朴素贝叶斯算法在审计抽样中的应用研究

Research on Application of Naive Bayes Algorithm in Audit Sampling

审计抽样是指审计人员对具有审计相关性的部分项目实施审计程序,根据抽样结果推断总体情况。非统计抽样作为广泛使用的审计抽样方法主要依据审计人员主观经验判断,抽样风险高且缺乏理论基础。本文将机器学习领域简单高效的朴素贝叶斯算法应用于审计抽样问题,以专项资金审计为例,提出贝叶斯算法审计抽样模型。通过抽样率与“三因素”分析法对模型抽样结果进行评估,验证贝叶斯算法审计抽样模型的可靠性。模型将审计人员职业经验判断与概率统计知识相结合,能够降低审计成本、提高审计效率、控制审计风险,并为人工智能审计、在线审计提供新思路。

Audit sampling refers to that auditors implement audit procedures on some projects with audit relevance and infer the overall situation according to the sampling results.As a widely used audit sampling method,non statistical sampling is mainly based on auditors'subjective experience judgment,with high sampling risk and lack of theoretical basis.In this paper,the simple and efficient naive Bayesian algorithm in machine learning field is applied to audit sampling problem.Taking special fund audit as an example,the Bayesian algorithm audit sampling model is proposed.Through the sampling rate and"three factors"analysis method to evaluate the model sampling results,verify the reliability of Bayesian algorithm audit sampling model.The model can reduce audit cost,improve audit efficiency,control audit risk,and provide new ideas for artificial intelligence audit and online audit.