Hi Ulrike
any set of three people will probably have five different opinions on
this, but I can see that this makes sense:
NA - not available, not measured, not recorded
NaN - result of an arithmetic computation that lies outside of the real
numbers; in that sense, "available".
However, this point of view then opens up the question why 'is.na(NaN)'
is 'TRUE'.
Best wishes
Wolfgang
On 07/09/10 21:18, Ulrike Grömping wrote:
Kevin,
I wouldn't mind NaN (although it seems a bit strange, because you
wouldn't expect a character to be a number), but I find it strange to
get the character string "NaN". is.na(as.character(NaN)) returns FALSE,
which is what I dislike.
Best, Ulrike
Kevin R. Coombes schrieb:
It seems to me that preserving information about the kind of number
(or not) present would be useful. I rather like the fact that
as.numeric(as.character(NaN))
and
as.numeric(as.character(Inf))
both work as the identity operator on numeric-like objects. (In this
context, note that both is.numeric(NaN) and is.numeric(Inf) both
return TRUE.) In your example, the character string "ee" does not
represent any number that I know about (at least in standard R).
Kevin
On 9/7/2010 11:23 AM, Ulrike Grömping wrote:
Dear DevelopeRs,
I am surprised about the outcome of the second command:
str(as.character(as.numeric("ee"))) str(as.character(log(-1)))
I would have expected a character NA. Is there an intention behind
this behavior?
Best, Ulrike
______________________________________________
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
--
Wolfgang Huber
EMBL
http://www.embl.de/research/units/genome_biology/huber
______________________________________________
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel