Manual color scales in ggplot

[Eliot Miller]

Hi all,

Apologies, but I know that this is a simple question. I can't seem to get what I want to happen. I have a set of polygons, and I have assigned values to them outside of ggplot, then converted those values to r,g,b colors. I have a data frame that schematically looks something like this:

test <- data.frame(plot=c(1,1,1,1,1,2,2,2,2,2), lat=c(1,1,2,2,1,3,3,4,4,3), long=c(1,2,2,1,1,3,4,4,3,3), poly.color=c(rep("#CE3B57",5),rep("#BA5D58",5)))

It would seem to me that I should be able to plot it like this:

test.plot <- ggplot(test) +
    geom_polygon(aes(x=long, y=lat, group=plot)) +
    scale_fill_manual(breaks=plot, values=poly.color)

Or perhaps like this:

test2 <- data.frame(plot=unique(test$plot),color=unique(test$poly.color))

test.plot <- ggplot(test) +
    geom_polygon(aes(x=long, y=lat, group=plot)) +
    scale_fill_manual(breaks=test2$plot, values=test2$color)

But neither of those work. What am I not understanding here?

Thanks for any help you can offer!

Cheers,
Eliot

[Honglong Wu]

can't understand what you want to do.

maybe you can try the following code: ggplot(test) + geom_polygon(aes(x=long, y=lat, group=plot,fill=poly.color))

--
--
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.

[Eliot Miller]

Hi, thank you very much for the response! This is what I want with a major caveat. Those codes I provided (e.g. "#CE3B57") are actual codes for colors. I want the polygons filled with the provided colors. You can see what I mean with this modified code:

test <- data.frame(plot=c("one","one","one","one","one","two","two","two","two","two"), lat=c(1,1,2,2,1,3,3,4,4,3), long=c(1,2,2,1,1,3,4,4,3,3), poly.color=c(rep("#CE3B57",5),rep("#BA5D58",5)))

test.plot <- ggplot(test) +
    geom_polygon(aes(x=long, y=lat, group=plot, color=factor(plot))) +
    scale_fill_manual(values=c("one"="red","two"="blue"))

The issue is that in real life, I have a color gradient I have made manually (I want to directly control each r, g, b element in the code so I don't want to give that control up to ggplot) with hundreds of levels. I do not want to manually type out "one"="#CE3B57", "eighthundredfiftyseven"="#CE3B55", etc.

Does that make sense?

Thanks!
Eliot

[Eliot Miller]

Hi all, sorry for the multiple posts. I eventually solved this. It's pretty simple.

Make a vector of the colors you want to plot. For instance, following the original example (I haven't figured out why in this test case I need "as.character" while in my real example I don't, but it doesn't seem important):

your.colors <- as.character(test$poly.color)

Then give the elements of that vector names that correspond to the polygon elements. Then just use it like below:

names(your.colors) <- test$plot

test.plot <- ggplot(test) +
    geom_polygon(aes(x=long, y=lat, group=plot, fill=factor(plot))) +
    scale_fill_manual(values=your.colors)

[Dennis Murphy]

Hi:

Here are two ways you could go about this, but I would slightly
rewrite your data frame code to avoid automatic conversion of color
strings to factor.

test <- data.frame(plot=c(1,1,1,1,1,2,2,2,2,2),
lat=c(1,1,2,2,1,3,3,4,4,3),
long=c(1,2,2,1,1,3,4,4,3,3),

poly.color=rep(c("#CE3B57", "#BA5D58"), each = 5),
stringsAsFactors = FALSE)


1. Use scale_fill_manual() as you did in your later posts.


2. A less well-known feature is scale_*_identity(), where the colors
themselves are the breaks and values (and the default labels. This
also works:

ggplot(test) +
geom_polygon(aes(x=long, y=lat, fill= poly.color)) +
scale_fill_identity(guide = "legend", labels = c(1, 2)) +
labs(fill = "plot")

By default, scale_identity() does not produce a guide, so you have to
specify the type of guide you want before you can change any of the
scale's properties (in this case, the labels).

Dennis

[Eliot Miller]

This is great, thanks Dennis!