数字化背景下皮革企业财务舞弊识别算法对比研究

Comparative Study on Fraud Identification Algorithms of Leather Enterprise Financial Data Mining under Digital Background

为判断不同财务舞弊识别算法优劣性,本文以我国某皮革上市企业财务报告为基础,对该企业财务报告舞弊问题进行挖掘。通过逻辑回归、决策树、支持向量机、神经网络4种识别算法识别效果进行对比后发现,神经网络算法在进行财务数据挖掘舞弊识别时的正确率、召回度、精准度、F1Score均最佳,其次为支持向量机。但是由于神经网络算法模型的搭建难度较大,模型较为复杂,需要根据企业或审计的具体需求进行灵活选择。

In order to judge the merits of different financial fraud identification algorithms, this paper took the financial reports of one leather listed enterprises in China as the basis, and excavated the fraud problems in the financial reports of the enterprises. Through the comparison of the recognition effects of the four recognition algorithms of logistic regression, decision tree, support vector machine and neural network, it was found that the neural network algorithm had the best accuracy, recall, accuracy and F1 Score in financial data mining fraud identification, followed by support vector machine. However, due to the difficulty of constructing the neural network algorithm model and the complexity of the model, flexible selection should be made according to the specific needs of enterprises or auditing.