The Research On The Trend Of Domestic Real Estate And The Forecast Of Sales Price Index Based On ARMA Model

This paper made some research on the development trend of real estate on the background of domestic real estate industry in recent decades.and on this bases, we made some forecast analysis of sales price index on ARMA model.In the third part of this paper, we analysised and discussed the change of domestic town resident population,the investment and development of real estate,the development of demestic real estate industry respectively.We used Granger causality test, stability analysis,ADF test and so on,and got the conclusions as below:the added value,the growth and the GDP proportion of demestic real estate industry from 1992 to 2007 are non-stationary sequences,but the contribution of GDP is stationary sequence.Then,we made regression on these four sequences,the regression equations as below: X11=896193.2-449.4383t+1.105133 Xt-11+0.155901 Xt-21 +0.335704Xt-31-0.088315 Xt-41-0.097913Xt-51+0.912867Xt-61. Xt2=-3315.636+1.663423t-0.548474 Xt-12,-0.013403 Xt-22 +0.137781 Xt-32+0.340514 Xt-42. Xt3=0.893319+0.824929 Xt-13-0.125995 Xt-23-0.030363 Xt-33 +0.500336 Xt-43-0.357476Xt-53△2Xt4=7.386182-0.003622t+0.35567△2Xt-14-1.84024△Xt-14Prediction results showed that in the coming decades,the added value of real estate would increase rapidly;the growth of real present estate rise and fall and would reach 30% in 2050; the GDP proportion of real estate wave in the nearby of 4.75%;the contribution of GDP increase slowly,and would reach 6.688% in 2050.In the fourth part of this paper,we made fitting and forecast analysis of the real estate sales price index on 156 sets of data from 1998-01 to 2010-12.In the solving of this model,first of all,we made stability analysis and found this sequence is non-stationary. Then we made single whole analysis and found it's first-order single whole sequence.Secondly,we made correlation analysis on the first-order differential sequence of sales price index,got the conclustion of that it's autocorrelative.Finally,we estimated the parameters of this model,and this is the emphasis and difficulty of this part.We comed up with a new solving method—GA-LS arithmetic innovatly based on RLS-LS arithmetic and GW-LS Arithmetic. GA-LS arithmetic is a arithmetic which uses GA arithmetic to estimate the parameters of LS approximate structure in ARMA model.It's not only overcomed the problems of RLS-LS arithmetic effectively,but also got more actual results.We not only used RLS-LS arithmetic and GA-LS arithmetic to estimate and forecast respectively,but also compared the results of them and proved the advantages of GA-LS arithmetic. We got the following parameters:(?)=[0,0.1,0,0.2,0.12, 0.2196,0.03,0.0998,0,0.0118,0.1030,0,0,0,0],δ1t=-1.0000,δ2t=0.0000,φ1t=-1.The prediction results showed that the real estate sales price index would present slight fluctuations in the next five years if the real estate policy unchang.And it's present growth trend overall,which would reach 114 in 2020.