Which method also penalizes coefficients, but can shrink some coefficients all the way to zero, performing feature selection?

Which method also penalizes coefficients, but can shrink some coefficients all the way to zero, performing feature selection?

• A. Ridge
• B. Features
• C. LASSO
• D. Outliers

Answer: (C)

This item is about regularization that promotes sparse models. LASSO uses an L1 penalty on the coefficients, which not only shrinks them but can push some to exactly zero when the penalty is strong enough. Those zero coefficients effectively remove those features, giving you automatic feature selection. Ridge regression uses an L2 penalty and tends to shrink coefficients toward zero but not to exactly zero in practice, so it doesn’t perform feature selection. The remaining options don’t describe a coefficient-penalizing method tied to sparsity.

This item is about regularization that promotes sparse models. LASSO uses an L1 penalty on the coefficients, which not only shrinks them but can push some to exactly zero when the penalty is strong enough. Those zero coefficients effectively remove those features, giving you automatic feature selection. Ridge regression uses an L2 penalty and tends to shrink coefficients toward zero but not to exactly zero in practice, so it doesn’t perform feature selection. The remaining options don’t describe a coefficient-penalizing method tied to sparsity.