Relative frequency histogram in R, ggplot

Question

I can draw relative frequency histogram in R, using lattice package:

a <- runif(100)
library(lattice)
histogram(a)

I want to get the same graph in ggplot. I tried

dt <- data.frame(a)
ggplot(dt, aes(x = a)) + 
geom_bar(aes(y = ..prop..))+
 scale_y_continuous(labels=percent)

but it doesn't work like that. What I should change in the code? Calculating relative frequency before graph is not an option for me.

I guess you are looking for `geom_histogram` – kath Aug 09 '19 at 13:21 — kath, Aug 09 '19 at 13:21
ggplot(dt, aes(x=a)) + geom_histogram() – Dr. Flow Aug 09 '19 at 13:23 — Dr. Flow, Aug 09 '19 at 13:23
No, i need relative frequency, not a count. – neringab Aug 09 '19 at 13:34 — neringab, Aug 09 '19 at 13:34

Axeman · Accepted Answer · 2022-05-11T15:14:56.800

24

You want a histogram, not a barplot, so:

ggplot(dt, aes(x = a)) + 
  geom_histogram(aes(y = after_stat(count / sum(count))), bins = 8) +
  scale_y_continuous(labels = scales::percent)

lattice:

ggplot2:

You can see that the binning algorithm works slightly different for the two packages.

edited May 11 '22 at 15:14

answered Aug 09 '19 at 14:00

Axeman

3

Another option is to use `geom_histogram(aes(y = stat(density)))` directly. https://ggplot2.tidyverse.org/reference/geom_histogram.html#computed-variables – Kiteration Jan 15 '22 at 06:57
@Kiteration, That gives a different plot. – Axeman May 11 '22 at 15:16

score 0 · Answer 2 · answered Aug 09 '19 at 13:41

0

You can try something like :

ggplot(data=df, aes(x=a)) + geom_bar(aes(y = (..count..)/sum(..count..)), group = 1)

answered Aug 09 '19 at 13:41

S.Gradit

No, with this method i only see three vertical lines, not a distribution. – neringab Aug 09 '19 at 13:54

2 Answers2