R-alpha: Re: R-beta: Bug or feature? [+ better histograms]

Peter Dalgaard BSA p.dalgaard@kubism.ku.dk
18 Aug 1997 11:51:57 +0200

Kurt Hornik <Kurt.Hornik@ci.tuwien.ac.at> writes:

> >>>>> Ross Ihaka writes:
> > Bill Venables writes:
> >> First a bit of an explanation.  Brian Ripley and I are firmly of
> >> the opinion that, to be any use in teaching at all, histograms
> >> should be nonparametric estimators of the probability density
> >> function.  That is, the vertical scale should be a relative
> >> frequency density scale.  It seems impossible to get this in R, so
> >> here is a function (+ a few extras) that does give you a "true"
> >> histogram, (unless you are peverse enough to request otherwise.)
> > Oh goody!  I HATE the present hist; partly because of the
> > frequency/density aspect, but mainly because if I ask for 10 cells I
> > expect to get 10 cells, not 7 or 35 or whatever.  I'd like propose
> > that the present hist get renamed hist.old (or something similar) and
> > that Bill's "truehist" become the true "hist".
> I'd say, get rid of the old one ...

Ugh. Please do NOT do that! Some of us need to teach histograms to
people who have difficulties with integrals (and also in recognizing
the connection between a barplot and a step function), so we have to
make do with "shape of curve" considerations, at least in the
beginning. The last thing we need is inexplicable y-axes.

The "# in bin" is not really problematic as long as the the bins are
of the same size and *much* easier to explain to students. 

Even for a statistician, the number is really what conveys the
accuracy estimate that you need when you try to evaluate whether a
histogram matches a given density.

Make it an option, e.g. hist(x,as.density=T). People who can
understand why the "sum of area of blocks should be 1" can also figure
out how to add an option to a function call.

I could easily do without the "prettification" silliness, though.

   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard@biostat.ku.dk)             FAX: (+45) 35327907
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch