Research On Sampling Of Tax-checking Based On Data Mining

Since gold-tax engineering construction started comprehensively in1994, the informationconstruction in Chinese tax sector has entered a fast development stage. CTAIS(China TaxationAdministration Information System) system has been widely applied and tax data, such as basicinformation and tax declaration of tax payer and so on, have been collected by provincial taxsector. But these so mass data are dispersing, business-orienting and lack of being arranged andintegrated, which cause so much island of information in tax sector. By using data miningtechnology, we can analyze and exact implied knowledge intelligently and automatically fromthese so mass data and it contributes to raising the depth and breadth of tax data using, providingdecision support for tax department, of which help to improve the efficiency and managementlevel. So, this research will have very important significance.This paper has combined work practice from tax inspection by tax department. It starts fromcurrent research on sampling of tax-checking, explain data mining technology, and, apply tostepwise method of discrimination analysis to filter out the critical index of tax-checking on thefoundation of43sample data; Secondly, establish a sampling of tax-checking math modelaccording to Logistic regression and come conclusions. Last, provide some related policyrecommendations from macro aspects.