Discussion questions#
Suppose two predictors are highly correlated and both are related to the response.
How would OLS, Ridge, and Lasso behave in that situation?
Which method would you prefer?
Would your answer depend on whether your goal is prediction, interpretation, or variable selection?