0

Histogram breaks

In base plot, when you use the hist function, it automatically calculates the breaks using the nclass.Sturges function, in ggplot however, you have to provide the breaks.

If I plot a histogram of the classical faithfull data I get the following graph:

data("faithful")
hist(faithful$waiting)

enter image description here

This works

I found out in this question, that you can mimic the 

library(tidyverse)
data("faithful")
brx <- pretty(range(faithful$waiting), n = nclass.Sturges(faithful$waiting), min.n = 1)
ggplot(faithful, aes(waiting)) + 
  geom_histogram(color="darkgray", fill="white", breaks=brx)

enter image description here

But this does not

But I would like to add the breaks within my ggplot function, so I tried this:

ggplot(faithful, aes(waiting)) + 
  geom_histogram(color="darkgray", fill="white", 
                 aes(breaks=pretty(range(waiting), 
                                   n = nclass.Sturges(waiting), 
                                   min.n = 1)))

Which gives me the following error:

Warning: Ignoring unknown aesthetics: breaks
Error: Aesthetics must be either length 1 or the same as the data (272): breaks, x

I understand what it means, but I can put Aesthetics of length one in aes, such as:

ggplot(faithful, aes(waiting)) + 
  geom_histogram(color="darkgray", fill="white", breaks=brx, 
                 aes(alpha = 0.5))

What am I doing wrong?

Z.Lin
  • 28,055
  • 6
  • 54
  • 94
Derek Corcoran
  • 3,930
  • 2
  • 25
  • 54
  • Based on your data, `brx` is a vector of length 13, not length 1. In any case `geom_histogram` (which uses the same aesthetics as `geom_bar`) does not understand `breaks` as an aesthetic mapping, so even something like `aes(breaks = 1)` would be ignored. – Z.Lin May 07 '18 at 01:41

1 Answers1

1

geom_histogram uses the same aes as geom_bar according to the documentation and breaks isn't one of those aesthetics. (See geom_bar)

In the working code chunk you pass breaks to the function geom_histogram directly and that works, but in the problematic chunk you pass it as an aesthetic, and ggplot complains.

This works for me and I think does what you want:

ggplot(faithful, aes(x = waiting)) + 
geom_histogram(color = "darkgray", fill = "white", 
             breaks = pretty(range(faithful$waiting), 
                               n = nclass.Sturges(faithful$waiting), 
                               min.n = 1))
Brendan A.
  • 1,268
  • 11
  • 16