R devel - Nov 1999 - tapply not simplifying to vector? (PR#320)

Hi, all.

The help file for tapply says that if simplify is true, and the result of
the calculation is always a scalar, then tapply will return a vector.

Nonetheless:
> t1 <- tapply(runif(10), rep(1:5, 2), mean)  
> is.vector(t1)  
[1] FALSE 
> is.array(t1) 
[1] TRUE
> 
I have found this in version 0.65.1 on an SGI running Irix 6.5, and on a
Mac running Linux-PPC.  I've also seen it on an Intel machine running
RedHat 6.0.

I noticed this because some results that I expected to be vectors couldn't
be combined into a data frame (Error: can't coerce array into a
data.frame).

It seems that the check and conversion simply never happens.  One way to
fix this is to simply add these few lines to the end of the function (just
before the value "ansmat" is returned: 

  if(length(dim(ansmat)) == 1){
    dim(ansmat) <- NULL
    names(ansmat) <- namelist[[1]]
  }

I wouldn't be surprised to find that there's a more elegant way to do
this.  Perhaps ansmat[index] should be returned instead of ansmat?  But I
haven't tested this extensively, so I don't know whether it might break
something else.

Hope this report helps.

Thanks for the terrific software.

Matt Wiener


-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

>>>>> "MatW" == Matthew Wiener
<mcw@ln.nimh.nih.gov> writes:
    MatW> The help file for tapply says that if simplify is true, and the
    MatW> result of the calculation is always a scalar, then tapply will
    MatW> return a vector.

    MatW> Nonetheless:

    >> t1 <- tapply(runif(10), rep(1:5, 2), mean)  
    >> is.vector(t1)  
    R> [1] FALSE 
    >> is.array(t1) 
    R> [1] TRUE

    <....> 

all correct.

The question really is if the bug is in the code or the documentation.

Here you get a 1D array, something quite rare (but appearing more and more
in recent discussions...).

One could argue that the current behavior of tapply() to always return an
array if simplify = TRUE  is more consistent, than what the doc says should
happen (coercing 1D arrays to vectors).
Otoh, the protoype  *does* return a vector in this case,
and 1D arrays can be a pain... in other places.

I'm pro fixing this.
and I'll do if I don't hear protests..


    MatW>   <....>

    MatW> Hope this report helps.

yes indeed, thank you!
R's quality is growing partly because of such fine bug reports!

    MatW> Thanks for the terrific software.

you're welcome!

Martin Maechler <maechler@stat.math.ethz.ch>
http://stat.ethz.ch/~maechler/
Seminar fuer Statistik, ETH-Zentrum  LEO D10	Leonhardstr. 27
ETH (Federal Inst. Technology)	8092 Zurich	SWITZERLAND
phone: x-41-1-632-3408		fax: ...-1228			<><

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

>>>>> "PD" == Peter Dalgaard BSA
<p.dalgaard@biostat.ku.dk> writes:
    PD> maechler@stat.math.ethz.ch writes:
    >> I'm pro fixing this.
    >> and I'll do if I don't hear protests..

    PD> Protest!...

    >> tapply(Ozone,list(Month=Month),mean,na.rm=T)
    PD> Month
    PD> May      June      July    August September 
    PD> 23.61538  29.44444  59.11538  59.96154  31.44828 

    PD> Notice the "Month" in the output.

    >> dput(tapply(Ozone,list(Month=Month),mean,na.rm=T))
    PD> structure(c(23.6153846153846, 29.4444444444444, 59.1153846153846, 
    PD> 59.9615384615385, 31.4482758620690), .Dim = 5, .Dimnames =
structure(list(
    PD> Month = c("May", "June", "July",
"August", "September")), .Names = "Month"))

    PD> Notice that the Month is the tag name of the single component of the
    PD> dimnames vector, it will disappear if you try coercing it to a
vector.

Completely agree with your argument.
 --> I'll change the documentation

Kurt: Another (R<->S) difference for the FAQ.



-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._