基于学习矢量量化算法的财务失败预测研究

实现有效的财务失败预测对于银行、投资者、企业和政府管理机构来说具有重要的意义，因而相关研究一直在金融信息处理领域中备受关注。近年来，神经网络方法被引入该领域并成为新的研究热点。文中分别利用160家和384家公司的财务数据作为训练集和测试集，首次将学习矢量量化(LVQ)算法应用至中国上市公司的财务失败预测模型的构建．井与传统的BP神经网络、对数回归模型、C4．5决策树等方法进行了实证分析比较。研究结果表明学习矢量量化算法与这些传统方法相比具有更高的预测精度，在此领域有着良好的应用前景。

Financial Failure Prediction Based on Learning Vector Quantization

Effective financial failure prediction is the important requirement of banks, corporations and government management organizations. And the related research is always a hot topic in finance information processing field. Recent years, neural networks have been introduced into this new field and have become new hot. This paper applies learning vector quantization to financial failure prediction based on companies in Chinese Stock Markets firstly and compares it with BP neural network, logistic regression and C4.5 models. The dataset includes 160 companies as train set and 384 ones as test set. The results show that LVQ method outperforms these models and has good application prospect in this field.