Date:         Tue, 13 May 2008 00:51:25 -0400
Reply-To:     Tom Smith <need_sas_help@YAHOO.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Tom Smith <need_sas_help@YAHOO.COM>
Subject:      Re: A Dataset Modification help Needed to create another dataset
Comments: To: Muthia Kachirayan <muthia.kachirayan@GMAIL.COM>

On Mon, 12 May 2008 21:15:07 -0400, Muthia Kachirayan
<muthia.kachirayan@GMAIL.COM> wrote:

>On Mon, May 12, 2008 at 3:10 PM, Tom Smith <need_sas_help@yahoo.com>
wrote:
>
>> I have a following dataset having variables:SUB, HR, PQ, HJ as bellow:
>>
>> SUB     HR        PQ                HJ
>> 1003   post      Hadsone          Cause
>> 1003   pre       Green bay        Location
>> 1004   post      Jamsed           Reason
>> 1004   post      Winona           City
>> 1004   Pre       PullMan          ZIp
>>
>> Need to create another dataset which will be by the variable SUB as
below
>> (means
>> it will have only one observation for each SUB variable's value ( Not a
>> proc report):
>>
>>
>>
>>
>>
>> SUB     HR        PQ                HJ
>> 1003   post      Hadsone          Cause
>>       pre       Green bay        Location
>> 1004   post      Jamsed           Reason
>>       post      Winona           City
>>       Pre       PullMan          ZIp
>>
>>
>> Can you help me ? I think by Proc tabulate we can do it. But
>> I forget the statement to use to create a dataset from Proc tabulate.
>> Please help me out even if you have any better approach other than
>> tabulate.
>> Thanks a lot for helping.
>>
>
>Tom,
>
>Howard gave a solution giving 5 records of output for the sample data set.
>If you have meant it to be one record per SUB, then here is an array
>approach. This solution assumes that there is no more than 3 rows per SUB
as
>a requirement for dimensioning the v[ ] array. You can add additional
>elements(3) to v[ ] per each row.
>
>data need;
>do _n_ = 1 by 1 until(last.sub);
>   set have;
>   by sub;
>   length v1 $8 v2 $9 v3 $8 v4 $8 v5 $9 v6 $8 v7 $8 v8 $9 v9 $8;
>   array s HR PQ HJ;
>   array v[9] v1-v9;
>   do over s;
>       v[(_n_ - 1) * 3 + _i_] = s;
>   end;
>end;
>drop hr pq hj;
>run;
>
>proc print data = need;
>run;
>
>Muthia Kachirayan

I had posted this problem earlier. But guesss I could not make it clear in
my last post. So I think I cofused those who tried to help me out. So I am
trying to make it clear below and more simple:

I have a following dataset having variables:SUB, HR as bellow:

SUB     HR
1003   post
1003   pre
1004   post
1004   post
1004   Pre
1005   post
1005   pre
1005   failed
1005   never tried

Need to create another dataset from it which will be by the variable SUB
as below(means it will have only one observation for each SUB variable's
value and all the values for HR for that perticular value of SUB will be
in the same record one after another. Plz do not use Proc Report. I need a
dataset.
SUB     HR
1003   post /
       pre
1004   post /post /
       Pre
1005   post / pre /
       failed /
       never tried