Question

I would like to rename the values in a legend without altering the custom colors that have already been set. Is there a way to set legend labels without using scale_color_manual? Currently, I have something like this:

norm <- rnorm(1000, 0 , .5)
gam <- rgamma(1000, 2)
beta <- rbeta(1000, 2, 3)
dist <- data.frame(Normal = norm, Gamma = gam, Beta= beta)
dat <- melt(dist, variable.name = "Distribution", value.name = "XValue")
plot1 <- ggplot(dat, aes(XValue, color = Distribution)) +
            stat_density(geom = "path", position = "identity", size = 2) +
            scale_color_manual(values = c("yellow", "black", "forestgreen"))

plot2 <- plot1 + scale_color_discrete(labels = c("Distribution 1",
                                "Distribution 2",
                            "Distribution 3"))

This however overwrites the manual colors. I will be changing the names in a different function from where I am setting colors so, unfortunately, I will not be able to use scale_color_manual( values =… , labels = …). The other option I thought of is to somehow get the colors used in plot1. I could then do something like:

colors <- plot1$colors_used
plot2 <- plot1 + scale_color_manual(labels = c("Distribution 1", 
                                               "Distribution 2",
                        "Distribution 3"),
                                      values = colors)

Any help would be much appreciated. Thanks!

Answer 1

It is possible to specify the label names in scale_colour_manual.

ggplot(dat, aes(XValue, color = Distribution)) +
  stat_density(geom = "path", position = "identity", size = 2) +
  scale_color_manual(values = c("yellow", "black", "forestgreen"),
                     labels = c("Distribution 1",
                                "Distribution 2",
                                "Distribution 3"))