ChatGPT解决这个技术问题 Extra ChatGPT

How can I change the Y-axis figures into percentages in a barplot?

https://i.stack.imgur.com/wuK3a.jpg


m
mkirzon

Use:

+ scale_y_continuous(labels = scales::percent)

Or, to specify formatting parameters for the percent:

+ scale_y_continuous(labels = scales::percent_format(accuracy = 1))

(the command labels = percent is obsolete since version 2.2.1 of ggplot2)


I liked that you don't have to type library(scales) for this.
And the reason scales::percent(accuracy = 1) does not work is because the *_format() versions create a function instead of...whatever percent() alone creates, is that correct?
D
Deena

In principle, you can pass any reformatting function to the labels parameter:

+ scale_y_continuous(labels = function(x) paste0(x*100, "%")) # Multiply by 100 & add %  

Or

+ scale_y_continuous(labels = function(x) paste0(x, "%")) # Add percent sign 

Reproducible example:

library(ggplot2)
df = data.frame(x=seq(0,1,0.1), y=seq(0,1,0.1))

ggplot(df, aes(x,y)) + 
  geom_point() +
  scale_y_continuous(labels = function(x) paste0(x*100, "%"))

+1 for no external dependency. I know that since Hadley is the author of both ggplot2 and scales, it shouldn't really matter—but this solution is still appreciated.
M
Michael Ohlrogge

ggplot2 and scales packages can do that:

y <- c(12, 20)/100
x <- c(1, 2)

library(ggplot2)
library(scales)
myplot <- qplot(as.factor(x), y, geom="bar")
myplot + scale_y_continuous(labels=percent)

It seems like the stat() option has been taken off, causing the error message. Try this:

library(scales)

myplot <- ggplot(mtcars, aes(factor(cyl))) + 
          geom_bar(aes(y = (..count..)/sum(..count..))) + 
          scale_y_continuous(labels=percent)

myplot

D
DanM

Borrowed from @Deena above, that function modification for labels is more versatile than you might have thought. For example, I had a ggplot where the denominator of counted variables was 140. I used her example thus:

scale_y_continuous(labels = function(x) paste0(round(x/140*100,1), "%"), breaks = seq(0, 140, 35))

This allowed me to get my percentages on the 140 denominator, and then break the scale at 25% increments rather than the weird numbers it defaulted to. The key here is that the scale breaks are still set by the original count, not by your percentages. Therefore the breaks must be from zero to the denominator value, with the third argument in "breaks" being the denominator divided by however many label breaks you want (e.g. 140 * 0.25 = 35).