Simple visually-weighted regression plots

There has recently been a lot of discussion of so-called “visually-weighted regression” plots.

Folk hero Hadley Wickham suggests that such plots would be easy to implement with ggplot2, and so I have attempted to prove him right.

The approach outlined in the following Gist would be easy to apply to any situation in which you have a matrix of replicated predictions or bootstrapped fits from a model — any such a matrix would just take the place of the simYhats object.

The end-product of this example

By d-sparks

Tags: MASS ggplot2 graphics reshape2 rstats