Display frequency instead of count with geom_bar() in ggplot

Question

On this page, they give the following example

library(ggplot2)
library(reshape2)
ggplot(data=tips, aes(x=day)) + geom_bar(stat="bin")

Instead of a count I'd like to have a frequency in y-axis. How can I achieve this?

Salado

Answer 1

Here's the solution which can be found in related question:

pp <- ggplot(data=tips, aes(x=day)) + 
      geom_bar(aes(y = (..count..)/sum(..count..)))

If you would like to label frequencies as percentage, add this (see here):

library(scales)
pp + scale_y_continuous(labels = percent)

Endoblast

What is the meaning of the dot-dot notation in ..count..? Do you have any reference to that? – Individualize

@Individualize Take a look at this question. – Endoblast

How would you obtain data labels for the frequencies, to be shown above each bar? – Marmalade

I really wish geom_bar generated the ..density.. variable like its cousin function geom_histogram does. – Dichlamydeous

I think think labels=percent in scale_y_continuous(labels = percent) doesnt work anymore – Lilylilyan

@Lilylilyan Running on ggplot2_3.1.0, scales_1.0.0 is still fine (R 3.5.1, Win 10). – Endoblast

Answer 2

now ..prop.. is available

ggplot(data=tips, aes(x=day)) + 
  geom_bar(aes(y = ..prop.., group = 1))

Atavism

This works if you are using facets, while the ..count.. option does not! – Shudder

@Shudder do you mean adding e.g. facet_wrap(~sex)? It works with both for me. – Atavism

I should have defined "works". It will create a figure if you use ..count../sum(..count..), but the frequency will sum to 1 over all of the facets (I think). If you use ..prop.., the frequency will sum to 1 in each facet. Maybe it depends on what you're trying to show. – Shudder

@Shudder ..count../sum(..count..) and ..prop work exactly the same for me as long as you set group = 1 – Atavism

ggplot(data=tips, aes(x=day)) + geom_bar(aes(y = ..prop.., group = 1)) + facet_wrap(~sex) and ggplot(data=tips, aes(x=day)) + geom_bar(aes(y = (..count..)/sum(..count..), group = 1)) + facet_wrap(~sex) do not yield the same plot for me. You can also remove group=1 in the ..count.. version and it looks the same. – Shudder

Answer 3

With ggplot2 >= 3.4.0, I think the recommended way is the following:

library("ggplot2")
library("reshape2")

ggplot(data = tips, aes(x = day)) + 
  geom_bar(aes(y = after_stat(prop), group = 1))

Sot