R help - Sep 2005 - Using unsplit - unsplit does not seem to reverse the effect of split

In data OME in MASS I would like to extract the first 5 observations per subject
(=ID). So I do
 
library(MASS)
OMEsub <- split(OME, OME$ID)
OMEsub <- lapply(OMEsub,function(x)x[1:5,])
unsplit(OMEsub, OME$ID)

- which results in 
 
[[1]]
[1] 1 1 1 1 1
[[2]]
[1] 30 30 30 30 30
[[3]]
[1] low low low low low
Levels: N/A high low
[[4]]
[1] 35 35 40 40 45
[[5]]
[1] coherent   incoherent coherent   incoherent coherent  
Levels: coherent incoherent
[[6]]
[1] 1 4 0 1 2

............
 
[[1094]]
[1] 4 5 5 5 2
[[1095]]
[1] 100 100 100 100 100
[[1096]]
[1] 18 18 18 18 18
[[1097]]
[1] N/A N/A N/A N/A N/A
Levels: N/A high low
There were 50 or more warnings (use warnings() to see the first 50)

warnings()
Warning messages:
1: number of items to replace is not a multiple of replacement length
2: number of items to replace is not a multiple of replacement length
3: number of items to replace is not a multiple of replacement length
....
 
According to documentation unsplit is the reverse of split, but I must be
missing a point somewhere... Can anyone help? Thanks in advance. S??ren

S??ren H??jsgaard <Soren.Hojsgaard at agrsci.dk> writes:
> In data OME in MASS I would like to extract the first 5 observations per
subject (=ID). So I do
>  
> library(MASS)
> OMEsub <- split(OME, OME$ID)
> OMEsub <- lapply(OMEsub,function(x)x[1:5,])
> unsplit(OMEsub, OME$ID)
> 
> - which results in 
>  
> [[1]]
> [1] 1 1 1 1 1
> [[2]]
> [1] 30 30 30 30 30
> [[3]]
> [1] low low low low low
> Levels: N/A high low
> [[4]]
> [1] 35 35 40 40 45
> [[5]]
> [1] coherent   incoherent coherent   incoherent coherent  
> Levels: coherent incoherent
> [[6]]
> [1] 1 4 0 1 2
> 
> ............
>  
> [[1094]]
> [1] 4 5 5 5 2
> [[1095]]
> [1] 100 100 100 100 100
> [[1096]]
> [1] 18 18 18 18 18
> [[1097]]
> [1] N/A N/A N/A N/A N/A
> Levels: N/A high low
> There were 50 or more warnings (use warnings() to see the first 50)
> 
> warnings()
> Warning messages:
> 1: number of items to replace is not a multiple of replacement length
> 2: number of items to replace is not a multiple of replacement length
> 3: number of items to replace is not a multiple of replacement length
> ....
>  
> According to documentation unsplit is the reverse of split, but I must be
missing a point somewhere... Can anyone help? Thanks in advance. S??ren
It only works if the first argument is or could have resulted from a
split on the second argument. That is clearly not the case when you
are creating subvectors. 

I have on occasion wanted an unsplit that worked without the 2nd
argument as in

unsplit(l, rep(seq(along=l), sapply(l,length)) )

but if you think about it, it's not really doing anything that
do.call("c",l) or do.call("rbind",l) won't do.

-- 
   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

On Tue, 2005-09-27 at 19:12 +0200, S伱伕ren H伱伕jsgaard
wrote:
> In data OME in MASS I would like to extract the first 5 observations
> per subject (=ID). So I do
>  
> library(MASS)
> OMEsub <- split(OME, OME$ID)
> OMEsub <- lapply(OMEsub,function(x)x[1:5,])
> unsplit(OMEsub, OME$ID)
> 
> - which results in 
>  
> [[1]]
> [1] 1 1 1 1 1
> [[2]]
> [1] 30 30 30 30 30
> [[3]]
> [1] low low low low low
> Levels: N/A high low
> [[4]]
> [1] 35 35 40 40 45
> [[5]]
> [1] coherent   incoherent coherent   incoherent coherent  
> Levels: coherent incoherent
> [[6]]
> [1] 1 4 0 1 2
> 
> ............
>  
> [[1094]]
> [1] 4 5 5 5 2
> [[1095]]
> [1] 100 100 100 100 100
> [[1096]]
> [1] 18 18 18 18 18
> [[1097]]
> [1] N/A N/A N/A N/A N/A
> Levels: N/A high low
> There were 50 or more warnings (use warnings() to see the first 50)
> 
> warnings()
> Warning messages:
> 1: number of items to replace is not a multiple of replacement length
> 2: number of items to replace is not a multiple of replacement length
> 3: number of items to replace is not a multiple of replacement length
> ....
>  
> According to documentation unsplit is the reverse of split, but I must
> be missing a point somewhere... Can anyone help? Thanks in advance.
> S伱伕ren

If you read the documentation for split/unsplit, you will also note that
in the Details section it says:

 'unsplit' works only with lists of vectors

as opposed to lists of data frames, which is the result of your split()
operation.

Also note that in the Value section, it indicates:

'unsplit' returns a vector for which 'split(x, f)' equals
'value'

as opposed to unsplit returning a data frame.


Thus, use:

  OME1 <- do.call("rbind", OMEsub)

where OME1 will be the result of rbind()'ing the data frames in the
OMEsub list.

See ?do.call for more information.

HTH,

Marc Schwartz