Isotonic Regression is THE Coolest Machine-Learning Model You Might Not Have Heard Of

Isotonic regression is a free-form linear model that can be fit to predict sequences of observations. However, there are two major differences between isotonic regression and a similar model like weighted least squares. An isotonic function must not be non-decreasing. This is because an isotonic function is a monotonic function, meaning a function that preserves or reverses a given order. Isotonic regressors use an interesting concept called " Order Theory."