The Applicability Of The Mass Appraisal Of Real Estate Tax Base By Random Forest Method

In recent years,with the continuous improvement of our country's economic,house prices in China have shown a rising trend.According to the actual situation,our government made a series of adjustments to the real estate tax,and actively promoted the reform of real estate tax.In 2011,we carried out a pilot reform of personal housing levy tax in Shanghai and Chongqing.In November 2017,the Ministry of Finance for the first time clearly present the principle "legislation first,fully authorized,step by step",to promote the real estate tax legislation and implementation.The determination of the real estate tax base is an important prerequisite for the full levy of the real estate tax.Therefore,this paper hopes to study the evaluation method of the real estate tax base.As a nonparametric method,random forest method does not need preset function form,it can deal with complex and difficult to match nonlinear relationship samples,and has high prediction accuracy and good noise immunity.Based on the experience of real estate tax base assessment at home and abroad,the random forest method is introduced into the framework of hedonic price theory,and the mass appraisal model of real estate tax base based on random forest is established.This paper will establish the mass appraisal model of real estate tax based on random forest for the empirical study of Xiamen City,listing three ways of data through the network,geographic information system and field survey collected in Xiamen City 33 District 231 real estate data for modeling and prediction,the model predicts very well.Based on the prediction results of random forest model,it is compared with multiple linear regression and regression decision tree model,showing the advantage of random forest method in mass appraisal of real estate tax base.