Removing highly correlated features before linear modeling can:
Select an answer to reveal the explanation.
Short Explanation
Dropping correlated features can stabilize linear coeffs — trees care less.
Full Explanation
Method-dependent feature filtering.