Research On Variable Selection Of Two Types Of Spatial Econometric Models

With the rapid development of the times,people are increasingly accelerating the exploration of economic issues.It is found that the influence of spatial factors on economic issues is becoming more and more significant.Spatial statistical research has emerged at the historic moment and has attracted the attention of many scholars.Because of the fast development of Internet,people can get a large number of data samples from computers,and each sample contains a lot of information Selecting the most valuable information from it makes variable selection one of the hot issues in modern statistical analysis.This paper studies the variable selection of two spatial econometric models:Firstly,the Adaptive Lasso variable selection problem of the first-order spatial autoregressive error autoregressive model is studied,and the penalty of the parameter estimation of the SARAR model is imposed.Under certain conditions,it proves that the method has good asymptotic properties and consistency.It is verified that the way can effectively identify non-zero parameters and zero parameters by numerical simulation,and can well implement the variable selection of the SARAR model.Secondly,for the Adaptive Lasso variable selection problem of the partial linear space autoregressive model,under certain conditions,it is proved that the method has good asymptotic property and consistency.Through numerical simulation,it is also verified that adaptive lasso method can effectively identify non-zero coefficient and zero coefficient.The results show that the method can well realize the variable selection of partial linear space autoregressive models.