Variable Selection For Exponential Family Distribution In Regression Model Under The Regularization Parameter

At present, analysis of high-dimensional data processing is the study of hot issues in statistics.The variable selection method is an effective method to deal with high dimensional data.Elastic Net method is the selection method for high dimensional data of variables,especially the micro array data, it can choose the strong related variables, and parameter estimation at the same time.Dantzig Selector(DS) method is for the sample size is much smaller than the dimension of the data type and variable selection method,it can select the variable and parameter estimation at the same time.This paper describes the Elastic Net and DS method, properties and application of Elastic Net regression model, Adaptive distribution of Elastic Net and Adaptive DS index of regular parameters. The main research contents and results are as follows:One is the selection of exponential family distribution regression model Elastic Net method variables new regularization parameter, group effect property distribution regression model Elastic Net method of the regularization parameter exponential family under the Adaptive Elastic Net, Oracle properties and methods; through the numerical simulation to verify the feasibility of Elastic Net method.Two is the choice of exponential family distribution regression model of ADS method variables new regularization parameter, Oracle property distribution regression model ADS method of regularization parameter exponential family under, and through the numerical simulation to verify the feasibility of this method.