Akaike Information Criterion And Its Application In Econometrics

Akaike information criterion has many important applications in the choice of econometric models,and it also has a wide range of applications in the fields of systems engineering and meteorology,this paper will be written from the information theory to the revolutionary innovation of Akaike information criterion,the way of model selection is the maximum likehood estimator before Akaike information criterion, which can connect the sample with overall together,and via the likelihood function to determine the optimal model, it becomes one link for sample analysis of the overall. But it is not the most economic way in the view of information entropy theory, the more greater the likelihood function means the more number of parameters, but it is not economy and strong practicality operability, and Akaike information criterion combined with entropy principle and the KL distance solves those problems, which can achieve a balance between model estimation accuracy and the number of parameters;then the paper will use the Monte Carlo simulation method on Akaike information criterion distribution in different sample, finding that the estimate effect of Akaike information criterion in large sample is not ideal, so it is necessary to introduce Bayse information criterion, as well as in different data skewness types for improved ICOMP criterion, and the MRC criterion for over fitting model; In the end the paper uses empirical analysis in the independence test,analysis of variance,factor analysis,and compare with other methods,finding that Akaike information criterion can not only accurately determine,but also avoid too much subjectivity,which makes that from art to science.