Application Research Of Data Mining Technology In Tax Audit

Under the background of the rapid development of data information,the level of data change in various industries in China is also rapidly accelerating,the trend of massive data information is irresistible,and the audit industry is also facing unprecedented challenges and opportunities.Although the application of computer audit has gradually replaced the traditional audit process and become the dominant tax audit,the audit risk is still increasing in the face of increasing data.Aiming at the problems of huge amount of audit data,limited knowledge and experience of auditors and difficult tracking of audit data in traditional tax audit methods,this paper uses computer-aided audit technology to collect,clean up,transform and analyze data,constructs tax audit database,and comprehensively utilizes data warehouse technology.Pattern recognition method,data analysis method and anomaly detection theory are the research methods.The tax data information is deeply excavated to obtain intrinsic pattern features.In this paper,decision tree algorithm and K-means-based clustering algorithm are used to discover the intrinsic discovery patterns hidden in tax data,supported by case studies.SPSS software is used as the implementation basis of the algorithm,and 12 kinds of tax audit characteristics in 43 financial audit cases are used as the model elements to construct the financial audit process.Problem may arise in data mining research,and enterprise tax integrity modeling and analysis,constructs a tax audit model based on data mining technology,and then verifies the application of the model by using auxiliary case data,and gives the results of the audit of enterprises with tax problems.The main contents of this paper are as follows:1.This paper expounds the research background and significance of tax auditing,analyzes on the research status of tax auditing at home and abroad and the main problems existing in the current tax auditing,and puts forward some suggestions for improvement.This paper expounds the characteristics of tax data and the general process and basic theory of data mining,which lays the theoretical foundation of this paper.2.Detailed analysis and summary of the advantages and disadvantages of each data mining technology,and gives the specific implementation process of each technology;Analyzes the data outlier detection algorithm,for each detection algorithm to clarify the advantages and disadvantages of its existence,to a certain extent,for the fourth sectionof the data preprocessing module provides a certain degree of.Theoretical basis.3.Based on the data of 43 enterprises,Establishes the database of tax inspection by using data mining technology.Based on the decision tree algorithm,the credibility of enterprises' tax payment is evaluated.When 100 parent nodes and 50 leaf nodes are selected in the decision tree model,the repeated data are trained for two rounds.The average accuracy of training is 73.1%,and the average accuracy of testing is 78.85%.That is to say,the average correct identification accuracy of tax-paying honest enterprises and non-tax-paying honest enterprises is 78.85%.Although the recognition accuracy is not ideal,this paper explores the feasibility of using decision tree algorithm in tax audit,which will be applied in the future.Foreshadowing.4.Constructs a tax audit model based on K-means clustering analysis algorithm.In the two clusters generated by the model,the iteration can converge to a stable value only once.The distance between the whole cluster center and the first cluster moves0.927,while the distance between the whole cluster center and the second cluster moves0.927.Moving 1.185,however,the central distance between the two clusters is 2.892,which shows that the deviation between the data is small,the convergence time to the stable value is shorter,and the number of iterations is less.The results show that the classification results obtained by K-means clustering algorithm are better than those obtained by decision tree algorithm.The feasibility of using these two methods in tax audit is explored,which lays a foundation for applying similar data mining technology to tax audit in the future.