Answer: The ridge is known as the L2 regularization because of the second-order natures of the shrinkage term, and it helps avoid situations where the parameter estimates are offsetting, with one having a large positive value and another a large negative value.

## Explanation Let's analyze each option: **Option A**: Incorrect. While LASSO is indeed L1 regularization (due to the L1-norm penalty), it doesn't just reduce parameters closer to zero - it can set them exactly to zero, effectively performing feature selection. **Option B**: Incorrect. Ridge regression is L2 regularization, not L1. Also, ridge regression shrinks parameters toward zero but doesn't set them exactly to zero. **Option C**: Incorrect. LASSO is L1 regularization, not L2. The L2 regularization refers to ridge regression. **Option D**: **Correct**. Ridge regression is L2 regularization (due to the L2-norm penalty term) and helps address multicollinearity by preventing parameter estimates from becoming too large with offsetting positive and negative values. This is particularly useful when dealing with highly correlated features. ### Key Differences: - **LASSO (L1)**: Uses absolute value penalty, can set coefficients exactly to zero, performs feature selection - **Ridge (L2)**: Uses squared penalty, shrinks coefficients toward zero but not exactly to zero, handles multicollinearity well Ridge regression's L2 penalty helps stabilize coefficient estimates when features are correlated, preventing the scenario where one coefficient becomes very large positive while another becomes very large negative to compensate.

Author: LeetQuiz Editorial Team