What does L1 (Lasso) regularisation do that L2 (Ridge) does not?
- A Drives some coefficients to exactly zero, performing feature selection
- B Handles correlated features better
- C Always improves accuracy
- D Speeds up training
Answer
Drives some coefficients to exactly zero, performing feature selection
L2 shrinks coefficients towards zero without eliminating them and distributes weight among correlated features more gracefully.





