Re: Question about aggregate() and statistics with greater than

[Sundar Dorai-Raj <sundar.dorai-raj@PDF.COM>]

Li, Mike wrote:

> Can you suggest a way to implement the aggregate function which allows 
> it to return more than a single value for each subset of the data?
> i.e. Suppose you have the following dataset:
> dat _ data.frame(ID=c(rep(1,4),rep(2,4),rep(3,4)), TIME=rep(c(0,0,1,1), 3))
>
> The aggregate function accepts arguments to "fun" that return single 
> statistics, such as mean.
> e.g. The following code is successful:
> aggregate(dat$TIME, list(dat$ID),mean)
>
> However, I want to have the aggregate function perform functions on the 
> subsetted data that return several values, such as by using the "unique" 
> function.
> i.e. The following code fails:
> aggregate(dat$TIME, list(dat$ID),unique)
>
> Can you suggest an efficient way to perform operations on subsets of data?
> Thanks much,
> Mike Li
Take a look at ?by:

by(dat$TIME, list(dat$ID), unique)

--sundar

P.S. "_" is deprecated as assignment as of S-PLUS 6.0 and should not be 
used.