In last decade, the lasso (least absolute shrinkage and selection operator) estimator is one of the most famous shrinkage and selection techniques in the multiple linear regression. Sometimes some prior knowledge like linear constraints on some important factors is at hand. Therefore, one may restrict the lasso estimator to combine this information within the lasso estimator. Since the lasso shrinks the coefficients already, a double shrinkage estimator is born.
Bienvenue à tous!