A data scientist wishes to ?t a linear model and has 100 predictors on hand. For simplicity of deployment, she wants no more than 10 of the predictors to have non-zero coeffcients. Suggest a subset selection method and a shrinkage method that could be used here.