본 논문에서는 베이지안 변수선택법들 중 SSVS, GVS, KM방법이 일반화선형모형인 로지스틱 회귀모형에서 어떤 차이를 보이는지 비교 분석하였다. 특히, 지시변수의 사전분포와 관측치의 크기가 미치는 영향에 주안점을 두고 살펴보았다.
Specification of linear predictor for a generalized linear model requares determining which variables to include. In this paper we discuss Bayesian methods for performing this variable selection. In particular we focus on logistic regression in generalized linear model. And the Bayesian variable selection methods which we consider are implemented using the WinBUGS software. The three methods (SSVS, GVS, KM) expand the usual regression equation to an equation that incorporates all possible subsets of predictors by adding indicator variables. And the vector of indicator variables dictates which predictors to include. The posterior distribution of the indicator vector is approximated by means of the Markov chain Monte Carlo. Then we select subsets with high posterior probabilities. To compare with three Bayesian variable selection methods, we apply this methods for logistic regression model by changing prior distribution of indicator variable and the number of observations about simulated data. Overall three methods`s simulation give similar results than definite differences. So to compare with three methods more objectively, we need to study more about prior distribution and also take account of the simulation design procedures.