Question

I would like to label each of the boxes in a barplot by their size(i.e number of observations in dataframe which are in the group).

e.g If the first variable has 3 levels and the second variable has 4 levels, I would like 12 labels.

(Also, is it possible to control the size or position of these labels)

Thank you for any help.

Answer 1

Here’s one way to do it, using the data VADeaths as an example (it will be in your R workspace by default, or if not, use library(datasets)).

bar <- barplot(VADeaths)
text(rep(bar,each=nrow(VADeaths)), as.vector(apply(VADeaths,2,cumsum)), 
     labels=as.vector(apply(VADeaths,2,cumsum)),pos=3)

It looks like this:

To modify the size of the font you can use text(...,cex=2) to make things twice the size they were, e.g.

Now, let’s explain this code so you know how to do it yourself!

First, let’s look at VADeaths: it’s a tally of deaths in each age group by category:

> VADeaths
      Rural Male Rural Female Urban Male Urban Female
50-54       11.7          8.7       15.4          8.4
55-59       18.1         11.7       24.3         13.6
60-64       26.9         20.3       37.0         19.3
65-69       41.0         30.9       54.6         35.1
70-74       66.0         54.3       71.1         50.0

Now, to do the text on the barplot, we basically draw the barplot, and then draw the text on top using R command text (see ?text).

text requires x,y coordinates and corresponding pieces of text to draw on the bar plot. We will give it the coordinates of each line in the bar plot to draw the text on.

To do this, see the “Value” section ?barplot. This function not only plots your bar plot, but returns the x coordinate of each bar. score!

> bar <- barplot(VADeaths)
> bar
[1] 0.7 1.9 3.1 4.3

Now all we need is y coordinates to go with our x coordinates.

Well, a stacked bar plot just tallies up the frequencies in VADeaths as you go along.
For example, in the ‘Rural Male’ group, the first line is drawn at 11.7, and the second is drawn at 11.7 + 18.1 = 29.8, the third at 11.7 + 18.1 + 26.9 = 56.7, and so on (see the values in VADeaths).

So, our y coordinates need to be cumulative sums going down the columns.

To calculate these for each column, we can use cumsum. For example

> cumsum(c(1,2,3,4,5))
[1]  1  3  6 10 15

Since we want to do this for each column in VADeaths, we have to use the function apply.

> apply(VADeaths,2,cumsum)
      Rural Male Rural Female Urban Male Urban Female
50-54       11.7          8.7       15.4          8.4
55-59       29.8         20.4       39.7         22.0
60-64       56.7         40.7       76.7         41.3
65-69       97.7         71.6      131.3         76.4
70-74      163.7        125.9      202.4        126.4

apply(VADeaths,2,cumsum) means: “For each column in VADeaths, calculate the cumsum of that”.
This gives us the y values for each line of the bar plot.

Let’s save these yvalues for further use:

> yvals <- as.vector(apply(VADeaths,2,cumsum))

The reason I use as.vector is just to flatten the matrix into a vector of values — it makes the plotting easier.

One last thing — my x values (that I stored in bar) only have one value per bar, but I need to expand it out so there’s one x value per line on each bar. To do this:

> xvals <- rep(bar,each=nrow(VADeaths))

This turns my previous x1,x2,x3,x4 into x1,x1,x1,x1,x1, x2,x2,x2,x2,x2, ..., x4,x4,x4,x4,x4.
Now my xvals match my yvals.

After this it’s simply a case of using text.

> text( xvals, yvals, labels=yvals, pos=3 )

The labels arguments tells text what text to put at the x/y positions.
The pos=3 means “draw each bit of text just above my specified x/y value”. Otherwise, the numbers would be drawn over the lines of the barplot which would be hard to read.

Now, there are many options for customising the position and size of text, and I suggest you read ?text to see them.

All this code condenses down to the two-liner I gave at the beginning of the answer, but this version might be a little more understandable:

bar <- barplot(VADeaths)
xvals <- rep(bar,each=nrow(VADeaths))
yvals <- as.vector(apply(VADeaths,2,cumsum))
text( xvals, yvals, labels=yvals, pos=3 )