Data Mining In Tax Analysis

With the development of modern information technology and artificial intelligence technology, human society has stepped into the information age. Chinese tax administration bureau has a great number of original tax data through the construction and wide usage of the tax information system, which can reflect the condition of the national economy. Knowledge discovered from data analyzing can help the macroscopical control of government.Traditional tax analysis is too simple to satisfy the needs of various users, which only contains simple qualitative description and year on year growth rate. Data in the CTAIS aren’t mined effectively and can’t transfer valuable information. The resources are wasted.The development of data mining technology provides method to solve this dilemma. Based on the technology of data mining and some classical algorithms, and combined with the existing tax analysis job, this paper uses regression, clustering, and factor analysis to improve tax analysis report. Besides, the case behind also verifies the feasibility of the methods.First, using regression mode to simulate the trend of the tax revenue, and find it like an exponential curve. Then using logistic regression mode to analyze the relationship between tax revenue and GDP, and the coefficient of the variable is the estimated value of tax elasticity. Besides, using K-means clustering algorithm to classify the provincial administrative units by the various types of tax revenue. These units are divided into three categories and the results are economic significant. In addition, using factor analysis to reduce the dimension of financial variables of publicly held companies, and the purpose is to monitor the tax resource of corporate income tax by the Z-score. The conclusion is when a Z-score is at the middle range, the company’s financial statements may be faked at a higher possibility which can break the stability of the tax resource. At last, using year on year growth, exponential curve and logistic regression curve to predict the tax revenue, and in this case, the exponential curve behaved the best. The tax information construction of China has made great achievements, and introducing data mining into tax analysis can improve the quality and efficiency of tax management.