Research on sampling method of tax-checking based on neural network

摘要

It is a core component of the Golden Tax Project that the application of information technology supports the tax-checking. According to some problems of inefficiency and poor accuracy in tax-checking sampling practices, learning from the current tax-checking sampling study, selects financial indicators of tax-checking sample of the value-added tax (VAT) based on gradually discriminant analysis (GDA), has a better solution to discriminant classifier of the “honest tax group” and “dishonest tax group”, and then using the technology of self-organizing map neural network (SOM), builds a intelligent analysis model on VAT sampling; Finally, uses the real data of 43 enterprises as an example to test, Finally, the use of 43 actual business data as an example the test, and the results of discriminant analysis were compared with that of statistical analysis, and the results show that the sampling effect of BP nets is remarkable.