filling up to the edge + 'where' is missing

[Ross Boylan]

I want to make a graph of a linear probability model, coloring the
background red for areas where the probability is an impossible value.

nofly <- data.frame(x=c(-1, -1, 3, 3, -1, -1, 3, 3),
y=c(-1, 0, 0, -1, 1, 2, 2, 1),
id=rep(1:2, each=4))
g <- ggplot(data=data.frame(x=0, y=.2), aes(x=x, y=y)) +
scale_y_continuous("Probability", limits=c(-1, 2)) +
scale_x_continuous(limits= c(-1, 3)) +
geom_point() +
geom_polygon(data=nofly, fill="red", alpha=0.3, aes(group=id)) +
geom_abline(intercept=0.2, slope=0.8)

This basically works, but there are 2 annoyances.

First, the area that is red does not extend out to the x limits of the
background, which go beyond the explicit limits I set. Is there a way
to get it to extend all the way?

Second, I didn't really want to plot the point. But everything I tried
without it produced the error
> g <- ggplot(data=data.frame(x=0, y=.2), aes(x=x, y=y)) +
+ geom_abline(intercept=0.2, slope=0.8)
> g
Error in as.environment(where) : 'where' is missing

Is there a way to get around that? A search found quite a few messages
discussing such error messages, but they seemed to have different causes
than mine (though perhaps just on the surface).

Thanks.
Ross Boylan

[Dennis Murphy]

The argument expand = in scale_*_continuous() controls the padding;
setting expand = c(0, 0) removes it. If you don't want points, don't
use geom_point(). Here's the simplest way I know of to reproduce your
plot sans the point:

nofly <- data.frame(x=c(-1, -1, 3, 3, -1, -1, 3, 3),
y=c(-1, 0, 0, -1, 1, 2, 2, 1),
id=rep(1:2, each=4))

library(ggplot2)
ggplot() +

geom_polygon(data=nofly, fill="red", alpha=0.3,

aes(x = x, y = y, group=id)) +
scale_y_continuous("Probability", expand = c(0, 0)) +
scale_x_continuous(expand = c(0, 0)) +
geom_abline(intercept=0.2, slope=0.8, size = 1)

The aesthetics you want are in nofly, so you can define them in the
geom_polygon() layer. It is legitimate to pass an empty argument list
to ggplot(), which will set up an empty base layer. Since
geom_abline() doesn't explicitly need a data frame when the intercept
and slope are vectors of length 1 (i.e., they are set rather than
mapped), you have everything you need. All I did was add a size
argument to geom_abline() to make it a bit thicker.

The following also works because geom_abline() doesn't use any of the
aesthetics defined in the ggplot() call and only one data frame is in
play:

ggplot(data=nofly, aes(x = x, y = y, group=id)) +
geom_polygon( fill="red", alpha=0.3) +
scale_y_continuous("Probability", expand = c(0, 0)) +
scale_x_continuous(expand = c(0, 0)) +
geom_abline(intercept=0.2, slope=0.8, size = 1)


Dennis

> --
> --
> You received this message because you are subscribed to the ggplot2 mailing
> list.
> Please provide a reproducible example:
> https://github.com/hadley/devtools/wiki/Reproducibility
>
> To post: email ggp...@googlegroups.com
> To unsubscribe: email ggplot2+u...@googlegroups.com
> More options: http://groups.google.com/group/ggplot2
>
> --- You received this message because you are subscribed to the Google
> Groups "ggplot2" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to ggplot2+u...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.