) for an unmatched sample of Chinese high-tech manufacture companies. The grid-search technique using 10-fold cross-validation is used to find out the best parameter value of kernel function of SVM. The experiment results show that the proposed SVM model outperforms conventional statistical methods and back-propagation neural network. In general, SVM provides a robust model with high prediction accuracy for forecasting financial distress of Chinese listed companies. It is also suggested that Chinese special-treated event adopted as cut-off line has some effect on the prediction accuracy of the models. (c) 2007 Elsevier Ltd. All rights reserved.