How to make variable width histogram in R with labels aligned to bin edges?

问题

I'm using ggplot2, which by default creates histograms with fixed bin widths and whose bin labels are plotted in the center of each bin.

What I want instead is a variable-width histogram whose bin labels are representative of the end points of each bin, like this plot:

desired plot

To produce this example plot, I manually entered the bin parameters and shifted the bins to align them with their end points:

income=data.frame(lx=c(0,10,25,50,100),rx=c(10,25,50,100,150),y=c(20,28,27,18,7))
income$width = income$rx-income$lx


ggplot(income, aes(lx+width/2,y/width)) + geom_bar(aes(width=rx-lx), color='black', stat='identity') + 
  scale_x_continuous(breaks=unique(c(income$lx,income$rx))) + labs(x='Income (thousands of $)', y='% per thousand $')

But I want to do this automatically, from the original data. (The original data can be approximated using the following code):

incomes=unlist(sapply(1:nrow(income), function(i) sample(income$lx[i]:(income$rx[i]-1),income$y[i],replace=TRUE)))
widths=unlist(sapply(1:nrow(income), function(i) rep(income$rx[i]-income$lx[i],income$y[i])))
incomes=data.frame(incomes, widths)

回答1:

You can produce a variable width histogram by specifying the desired breaks in geom_histogram. Use y=..density.. (rather than the default, which is based on counts), so that the bars will be normalized to their proportion of the total bar area.

breaks = c(0,10,25,50,100,150)

ggplot(incomes, aes(incomes)) +
  geom_histogram(aes(y=..density..),
                 color="black", fill="grey40", breaks=breaks) +
  scale_x_continuous(breaks=breaks)

来源：https://stackoverflow.com/questions/37766893/how-to-make-variable-width-histogram-in-r-with-labels-aligned-to-bin-edges

标签

ggplot2