How to change labels (legends) in ggplot?

Question

My code is like below, I want to change the label of the ggplot, but R always remind me:

Error in unit(tic_pos.c, "mm") : 'x' and 'units' must have length > 0

What should I do?

ggplot(mat,aes(x=sales,col=type))+
  geom_density()+labels("red_sold","blue_sold","yellow_sold")

Solution 1

Is mat$type a factor? If not, that will cause the error. Also, you can't use labels(...) this way.

Since you did not provide any data, here's an example using the built-in mtcars dataset.

ggplot(mtcars, aes(x=hp,color=factor(cyl)))+
  geom_density()+
  scale_color_manual(name="Cylinders",
                       labels=c("4 Cylinder","6 Cylinder","8- Cylinder"),
                       values=c("red","green","blue"))

In this example,

ggplot(mtcars, aes(x=hp,color=cyl))+...

would cause the same error that you are getting, because mtcars$cyl is not a factor.

Solution 2

As a complement to @jlhoward's answer, scale_color_manual() is more intended to customize the color scale (the actual colors that will be displayed).

For your case, you might rather use scale_color_discrete():

ggplot(mtcars, aes(x=hp,color=factor(cyl)))+
    geom_density()+
    scale_color_discrete(name="Cylinders",
                         labels=c("4 Cylinder","6 Cylinder","8- Cylinder"))

This is quicker, but it is dependant on the factor levels order, which might lead to irreproducibility. You might want to specify the breaks argument to minimise the risk of error (and customize the order in the legend):

ggplot(mtcars, aes(x=hp,color=factor(cyl)))+
    geom_density()+
    scale_color_discrete(name="Cylinders",
                         breaks=c(8,6,4),
                         labels=c("8 Cylinder","6 Cylinder","4 Cylinder"))

More info on https://ggplot2-book.org/scales.html.

Comments

[Dan Chaltiel]

Of note, it is possible to change name and labels with keeping the defaults values by using scale_color_discrete() instead.

[Fabian Winkler]

@DanChaltiel Maybe you could post your comment as a regular answer, so it's easier to find?