Introduction linear regression with gradient descent This tutorial is a rough introduction into using gradient descent algorithms to estimate parameters (slope and intercept) for standard linear regressions, as an alternative to ordinary least squares (OLS) regression with a maximum likelihood estimator.