Adaptive Variable Selection For Multiple Response Longitudinal Datay

In economics,medicine,sociology and other fields of research work,often get the longitudinal data as a result,even the multiple responses of longitudinal data.This type of data can depict the characteristics of individuals in different times different indicators.In addition,the longitudinal data of variable selection can help researchers extract has important influence on the response variable indicators,resulting in a more effective model to explain the data.In this paper,the adaptive variable selection method is mainly used to discuss the longitudinal data joint regression analysis in the case of multiple responses.Secondly,Adaptive LASSO was used for parameter estimation and variable selection,and BIC criterion was used to discuss the selection of Adaptive parameters.Then,the joint modeling method in this paper is compared with the glmmLasso modeling method.Finally,data sets from the European depression study support our approach as a real data example.The results show that the method proposed in this paper has great advantages in terms of selection accuracy and estimation accuracy.Especially when the dimension of random effect increases,the results obtained by Adaptive LASSO joint modeling are more stable and closer to the real model than those obtained by independent modeling.