[R] Odp: splitting vector into equal sets

[Petr PIKAL]

utkarshsinghal <utkarsh.singhal at global-analytics.com> napsal dne 
16.04.2009 17:20:10:

> Hi Pikal,
> 
> Thanks for your comments and apologies for not providing a clear 
example. But 
> you have completely ignored the small example I mentioned:

Well, this is not exactly an example. So let us clarify things a bit. You 
do not explain how to handle your vector at all.

One option
split(1:20, factor(1:3))

splits your vector to 3 groups. It recycles 1:3 so every third observation 
is put into one vector.
This function randomly samples a vector and gives you list with 3 almost 
equal groups and can be easily expanded to arbitrary number of groups.

function(x) {
x1<-sample(x, ceiling(length(x)/3))
x.<-x[-x1]
x2<-sample(x., ceiling(length(x.)/2))
x3<-x[-c(x2,x1)]
list(x1,x2,x3)
}

Here are two functions which compute how many members are in each group 
and how many are left.
> 20%%3
[1] 2
> 20%/%3
[1] 6

split(x, rep(letters[1:3], each=length(x)%/%3))
and this splits your vector to 3 portions and put last 2 values to the 
first vector.

I think I gave you some insights how to handle such problem.

> 
> I want to split rnorm(20) into three equal groups.
> Note that here number of observations is not a multiple of number of 
groups. 
> For that I want an option where I can specify how to treat these extra 

which extra observations, first 2, last 2, arbitrary 2?

> observations, i.e.,  to put these observations into the 1st group or the 
last 
> group or one in each  group starting from 1st or starting from last

Only **you** know how exactly **you** want to handle it so **you** need to 
elaborate a function **yourself**. If the function shall split arbitrary 
length vector to arbitrary number of groups and put extra values to 
arbitrary groups it would not be a simple case and it would need a little 
bit of programming work.

> Also I dont want to calculate the number of observations going in each 
group beforehand.

You do not need to calculate it yourself. R will do it for you if you tell 
it.

Regards
Petr

> 
> Regards
> Utkarsh
> 
> 
> Petr PIKAL wrote: 
> Hi
> 
> r-help-bounces at r-project.org napsal dne 16.04.2009 15:23:15:
> 
> 
> Hi R,
> 
> I have explored R archives a lot but couldn't find an efficient way of 
> doing the following:
> 
> I want to split a vector into sets of equal sizes. Is there any inbuilt 
> function of doing so with the option of specifying how to treat the 
> remaining observations. For example: suppose I want to split 20 
> observations in 3 groups, then I also want the option to put the extra 
> two observations into the 1st group or the last group or one in each 
> group starting from 1st or starting from last.
> 
> I have already tried the "cut" & the "quantcut" function but of no use.
> Any help will be appreciated. Note that I am looking for an R function 
> rather than lines of code.
> 
> 
> Did you try split?
> e.g.
> split(rnorm(12), rep(1:3,4))
> gives list with 3 vectors. Then you can use lapply, sapply or other list 

> functions to make some computation.
> 
> Or I could be completely wrong what you really want as you did not 
provide 
> any example.
> 
> Regards
> Petr
> 
> 
> 
> Regards
> Utkarsh
> 
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide 
> 
> http://www.R-project.org/posting-guide.html
> 
> and provide commented, minimal, self-contained, reproducible code.
> 
> 
> 
> 
>