R devel - Sep 2008 - aggregate() does not return POSIXct object correctly (PR#12887)

Full_Name: Rene Locher
Version: 2.7.2 Patched (2008-09-12 r46541)
OS: XP
Submission from: (NULL) (160.85.231.40)


dat <- data.frame(event=factor(c("A","A","B")),
                 
time=as.POSIXct(c("2008-01-10","2008-01-01","2008-01-04")))

min(dat$time)
## "2008-01-01 CET"
## as expected

aggregate(dat$time,by=list(event=dat$event),min)
## results in
##    event          x
##  1     A 1199142000
##  2     B 1199401200

## I expected:
##    event                x
##  1     A "2008-01-01 CET"
##  2     B "2008-01-04 CET"

## my system:

sessionInfo()
## R version 2.7.2 Patched (2008-09-12 r46541)
## i386-pc-mingw32
##
## locale:
##
LC_COLLATE=German_Switzerland.1252;LC_CTYPE=German_Switzerland.1252;LC_MONETARY=German_Switzerland.1252;LC_NUMERIC=C;LC_TIME=German_Switzerland.1252
##
## attached base packages:
## [1] stats     graphics  grDevices utils     datasets  methods   base

Kind regards

Rene Locher                                     E-Mail: rene.locher at zhaw.ch
Institute for Data Analysis and Process Design     Tel:    +41 58 934 7810
Zurich University of Applied Sciences              Fax:    +41 58 935 7810
Rosenstrasse 3                   
Postfach
CH-8401 Winterthur                                http://www.idp.zhaw.ch

rene.locher at zhaw.ch wrote:
> Full_Name: Rene Locher
> Version: 2.7.2 Patched (2008-09-12 r46541)
> OS: XP
> Submission from: (NULL) (160.85.231.40)
>
>
> dat <-
data.frame(event=factor(c("A","A","B")),
>                  
time=as.POSIXct(c("2008-01-10","2008-01-01","2008-01-04")))
>
> min(dat$time)
> ## "2008-01-01 CET"
> ## as expected
>
> aggregate(dat$time,by=list(event=dat$event),min)
> ## results in
> ##    event          x
> ##  1     A 1199142000
> ##  2     B 1199401200
>
> ## I expected:
> ##    event                x
> ##  1     A "2008-01-01 CET"
> ##  2     B "2008-01-04 CET"
>   
This is as documented, possibly annoying, but not a bug.

aggregate() is documented to build on tapply() for which the discarding
of class is documented on ?tapply. The root cause is unlist():
> tapply(dat[["time"]],dat$event,min,simplify=FALSE)
$A
[1] "2008-01-01 CET"

$B
[1] "2008-01-04 CET"
> unlist(tapply(dat[["time"]],dat$event,min,simplify=FALSE))
         A          B
1199142000 1199401200

and a partial rationale is that unlist() wouldn't know what to do if the
arguments had different classes.
The workaround is, of course, just to stick the class back on.




-- 
   O__  ---- Peter Dalgaard             ?ster Farimagsgade 5, Entr.B
  c/ /'_ --- Dept. of Biostatistics     PO Box 2099, 1014 Cph. K
 (*) \(*) -- University of Copenhagen   Denmark      Ph:  (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)              FAX: (+45) 35327907