Variable Selection Based On Longitudinal Survival Data Model

Vertical and survival data are often found in the field of industry,psychology,economics,medical research and biological research.The joint model of the two datas is a very common model in medical research.The number of variables needed to be measured in the actual study is very large,and the number of variables that we need to estimate in this huge measurement also increases.In order to improve the accuracy of establishing joint models,we must choose the most explanatory covariates that have a significant impact on the response variables.When dealing with more variables and other data,the penalty likelihood method is less than that of the traditional variable selection method.The model is more concise,and the running time of the program is shorter,and the fitting and prediction effect is better.In this paper,the MLE method,Lasso penalty,alasso and SCAD penalty are used to select the variable of the vertical survival data model.The EM algorithm is used to solve the penalty likelihood estimation.Finally,the maximum likelihood estimation,Lasso penalty,alasso and SCAD penalty are used to estimate the parameters of the model and select variables.