How to make stat_binhex shown on a log scale in ggplot2

Question

I have a 2d hexagon density plot with many points. I would like the counts within the hexagons to be displayed on a logarithmic scale, but I can't figure out how to do this through ggplot2.

Here is a simple example:

x <- runif(1000, 50, 100) 
y <- rnorm(1000, mean = 10, sd = 8)

df <- as.data.frame(cbind(x, y))

ggplot(df, aes(x, y)) + stat_binhex()

score 12 · Answer 1 · answered Apr 24 '18 at 19:15

12

Another way is to pass trans = argument to scale_fill_* used.

ggplot(df, aes(x, y)) + geom_hex() + scale_fill_gradient(trans = "log")
ggplot(df, aes(x, y)) + geom_hex() + scale_fill_viridis_c(trans = "log")

answered Apr 24 '18 at 19:15

Taekyun Kim

146
1
3

1

The automatic tickmark selection and numbering is a lot less nice in this solution, but otherwise I like it. – Nick Jun 10 '20 at 12:29
4

trans="log10" or trans="log2" gives nicer tickmarks – andyyy Mar 29 '21 at 13:09

Artem Fedosov · Accepted Answer · 2014-10-15T08:26:52.127

10

There is a fill aesthetics that defaults to ..count.. when you do not specify it in stat_binhex. The code below produces same plot as your original code.

ggplot(df, aes(x, y)) + stat_binhex(aes(fill=..count..))

enter image description here

If you want to have a log scale for counts, then the solution is pretty straight forward:

ggplot(df, aes(x, y)) + stat_binhex(aes(fill=log(..count..)))

enter image description here

edited Oct 15 '14 at 08:26

answered Oct 15 '14 at 08:03

Artem Fedosov

2,163
2
18
28

2

This doesn't work for me in ggplot2_2.1.0: Error in eval(expr, envir, enclos) : object 'count' not found – daknowles Mar 30 '16 at 05:32
1

@daknowles this is most likely caused by a bug in ggplot2: [#1608](https://github.com/hadley/ggplot2/issues/1608) – FloE Jul 19 '16 at 14:39
As of ggplot 3.4.0 one should use `after_stat(count)` instead of `..count..` as the latter is deprecated – nebroth Jun 16 '23 at 14:04

How to make stat_binhex shown on a log scale in ggplot2

2 Answers2