| The research on the regression model is an important part of mathematical statistics.The parameter regression model can not meet the flexibility of data in practical problems because of its fixed form.Non-parametric regression model,although it does not need to be selected in advance,and it reflects information more accurately,but it faces a problem of high dimension.However,the semi-parametric model has the advantages of both of the two models,and it contains both parameter and non-parametric components.It can not only reduce the risk of miscarriage of justice,but also overcome the " curse of dimensionality ",has become a hot issue in recent years.It is used in various fields widely.First,study on the semi-parametric,which combines the model of non-parametric regression model and spatial variable coefficient model,Then the two-step estimation of the model is proposed,and the accurate expression of the parameters of nonlinear function and spatial variation coefficient in semi-parametric model is given.Finally,the numerical simulation results show that the estimation value is good with the real value,and the precision of the method.The second,construct a model of Harbin commercial housing prices based on spatial variable coefficient regression model by using Harbin each district housing related data that from 2014 to 2015 is given.The influence factors of housing space difference in Harbin are discussed and analyzed.Research shows that schools,hospitals and greening parameters are the main factors which affect the spatial differences of housing prices.And parks,metros have smaller effects on housing prices.Besides,commercial centers and building ages can improve the housing prices.Finally,the spatial variable coefficient regression model is compared with the global space least square method,further shows the superiority of the spatial variable coefficient model in solving the spatial data relative to the ordinary model. |
PDF Full Text Request