hist(.) -- S compatibility : [a,b) versus (a,b] and ....

Martin Maechler Martin Maechler <maechler@stat.math.ethz.ch>
Wed, 7 Jan 1998 10:24:06 +0100

Looking at Albrech Gebhardt's bug report to R-help (Dec.23)
and proposal for improvement,

I've been also trying several examples and comparing results with S-plus.
Note that we have decided to NOT be fully compatible to S with our hist(.)
function since we found S to be really `wrong' and `unuseful':
S: - The value returned by hist(.) depends on plot=T or plot=F
R: - We return a list(.) either way which contains additional useful things
     anyway, see,  hist(.) in 0.60 or 0.61.
   - we will have a ``label = FALSE'' argument as proposed by Albrecht

However, there are several  incompatibilities:

1) S uses half-open  (a,b] intervals 
   where for the left most cell,  [a,b] is used when
	``include.lowest  =  T'' (which is default).

   R currently uses half-open  [a,b) intervals.

I propose to change R here, including the "include.lowest = TRUE" argument

2) S has    ``probability = F'',
   we have  ``freq = NULL''  (which defaults to FALSE for equidistant breaks
			      and  TRUE for non-equidistant breaks).

I think we don't want to change the better default here.
I wonder if we should change the argument name from 'freq' to 'probability'

If I don't hear objections or better proposals from you,
I'll change "1)" and leave "2)".

Martin Maechler <maechler@stat.math.ethz.ch>		 <><
Seminar fuer Statistik, SOL G1
ETH (Federal Inst. Technology)	8092 Zurich	 SWITZERLAND
phone: x-41-1-632-3408		fax: ...-1086
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