Study On Application Of Random Forest In Mass Appraisal Of Second-hand House

In recent years,Chinese real estate market is strongly hot and the price of new home market is rising year by year,resulting in many buyers transfer their attention to the second-hand housing market has become more active.With the increase of trading quantities of second-hand housing,the demand of registration of real estate,bank mortgages and tax-collection,prompting the need for mass appraisal technology.Therefore,through the analysis of mass appraisal technology,this article aims to explore a set of mass appraisal model of second-hand housing of high precision,high speed and strong applicability,combined with domestic actual situation of second-hand housing market.This article,guided by the market comparison method,analyzes factors of second-hand price from the public,building factors,regional factors and other factors and determines factors of model which can be used to search for comparable buildings and to build the mass appraisal model.In the process of finding comparable cases,the strong classification ability of support vector machine find the most comparable three buildings in same area of assessment,dividing smaller homogeneous area from the buildings of the same area of assessment.Considering the nonlinear relationship between the characteristics and the price,this article selects random forest to build the model of mass appraisal because it's good stability and not easily overfitting.And then,this article expounds the superiority of the model through introducing the principle of random forests.We build the model with 194 actual transactions of comparable buildings and confirm two important parameters,then,we use the model to estimate 19 cases which are from estimated building.According to results of the model show that the range of estimating error is within the specification requirements.The model which is running quickly and stable strongly,can be used to evaluate the price of mass appraisal of second-hand house.