Variable selection for skew-normal mixture of joint location and scale models

Abstract Although there are many papers on variable selection methods based mean model in the finite mixture of regression models, little work has been done how to select significant explanatory variables modeling variance parameter. In this paper, we propose and study a novel class models: skew-normal joint location scale models analyze heteroscedastic data coming from heterogeneous population. The problem for proposed is considered. particular, modified Expectation-Maximization(EM) algorithm estimating parameters developed. consistency oracle property penalized estimators established. Simulation studies conducted investigate sample performance methodologies. An example illustrated by