Statistical Analysis Of Local Fiscal Revenue Based On Kernel Method

The essence of fiscal revenue,a distributional relationship,is the monetary fund raised by the government to fulfill its functions,promote public policies,and provide public goods and services.The form of fiscal revenue in China is diversified,including taxes,profits,debts,fees,etc.,namely,tax revenue,profits paid by state-owned enterprises,debt income,and other income.Local fiscal revenue includes local fiscal budget revenue and extra-budgetary income,of which local fiscal budget revenue is the main source.Local fiscal budget revenue consists of local fiscal tax revenue and local fiscal non-tax revenue.Local fiscal tax revenue includes: value-added tax,corporate income tax,personal income tax,property tax,stamp duty,etc.Local fiscal non-tax revenue includes special income,administrative income,and fines and confiscation.The local fiscal revenue not only reflects the specific effects of the local financial system operation and the implementation of the income system,but also reveals the inherent changes of local economic growth,which is an important indicator for judging the level of regional economic development.Combining theoretical research and empirical analysis,the comprehensive evaluation index system of local fiscal revenue selects 19 representative indicators,and then uses principal component analysis and K-means cluster to analyze the local fiscal revenues of different provinces and cities nationwide.Studying the main factors affecting fiscal revenue,the final most influencing factors affecting local fiscal revenue are VAT,corporate income tax,personal income tax,urban maintenance and construction tax,real estate tax,stamp duty,land value-added tax,vehicle and vessel tax,deed tax,special The “real economy tax” factor consisting of 12 factors including income,fines and no income,and other non-tax revenues.Secondly,the“state-owned capital and land income” factor consists of five factors: urban land use tax,cultivated land occupation tax,administrative business fee income,state-owned capital operating income,and state-owned resources paid use.According to the results of principal component analysis,the financial revenues of different provinces and cities are ranked and classified.The kernel principal component analysis and kernel K-means clustering are introduced.The polynomial kernel function and the Gaussian kernel function are taken as examples to find the optimal kernel function parameters,and the different types of kernel functions and corresponding parameters are analyzed for feature extraction performance and clustering.The effect of the effect.Finally,it is concluded that when the number of principal elements is consistent,the range of data research increases from linear to nonlinear,which makes the feature extraction using kernel principal component analysis better than the principal component analysis.It is verified by Silhouette Value that the effect of kernel K-means clustering is better than that of K-means clustering when the number of classifications is the same.