OLS
Ordinary least squares: the estimator that picks the coefficients minimising the sum of squared residuals, .
When to use
The closed form is , though scikit-learn uses numerical optimisation instead of inverting the matrix. OLS predictions do not change with feature scaling.
See DS Lec 4.