R graphics: add labels to clustered/dodged bar chart

Question

This is a basic plotting question:

I need to add labels to a clustered/dodged bar chart. I have looked at several examples using text(), but cannot seem to position the labels correctly.

teachers <- c("A", "B","C", "D", "E")
mean_pre_scores <- c(10, 11, 12, 10,9)
mean_post_scores <- c(12,15,17,13,12)
pre_post <- data.frame(mean_pre_scores, mean_post_scores)

pre_post <- as.matrix(pre_post)
barplot((t(pre_post)), beside = T, names = teachers, legend = c("pre", "post"), 
        ylim = c(0,20), args.legend = list(x="bottomright"),  axes = T, main = "Unit 1 Test",
        col=c(26,51))

I want to modify this plot so that the values will be displayed above the bars. It would also be helpful to know how to show the values inside the bars.

This [post] (http://stackoverflow.com/questions/3626324/r-graphics-add-labels-to-stacked-bar-chart) discusses the same issue, you might have to tinker with Y-position of the labels for your dataset — Silence Dogood, Apr 25 '14 at 13:37
@Vivek Yes, I saw this post, but am having trouble positioning the labels. — windy, Apr 25 '14 at 13:40

Tyler Rinker · Accepted Answer · 2014-04-25T14:26:58.663

3

I think this is what you're after:

z <- barplot((t(pre_post)), beside = T, names = teachers, 
    legend = c("pre", "post"), ylim = c(0,20), 
    args.legend = list(x="topright"),  axes = T, 
    main = "Unit 1 Content Pre Test", col=c(26,51))

text(cex=1, x=c(z[1, ], z[2, ]), y=c(pre_post) + par("cxy")[2]/2, c(pre_post), xpd=TRUE)

enter image description here

To move the text inside the bars simply use subtractions as in:

text(cex=1, x=c(z[1, ], z[2, ]), y=c(pre_post) - par("cxy")[2]/2, c(pre_post), xpd=TRUE)

edited Apr 25 '14 at 14:26

answered Apr 25 '14 at 13:40

Tyler Rinker

108,132
65
322
519

Can I suggest moving the legend up top? – Rich Scriven Apr 25 '14 at 13:44
@Richard agreed, I did so. – Tyler Rinker Apr 25 '14 at 14:27

Rich Scriven · Answer 2 · 2014-04-25T14:07:32.013

When you say "above the bars", do you mean directly above them? Or on the top axis?

Here's an example with them on top.

And now it's clear you wanted the numeric values on the bars instead of the letters on the top axis. I'll leave it here anyway, and post a second result below.

> barplot((t(pre_post)), beside = T, legend = c("pre", "post"), ylim = c(0,20),
          args.legend = list(x="bottomright"),  axes = T, col=c(26,51))
> axis(3, at = c(2, 5, 8, 11, 14), labels = teachers)
> title(main = "Unit 1 Content Pre Test", line = 3)
> box()

enter image description here

Here's the correct result. I moved the legend to the top right corner where it won't be covering any bars. A bit easier to read that way.

> barplot((t(pre_post)), beside = TRUE, legend = c("pre", "post"), 
          main = "Unit 1 Content Pre Test", ylim = c(0,20), 
          args.legend = list(x="topright"), col=c(26,51))
> text(x = c(1.5,2.5,4.5,5.5,7.5,8.5,10.5,11.5,13.5,14.5), 
       y = txt.height+1, labels = as.character(txt.height))
> box()

enter image description here

Hmm we interpreted this differently but I definitively get you're thinking. I guess the OP will have to clarify. — Tyler Rinker, Apr 25 '14 at 13:42
I think you got it. She did say "values". I ran right past the meaning of that. — Rich Scriven, Apr 25 '14 at 13:43

Jaap · Answer 3 · 2017-10-21T09:59:35.930

A ggplot2 solution:

# get the data into the right format
pre_post <- data.frame(teachers,mean_pre_scores, mean_post_scores)
library(reshape2)
prepost <- melt(pre_post, id = "teachers")

# create the plot
ggplot(prepost, aes(x = teachers, y = value, fill = variable)) +
  geom_bar(stat = "identity", position = "dodge") +
  geom_text(aes(label = value), vjust = -0.5, position = position_dodge(0.9)) +
  theme_bw()

the result:

enter image description here

score 0 · Answer 4 · edited May 23 '17 at 10:32

Adapted from this post (R graphics: Add labels to stacked bar chart)

For inside labels:

teachers <- c("A", "B","C", "D", "E")
mean_pre_scores <- c(10, 11, 12, 10,9)
mean_post_scores <- c(12,15,17,13,12)
pre_post.df <- data.frame(mean_pre_scores, mean_post_scores)

pre_post <- as.matrix(pre_post.df)
b<-barplot((t(pre_post)), beside = T, names = teachers, legend = c("pre", "post"), 
     ylim = c(0,20), args.legend = list(x="bottomright"),  axes = T, main = "Unit 1 Content Pre Test",
     col=c(26,51))


ypos.inside<-apply(pre_post, 2, function(x) x -1 )
ypos.inside <- t(ypos.inside)

text(b, ypos.inside, pre_post)

enter image description here

For outside labels:

teachers <- c("A", "B","C", "D", "E")
mean_pre_scores <- c(10, 11, 12, 10,9)
mean_post_scores <- c(12,15,17,13,12)
pre_post.df <- data.frame(mean_pre_scores, mean_post_scores)

pre_post <- as.matrix(pre_post.df)
b<-barplot((t(pre_post)), beside = T, names = teachers, legend = c("pre", "post"), 
     ylim = c(0,20), args.legend = list(x="bottomright"),  axes = T, main = "Unit 1 Content Pre Test",
     col=c(26,51))

ypos.outside<-apply(pre_post, 2, function(x) x +1 )
ypos.outside <- t(ypos.outside)
text(b, ypos.outside, pre_post)

enter image description here

This approach is less generalizable than the one I provide in that if the scale becomes extreme (change one number to something like 100) the relative label placement changes. — Tyler Rinker, Apr 25 '14 at 14:33

R graphics: add labels to clustered/dodged bar chart

4 Answers4