639. R-Squared and Irrelevant Features
medium

A data scientist adds 20 new random noise features to a linear regression model. R² on the training set increases. Why is this misleading?