Research On Random Forest Model In The Mass Appraisal Of Tax Base Of The Property

Since the reform of tax implemented in 1994, the financial powers of local government has benn shrinking while the powers of deciding things has extended., which force the local government to find new resources of revenue. Many scholars focus on the design of tax system instead of assessment of property tax base. In fact, assessment of property tax base is the important premise of collection property tax, so the study focus on the assessment of property tax base.The paper modelling from the experience of foreign tax base assessment, we study the assessment of tax base from the angle of consumer demand..Characteristics of current price theory which has been widely applied in the field of real estate price evaluation, the model is based on hypothesis of linear model. And the base of the large number of characteristic variables involved in th e assessment, linear hypothesis may cause error to the assessment.Finally, the study build random forest model to assess the tax base of Nangang in Harbin based on 200 groups of data collected from quotation data and data of electronic map measurement. Then compare the results of random forest, regression of traditional multiple linear and regression prediction of decision tree. The reference of random forest model to the research of mass appraisal of real estate tax base, enrich the study of price theory and practice aspects. Practices for tax base assessment of mass appraisal model, give;s a new reference of model prediction, the improved accuracy and stability of model is conducive to the tax base mass appraisal in the actual application of the promotion.