Regularization: Lasso vs Ridge

Your housing price model is overfitting — it nails the training data but falls apart on new houses. The culprit: too many features and not enough constraint. Regularization is the fix, but you have two options. Ridge (L2) shrinks all coefficients evenly. Lasso (L1) drives some all the way to zero, effectively selecting features for you. Train both, compare their coefficients, and see which features Lasso considers irrelevant.