Dummy-Variable Trap
Perfect multicollinearity caused by keeping every dummy (one-hot) level of a categorical variable alongside an intercept; the fix is to drop one level per category.
When to use
In code, this is pd.get_dummies(..., drop_first=True), used in the regression pipeline of Lecture 4 and the insurance-charges problem set. Without it linear-regression coefficients are not identifiable.
See Lec 4.