财务困境预测:数据挖掘方法的比较与运用

近年来,数据挖掘方法在商业领域的应用方兴未艾。文章尝试将数据挖掘方法引入财务困境预测的问题中,并以上市公司的实际财务数据为出发点,全面比较了逻辑回归、神经网络和决策树等分类算法在上市公司财务困境预测问题上的优劣。结果表明决策树在预测准确率、波动性以及可解释性上具有综合优势。文章还提出了不同程度财务困境的新概念,并对这个问题进行了决策树建模。

Financial Distress Prediction: The Comparison and Application of Data Mining Models

Recently,many research efforts have been put on the applications of data mining methods to business domain.This paper aims at using data mining methods to predict the financial distresses.Three classifiers,i.e.,logistic regression,artificial neural networks,and decision tree,have been modeled on real-world financial data,and the results show the superior performances of decision tree measured by the predicting accuracy,the fluctuation of accuracy,and the explicability.In addition,a new concept of financial distress in different levels has been proposed and modeled by decision tree.