Enterprise tax risk prediction method

Abstract

The enterprise tax risk prediction method comprises the steps of calculating a prior probability matrix and a state transition matrix of a tax risk according to enterprise risk information and a total number of enterprises; calculating the risk state probability of the tax risk item according to the prior probability matrix and the state transition matrix; calculating a target number corresponding to the tax risk items; obtaining enterprise information, constructing a risk path link between an enterprise and a risk according to the enterprise information, and generating a multi-dimensional matrix according to the risk path link; and according to a target number corresponding to the risk item, screening out a target number of enterprises from the enterprises corresponding to the tax risk item according to a sequence of the enterprise risk item probabilities from high to low, and outputting the target number of enterprises as a prediction result. The embodiment of the invention provides a tax risk active prediction model based on an enterprise relation chain, and aims to solve the problems that a traditional tax risk prediction result is inaccurate, the enterprise risk quantity is separated from macroeconomy and the like.

Description

Technical field [0001] The present application relates to the field of electronic information technology, and more particularly to an enterprise tax risk prediction method. Background technique [0002] At present, tax risk is mainly based on tax information checking, which takes a lot of manpower and lacks grasp of the overall taxation, and due to the inspection of the reporting information, the taxation approach passes It is often impossible to obtain the risk of enterprises or regions in time. For example, the classic risk prediction method has neural network models such as decision tree, random forest, logic regression and other machine learning models. These models can have a certain prediction of certain types of tax risk through historical data, but it is difficult to obtain more accurate prediction results. . Inventive content [0003] In view of this, this embodiment provides a tax risk active prediction model based on the corporate relationship chain, which is used to ...

AI Technical Summary

Problems solved by technology

[0003] In view of this, this embodiment provides an active tax risk prediction model based on the enterprise relationship chain to overcome the inaccurate traditional tax risk prediction results and the disconnection between the number of enterprise risks and the macro economy.

Images