2

Sorry for the newbie R question...

I have a data.frame that contains measurements of a single variable. These measurements will be distributed differently depending on whether the thing being measured is of type A or type B; that is, you can imagine that my column names are: measurement, type label (A or B). I want to plot the histograms of the measurements for A and B separately, and put the two histograms in the same plot, with each histogram normalised to unit area (this is because I expect the proportions of A and B to differ significantly). By unit area, I mean that A and B each have unit area, not that A+B have unit area. Basically, I want something like geom_density, but I don't want a smoothed distributions for each; I want the histogram bars. Not interleaved, but plotted one on top of the other. Not stacked, although it would be interesting to know how to do this also. (The purpose of this plot is to explore differences in the shapes of the distributions that would indicate that there are quantitative differences between A and B that could be used to distinguish between them.) That's all. Two or more histograms -- not smoothed density plots -- in the same plot with each normalised to unit area. Thanks!

Improve this question
1
  • You are going to want to be more explicit with your question - code examples, or even a drawn picture. Regardless, I think this link should answer your question. Commented Nov 4, 2014 at 17:28

1 Answer 1

Reset to default
7

Something like this?

# generate example set.seed(1) df <- data.frame(Type=c(rep("A",1000),rep("B",4000)), Value=c(rnorm(1000,mean=25,sd=10),rchisq(4000,15))) # you start here... library(ggplot2) ggplot(df, aes(x=Value))+ geom_histogram(aes(y=..density..,fill=Type),color="grey80")+ facet_grid(Type~.)

Note that there are 4 times as many samples of type B.

You can also set the y-axis scales to float using: scales="free_y" in the call to facet_grid(...).

Improve this answer
Sign up to request clarification or add additional context in comments.

1 Comment

Thanks, this is almost exactly what I need; if I plot both histograms on the same y-axis (that is, no multiple facets) with alpha=0.something, I get exactly what I want. Very helpful! (Interestingly, I've noticed that if any Type is NA, they also get a histogram, but no legend entry. This might happen if you have no info on whether the things being measured are of type A or of type B.)

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.