Study On The Price Evaluation Model Of Hangzhou Second-hand Housing

With the rapid development of the real estate industry,the second-hand housing transaction market is growing,and the quality and speed requirements for second-hand housing price assessment are significantly improved.Rapid batch valuation can not only provide a reasonable reference for the market,but also reduce the credit risk in real estate commercial loans,prevent the "yin and yang contract"from appearing,and prepare for the subsequent real estate tax reform.However,the existing literature on the residential price evaluation model is mostly based on the second-hand house listing price,can not reflect the real transaction price,and does not systematically summarize the characteristic factors affecting the real estate price.The index selection is mainly based on the listing data,and there is a certain one-sidedness and Subjectivity.Based on the above realistic theory and demand background,this paper first systematically reviews batch evaluation,feature price theory and methods.Subsequently,on the basis of extensive reading of the literature,the main factors affecting real estate prices were summarized.At the same time,based on the actual situation of the housing market in Hangzhou in 2018,21 characteristic variables were selected to construct the evaluation index system.Finally,15 feature variables are selected for model construction based on correlation analysis and feature variable importance evaluation.Based on this,this paper selects traditional multiple regression model and support vector regression model in machine learning and random forest model to construct second-hand housing.Price evaluation model.The study found that:(1)The most important factors affecting the price of second-hand housing in Hangzhou are the distance from the city center,the key school district,the building area,the property fee,and the university.The more important factors include the age of the house,the subway,the kindergarten,the top three hospitals,and the total floor.,commercial complex,floor area ratio,completed subway,floor,general factors are greening rate,under construction and planning subway,floor score 1,floor score 2.(2)The traditional multiple linear regression model has large error.The random forest model has the highest matching accuracy and the best goodness of fit.In summary,the random forest model has a low false positive rate and good stability,and is suitable for promotion in the field of real estate batch automatic evaluation.