The Research About Evaluation Model Of Compilation Method Of The House Price Index For New Building In China

China have a dozen national or regional HPI(house price index)currently.It has initially formed system and made an important contribution to real estate market after continuous development and improvement of the index itself.But at the same time the situation that the HPI which have large differences ofen appear in real estate market.And the published data often make people feel distortion.Obviously it is disadvantageous to the stability and healthy development of the real eatate market.But the current domestic research on index evaluation system is very few,and needs to be further discussed.Therefore,to build a comprehensive evaluation system and establish a quantitative standard about the HPI methodology has important theoretical significance and social practical significance.The most difficult point is that there is no accurate index reference value for the evaluation system of HPI.This paper based on the the simulation theory and the basic statistical theory by designing HPI and combined with statistical distribution to generate simulated data.And then establish the evaluation model by comparing the designed HPI and the HPI which cacluated from the simulated data by different compilation method.In data simulation stage,this paper discusses the statistical distribution of house price and the characteristic variables and establish the simulation model based on the commercial bank lending data of Xiangyang city in 2014.Firstly the differences between ZFHPI(Zhongfang house price index which published by China Real Estate Index System)and GFHPI(Guofang house price index which published by National Bureau of Statistics)were compared and analyzed.Secondly,the decision trees were established to have preliminary understanding of the original data after data preprocessing.Then use the kernel density estimation to explore the distribution density of the characteristic variables.thirdly,linear mixed effects model were constructed to analyse the interactive effect of located floor with the total floor and its impact on prices.Through establish the semiparametric mixed effect model and discover that good results have been achieved in the housing price forecast period,which have solved the problem that the goodness of fit of linear mixed effects model are not very good.The transaction information can be simulated by the statistical distribution and the simulation model when the initial house price index is designed.We also establish a database to manage the data effectively.In order to ensure the degree of anastomosis of the simulated data and the actual trading market when evaluate the HPI,this paper tests the simulation model and the simulated data at last and finds that the simulation results is satisfactory.After the inspection of simulation data,we compare the calculation of the HPI with the initial specified HPI through repetition of sampling calculation by using different sampling method.This paper first adopts the mean absolute error and time series analysis methods to evaluate the index.Then using analytic hierarchy process(AHP)to establish the HPI evaluation system and put forward the improved method based on advantage for identifying dynamic weight coefficient.Then evaluate the two HPI and give the conclusion that GFHPI is better than ZFHPI in whole at the end of the section,but both compilation model needs to be improved at the same time.The evaluation methods also provides a quantitative evaluation criterion for other various HPI methodology.Finally,the research results and innovation points are summarized.And then explore the problem that should be delve into to a certain extent.