[R] ave gives unexpected NA's

[Peter Dalgaard]

Wolfram Fischer <wolfram at fischer-zim.ch> writes:

> [R 2.0.0 on Linux]
> 
> I tried:
> > df <- data.frame(
>      grp1=factor( c('A' ,'A' ,'A' ,'D', 'D' ) ) ,
>      grp2=factor( c('a1','a2','a2','d1','d1') )
>      )
> > df
>   grp1 grp2 val
> 1    A   a1   1
> 2    A   a2   2
> 3    A   a2   4
> 4    D   d1   8
> 5    D   d1  16
> 
> I got:
> > with( df, ave( val, grp1, grp2, FUN=sum ) )
> [1]  1 24 24 NA NA
> 
> I have expected to get:
> [1]  1 6 24
> 
> Do I misunderstand something with `ave' or is there a bug?

Both. You should have expected 1 6 6 24 24.

Apparently the culprit is this line inside ave()

  unlist(lapply(split(x, g), FUN))[g]

which acts up if there are empty levels of g. It works with

  unlist(lapply(split(x, g), FUN))[as.character(g)]

but that doesn't strike me as the right solution.

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907