Superscripts within ggplot2's axis text

Question

I would like to create a graph that has superscripts on the axis instead of displaying unformatted numbers using ggplot2. I know that there are a lot of answers which change the axis label, but not the axis text. I am not trying to change the label of the graph, but the text on the axis.

Example:

x<-c('2^-5','2^-3','2^-1','2^1','2^2','2^3','2^5','2^7','2^9','2^11','2^13')
y<-c('2^-5','2^-3','2^-1','2^1','2^2','2^3','2^5','2^7','2^9','2^11','2^13')
df<-data.frame(x,y)
p<-ggplot()+
  geom_point(data=df,aes(x=x,y=y),size=4)
p

So I would like the x-axis to display the same numbers but without the carrot.

Or tweak this: https://stackoverflow.com/questions/36920492/squared-superscript-in-ggplot2-yaxis-label-in-r?rq=1 — NelsonGon, Jun 09 '19 at 16:58
This is not a duplicate because those are axis labels, this is axis text. — Jack Armstrong, Jun 09 '19 at 16:59
Are you aware the `x` and `y` are character strings, not numbers? You are plotting factors, meaning, the integers from 1 to 11. — Rui Barradas, Jun 09 '19 at 17:03
Yes I designed them like that because if it is numerical then you are plotting numbers like any other graph, I would like to keep them as characters, but you could also have factors. — Jack Armstrong, Jun 09 '19 at 17:05
So, would you like to evaluate the number equal to the text on the axis or simply remove the caret? — NelsonGon, Jun 09 '19 at 17:12

NelsonGon · Answer 1 · 2019-06-09T20:34:58.977

EDIT: A purely base approach:

df %>% 
  mutate_all(as.character)->new_df
res<-unlist(Map(function(x) eval(parse(text=x)),new_df$x))#replace with y for y
to_use<-unlist(lapply(res,as.expression))
split_text<-strsplit(gsub("\\^"," ",names(to_use))," ")
join_1<-as.numeric(sapply(split_text,"[[",1)) #tidyr::separate might help, less robust for numeric(I think)
join_2<-as.numeric(sapply(split_text,"[[",2))
to_use_1<-sapply(seq_along(join_1),function(x) parse(text=paste(join_1[x],"^",
                                                join_2[x])))

The above can be reduced to less step, I posted the stepwise approach I took. The result for only x, the same can be done for y:

new_df %>%   
ggplot()+
  geom_point(aes(x=x,y=y),size=4)+
 scale_x_discrete(breaks=df$x,labels=to_use_1)#replace with y and scale_y_discrete for y

Plot:

Original and erroneous answer:

I have deviated from standard tidyverse practice by using $, you can replace it with . and it might work although in this case it's not really important since the focus is on labels.:

    library(dplyr)
   df %>% 
          mutate(new_x=gsub("\\^"," ",x),
                 new_y=gsub("\\^"," ",y))->new_df
        new_df %>%   
        ggplot()+
          geom_point(aes(x=x,y=y),size=4)+
         scale_x_discrete(breaks=x,labels=new_df$new_x)+
          scale_y_discrete(breaks=y,labels=new_df$new_y)

I think this is a good start, but now the numbers are out of order — Jack Armstrong, Jun 12 '19 at 16:15
Not sure why we have different results though. Are you using the data in the question or some other data? Been a few days since I looked at this so might need to see the code you're using and/or data. — NelsonGon, Jun 12 '19 at 19:19

Rui Barradas · Accepted Answer · 2019-06-12T21:31:32.490

This can be done with functions scale_x_log2 and scale_y_log2 that can be found in GitHub package jrnoldmisc.

First, install the package.

devtools::install_github("jrnold/rubbish")

Then, coerce the variables to numeric. I wil work with a copy of the original dataframe.

df1 <- df
df1[] <- lapply(df1, function(x){
  x <- as.character(x)
  sapply(x, function(.x)eval(parse(text = .x)))
})

Now, graph it.

library(jrnoldmisc)
library(ggplot2)
library(MASS)
library(scales)

a <- ggplot(df1, aes(x = x, y = y, size = 4)) + 
  geom_point(show.legend = FALSE) +
  scale_x_log2(limits = c(0.01, NA), 
                labels = trans_format("log2", math_format(2^.x)),
                breaks = trans_breaks("log2", function(x) 2^x, n = 10)) +
  scale_y_log2(limits = c(0.01, NA),
                labels = trans_format("log2", math_format(2^.x)),
                breaks = trans_breaks("log2", function(x) 2^x, n = 10))
a + annotation_logticks(base = 2)

Edit.

Following the discussion in the comments, here are the two other ways that were seen to give different axis labels.

Axis labels every tick mark. Set limits = c(1.01, NA) and function argument n = 11, an odd number.
Axis labels on odd number exponents. Keep limits = c(0.01, NA), change to function(x) 2^(x - 1), n = 11.

Just the instructions, no plots.

The first.

a <- ggplot(df1, aes(x = x, y = y, size = 4)) + 
  geom_point(show.legend = FALSE) +
  scale_x_log2(limits = c(1.01, NA), 
                labels = trans_format("log2", math_format(2^.x)),
                breaks = trans_breaks("log2", function(x) 2^(x), n = 11)) +
  scale_y_log2(limits = c(1.01, NA),
                labels = trans_format("log2", math_format(2^.x)),
                breaks = trans_breaks("log2", function(x) 2^(x), n = 11))
a + annotation_logticks(base = 2)

And the second.

a <- ggplot(df1, aes(x = x, y = y, size = 4)) + 
  geom_point(show.legend = FALSE) +
  scale_x_log2(limits = c(0.01, NA), 
               labels = trans_format("log2", math_format(2^.x)),
               breaks = trans_breaks("log2", function(x) 2^(x - 1), n = 11)) +
  scale_y_log2(limits = c(0.01, NA),
               labels = trans_format("log2", math_format(2^.x)),
               breaks = trans_breaks("log2", function(x) 2^(x - 1), n = 11))
a + annotation_logticks(base = 2)

Is it possible to have the tick marks on the odd numbers or every tick mark instead of only the even ones? — Jack Armstrong, Jun 12 '19 at 16:34
@JackArmstrong Yes, it is, I have just tested `limits = c(1.01, NA)` on both `sclae_*_log2` calls and it now prints on every tick mark. — Rui Barradas, Jun 12 '19 at 17:16
Instead of using limits, within breaks, = trans_breaks("log2", function(x) 2^(x-1), n = 10)) — Jack Armstrong, Jun 12 '19 at 19:08
@JackArmstrong You are absolutely right, didn't think of that. But it will give the labels on odd exponents only, with me `limits` gives them on every tick mark. — Rui Barradas, Jun 12 '19 at 19:16
@JackArmstrong OK, got it. I had changed argument `n` in `function(x) 2^(x), n = 11` to an odd number. Plus `limits = c(1.01, NA)` it gives labels on every tick mark. Should I edit the answer with this? — Rui Barradas, Jun 12 '19 at 19:20
If you want to with other possible ways, sure. No need for pictures I think. People should be able to read. — Jack Armstrong, Jun 12 '19 at 20:16

eipi10 · Answer 3 · 2019-06-25T16:03:56.890

You can provide a function to the labels argument of the scale_x_*** and scale_y_*** functions to generate labels with superscripts (or other formatting). See examples below.

library(jrnoldmisc)
library(ggplot2)

df<-data.frame(x=2^seq(-5,5,2),
               y=2^seq(-5,5,2))

ggplot(df) +
  geom_point(aes(x=x,y=y),size=2)  +
  scale_x_log2(breaks=2^seq(-5,5,2),
               labels=function(x) parse(text=paste("2^",round(log2(x),2))))

ggplot(df) +
  geom_point(aes(x=x,y=y),size=2)  +
  scale_x_continuous(breaks=c(2^-5, 2^seq(1,5,2)),
                     labels=function(x) parse(text=paste("2^",round(log2(x),2))))

ggplot(df) +
  geom_point(aes(x=x,y=y),size=2)  +
  scale_x_log10(breaks=10^seq(-1,1,1),
                labels=function(x) parse(text=paste("10^",round(log10(x),2))))

Superscripts within ggplot2's axis text

3 Answers3