Description

This scatterplot depicts a case where the data show several order violations. The animation shows how the isotonic regression algorithm resolves the order violations by taking an average going back in variable x until the order violation is resolved. The severity of monotonicity means the algorithm has to go back more steps in predictor x to compute a mean that resolves the order violation. The resulting regression curve begins to look more like a step function than a smooth curve, as shown in the first isotonic animation.

Richard Gonzalez, University of Michigan

gonzo at umich dot edu

Click here to give feedback on this site or report problems.