Date:         Sun, 16 May 2010 14:40:07 -0400
Reply-To:     OR Stats <stats112@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         OR Stats <stats112@GMAIL.COM>
Subject:      Re: Average the Dups
Comments: To: "Zack, Matthew M. (CDC/ONDIEH/NCCDPHP)" <mmz1@cdc.gov>
In-Reply-To:  <2529A281D3499446A3C931EE75C2942501D6F580@LTA3VS012.ees.hhs.gov>
Content-Type: text/plain; charset=ISO-8859-1

Hello:

I do like how this is more parsimonious and will try it for averaging the
dups.  In the future, if we need to change how we calculate the X & Y for
our dup rows.... we may not be using PROC Summary....  What would be the
easiest (safest) way of defining our own function for when a dup occurs in
the original database?

Thank you!

On Sun, May 16, 2010 at 11:27 AM, Zack, Matthew M. (CDC/ONDIEH/NCCDPHP) <
mmz1@cdc.gov> wrote:

> Why not use a procedure like PROC SUMMARY instead of DATA step solutions
> that may be
> error-prone?
>
> ========================================================================
> =================
>  proc sort;
>    by id;
>  run;
>
>  proc summary nway;
>    by id;
>    var x y;
>    output out=avvars(keep=id xmean ymean)
>           mean=xmean ymean;
>  run;
> ========================================================================
> ==================
>
> Matthew Zack
>  -----Original Message-----
> From: SAS(r) Discussion [mailto:SAS-L@LISTSERV.UGA.EDU] On Behalf Of OR
> Stats
> Sent: Sunday, May 16, 2010 8:58 AM
> To: SAS-L@LISTSERV.UGA.EDU
> Subject: Average the Dups
>
> Hello:
>
> I have a dataset that is
>
> id x y
>
> Every once in a while in the dataset there would be two consecutive rows
> with the same unique id.  In cases of these duplicates, I would like to
> take
> the average of the two rows.  How can I do this in SAS?  For example the
> data would like this
>
> 1   x1   y1
> 2   x2   y2
> 3   x3a y3a
> 3   x3b y3b
> 4   x4   y4
> .
> .
> .
> 74  x74a  y74a
>  74  x74b  y74b
>
> .
> .
> .
> The final table would look like
>  1   x1   y1
> 2   x2   y2
> 3   Average(x3a,x3b)  Average(y3a,y3b)
> 4   x4   y4
> .
> .
> .
> 74 Average(x74a,x74b)  Average(y74a,y74b)
>
> .
> .
> .
> Can we do this in a single datastep or PROC SQL?  Which is faster?
>
>
>