hist.default()$density
Dear developers,
since running 'example(hist)' produces
...
hist> sum(r$density * diff(r$breaks)) # == 1
[1] 0.9999999
...
I suppose that the current behaviour of hist() is not as intended (and
documented).
So, please find attached (and inline below) a (trivial) patch for
hist.default().
Best wishes
Martin
Index: src/library/graphics/R/hist.R
===================================================================
--- src/library/graphics/R/hist.R (revision 51652)
+++ src/library/graphics/R/hist.R (working copy)
@@ -111,7 +111,7 @@
stop("negative 'counts'. Internal Error in C-code for \"bincount\"")
if (sum(counts) < n)
stop("some 'x' not counted; maybe 'breaks' do not span range of 'x'")
- dens <- counts/(n*h)
+ dens <- counts/(n*diff(breaks))
mids <- 0.5 * (breaks[-1L] + breaks[-nB])
r <- structure(list(breaks = breaks, counts = counts,
intensities = dens,
Martin Becker wrote:
Dear developers, the current implementation of hist.default() calculates 'density' (and 'intensities') as dens <- counts/(n*h) where h has been calculated before as h <- diff(fuzzybreaks) which results in 'fuzzy' values for the density, see e.g.
tmp <- hist(1:10,breaks=c(-2.5,2.5,7.5,12.5),plot=FALSE) print(tmp$density,digits=15)
[1] 0.0399999920000016 0.1000000000000000 0.0600000000000000 Since hist.default()$breaks are not the fuzzy breaks used for the calculation of dens, the sum of the bins' area is significantly different from 1 in many cases, see e.g.
print(sum(tmp$density*diff(tmp$breaks)),digits=15)
[1] 0.999999960000008 Is this intended, or should the calculation of dens read dens <- counts/(n*diff(breaks)) instead (or should hist.default()$breaks return the fuzzy breaks)? Best wishes Martin
Dr. Martin Becker Statistics and Econometrics Saarland University Campus C3 1, Room 217 66123 Saarbruecken Germany -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: hist.diff URL: <https://stat.ethz.ch/pipermail/r-devel/attachments/20100409/d0c95a61/attachment.pl>