Research On Sampling Of Tax-checking Based On Data Mining

This paper focuses on how to make use of data mining technology to implement tax-checking sampling, organize the indexes of tax-check, understand the inspection object in depth, determine the enterprise whether or not has doubts and predict the unhonest business. This method can reduce the inspection costs, make the audit work more targeted and certainty and improve efficiency of inspection. According to the characteristics of this research, it used some research methods, such as the qualitative analysis and quantitative analysis, the combination of inheritance and innovation. The main contents of the study can be summarized as follows.Firstly, in explaining the research background and research significance, this paper analyzed tax-checking sampling theory of domestic and foreign and recent research related technologies of data mining and elaborated basic theory of tax check and the basic knowledge of data mining. It laid a theoretical foundation for this paper.Secondly, based on analyzed the source index target of tax check, this paper determine items of taxation, made use of the level of analysis and discriminant analysis methods to respectively analyze qualitative indicators and quantitative indicators and finally determine the index system for tax inspectors.Once again, based on data warehouse technology of data mining and past database of tax, this paper set up data warehouse of tax check, and based on that mined the strong association rules using the Apriori algorithm of association rules, which is most likely related to target tax evasion.Lastly, According to previous knowledge and methods, this paper set up a decision-making system of tax-checking sampling in Hebei Province, this system really realize the computer aided casing, to make up for some shortcomings in the artificial tax check. It gave full play to the superiority of modern information technology.