[R] How to replicate SAS by group processing in R

[David Winsemius]

On Oct 10, 2012, at 11:09 AM, ramoss wrote:

> Hello,
> 
> I am trying to re-code all my programs from SAS into R.
> 
> In SAS I use the following code:
> 
> proc sort data=upper;
> by tdate stock_symbol expire  strike;
> run;
> data upper1;
>  set upper;
>  by tdate stock_symbol expire  strike;

I must have forgotten my SAS. (It was a lng time ago I will admit.)  Would that have succeeded with the inclusion of 'strike' in that 'by' list?

>  if first.expire then output;
>  rename strike=astrike;
> run;
> 
> on the following data set:
> 
> tdate	stock_symbol	expiration 	strike
> 9/11/2012	C	                   9/16/2012	11
> 9/11/2012	C	                   9/16/2012	12
> 9/11/2012	C	                   9/16/2012	13
> 9/12/2012	C	                   9/16/2012	14
> 9/12/2012	C	                   9/16/2012	15
> 9/12/2012	C	                   9/16/2012	16
> 9/12/2012	C	                    9/16/2012	17
> 
> to get the following results:
> tdate	stock_symbol	expiration 	strike
> 9/11/2012	C	                   9/16/2012	11
> 9/12/2012	C	                   9/16/2012	14

> dat[tapply(1:nrow(dat), list( dat$stock_symbol,  dat$tdate), FUN= function(x) head(x,1) ), ]
      tdate stock_symbol expiration strike
1 9/11/2012            C  9/16/2012     11
4 9/12/2012            C  9/16/2012     14
> 

> 
> How would I replicate this kind of logic in R?
> I have seen PLY & data.table packages mentioned but don't see how they would
> do the job.

You must mean the 'plyr' package;  there is no "PLY'. I'm sure the 'ddply' function or data.table could do this.

Here's another way with the R 'by' function which is then row-bound using 'do.call':

> do.call( rbind, by(dat, list( dat$stock_symbol,  dat$tdate), FUN= function(x) head(x,1) ) )
      tdate stock_symbol expiration strike
1 9/11/2012            C  9/16/2012     11
4 9/12/2012            C  9/16/2012     14

-- 

David Winsemius, MD
Alameda, CA, USA