Scikit-Learn offers numerous state-of-the-art models for prediction (regression and classification). Linear models (e.g. Ridge, Logistic Regression) are the simplest of these models. They have pratical benefits such as interpretability and limited computation time while offering the best performance for some applications. This talk will cover the basics of these models with examples and demonstrate how they can scale to datasets that do not fit in memory or how they can incorporate simple polynomial non-linearities.