Research And Application On Sampling Of Tax Enforcement Inspection Based On Data Mining

With the rapid development of China's economy,great changes have taken place in the entire society.Personal taxes and corporate taxes play an important role in the national economy,the tax department needs to increase its internal supervision while adapting to the development of the times.As the first step in the work of law enforcement inspections,the selection of cases is a top priority.Nowadays,the commonly used method is computer-aided selection based on indicators.This method can find out all the doubtful cases based on indicators.However,the indicator system used in this method is based on many years of experience in selection of cases,and by default this indicator system is absolutely reliable,the actual selection time is too long.Therefore,we want to join data mining techniques and actively seek out the best solution.We will use the preprocessed data directly to generate clues,identify high-risk data from massive data,provide important information for law enforcement inspectors,and to judge the rationality of these clues,realize computer aided used based on data mining.This thesis is based on the "the platform of Risk monitoring,of local taxes at the provincial level,the data mining technology is used to study and realize the selection of tax law enforcement inspection,it focuses on the k-mean clustering algorithm to do a lower dimensional clustering operation on the initial processing of the data,according to the similarities among taxpayers,taxpayers can be grouped together to analyze high-risk taxpayers,using the Apriori algorithm to find the law of occurrence of doubtful data from the sample data,find out the combination of indicators that will appear with a large probability as a prediction rule.Combining the obtained forecasting rules with the current irregularities and predicting possible violations by taxpayers,the competent tax authorities will supervise these potentially infringements,and select the objects and directions that need supervision to achieve the election.Through the understanding and research of clustering and association rules,the system of tax enforcement inspection based on data mining technology is implemented under the existing working framework of "the platform of Risk monitoring ".The efficiency of the selection is greatly improved and the effect of the selection is achieved.This thesis is valuable.Through continuous research,it is found that some technologies have not yet reached the 100 percent accuracy,but this thesis is of reference to other researchers.