Research On Variable Selection In Spatial Econometrics

As we all know,it is very important to select variables effectively in the study of economic problems.The selection of effective inbreeding and dependent variables is also the first step in establishing economic models to analyze the problem.At present,many scholars’ research on variable selection in spatial econometrics mainly focuses on Bayesian variable selection method and classical AIC method.In this paper,the spatial nesting variable selection model(GNS)is further extended to the spatial nesting variable selec-tion model(GSM)with residual spatial lag term by using the generalized information criterion.Monte Carlo is used to simulate the effect of GSM model on variable selec-tion.Through comparison,it can be seen that GSM model can more effectively avoid the misselection and missed selection of variables.Secondly,this paper introduces three spatial intersection selection methods of Zeller,Occam Window and Gibbs sampling in Bayes method,and uses these three methods to effectively analyze the simulation data From the data analysis results,it can be seen that Zeller,Occam and Gibbs sampling can also reduce the misselection and missed selection of variablesIn the first chapter,the background,theoretical significance and research status of shellfish variable selection method and GIC method are introduced at first.Finally,the main research work of this article is briefly introducedIn the second chapter,the generalized embedded spatial variable selection model is further extended to GSM model,and the effect of GSM model is analyzed and demon-strated by using the simulation data of futures tradingIn the third chapter,the three variable selection methods of Zaller Occam Win-dow and Gibbs sampling are briefly introduced,with emphasis on analyzing the effect of Bayesian variable selection by considering different β and α values.