Penalty, Shrinkage and Pretest Strategies:Variable Selection and Estimation