Manual color and linetype, and additional geom with fixed color

[Sebastian Schubert]

Hi,

I have many lines in one plot and to differentiate between them I want to combine color and linetype following the suggestion here:

http://stackoverflow.com/a/11344735/1463740

This works fine. However, I want to add another geom with a fixed color. This color overwrites all other colors, but only when using the combination of color and linetype as described above. Here is an example:

#############################

library(ggplot2)

library(RColorBrewer)

d <- data.frame(x=c(rep(1,10), rep(2,10)),

                y=c(seq(1,10), seq(2,11)),

                model=letters[seq(1, 10)])

p1 <- ggplot(d, aes(x=x, y=y, color=model)) + geom_line()

p2 <- ggplot(d, aes(x=x, y=y, color=model, linetype=model)) + geom_line() +

    scale_linetype_manual(values = c(rep("solid", 5), rep("dashed", 5))) +

    scale_color_manual(values = c(brewer.pal(5, "Paired"), brewer.pal(5, "Paired")))

## similar colors

p1 

## works as expected but similar colors

p1 + stat_summary(fun.y=mean, colour="black", geom="line")

## easier to distinguish 

p2 

## all elements black 

p2 + stat_summary(fun.y=mean, colour="black", geom="line")

#############################

Is this a bug or a feature? What can I do?

Thanks a lot!

Sebastian

[Ista Zahn]

I would consider that a bug. Fortunately it's easy to work around:

ggplot(d, aes(x=x, y=y, color=model, linetype=model)) +

stat_summary(fun.y=mean, color="black", geom="line") +

geom_line() +
scale_linetype_manual(values = c(rep("solid", 5), rep("dashed", 5))) +
scale_color_manual(values = c(brewer.pal(5, "Paired"),
brewer.pal(5, "Paired")))

Note that the order matters--geom_line() has to come after stat_summary().

Best,
Ista

> --
> --
> 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.

[Sebastian Schubert]

As far as I can see, this solves the color issue but there is no line from

stat_summary(fun.y=mean, color="black", geom="line")

anyway. I've just realized that

p2 + stat_summary(fun.y=mean, colour="black", geom="line")

in my example does also not create the line...

So apparently not only the color is a problem here.

Thanks
Sebastian

[Ito, Kaori (Groton)]

Maybe the easiest fix for this problem is calculate mean outside:

tab <- ddply(d, .(x), summarize, y=mean(y))
p2 + geom_line(data=tab, aes(color=NULL, linetype=NULL))

[Ista Zahn]

I take it back, the only bug was in my brain, probably due to
insufficient coffee. The issue is simply that stat_summay is creating
separate lines for each model, because of the linetype mapping. You
can make it create only a single line by overriding the group mapping,
like this:

p2 + stat_summary(aes(group = 1), fun.y=mean, colour="black", geom="line")

Best,
Ista

On Tue, Jun 24, 2014 at 6:53 AM, Sebastian Schubert
<schube...@gmail.com> wrote:

[Sebastian Schubert]

Hi Ista,

thanks a lot, this explains the behavior. Maybe more consistent is to
explicitly set the linetype as well:

p2 + stat_summary(fun.y=mean, colour="black", linetype=1, geom="line")

Best wishes,
Sebastian