Date:         Thu, 26 Jul 2001 09:39:06 -0400
Reply-To:     "Diskin, Dennis" <Dennis.Diskin@PHARMA.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "Diskin, Dennis" <Dennis.Diskin@PHARMA.COM>
Subject:      Re: Data restructuring
Comments: To: Paolo ORIFICI <Orifici@LACAJA.COM.AR>
Content-Type: text/plain

Paolo,

Here's one way using datastep programming logic:

data  test;
length type $ 8;
INPUT CLAIM TYPE PAID;

cards;
 1              body                  100
 1              body                  70
 1              body                  30
 2              material              5
 2              material              7
 3              body                  70
 3              material              3
 3              material              8
 3              body                  30
 4              material              12
 4              body                  20
 5              body                  120
;

proc sort data=test;
by claim;

data test(drop=paid type);;
set test;
by claim;
retain type_1 paid_sum;
if first.claim then
   do;
      paid_sum = 0;
      type_1 = type;
   end;
paid_sum + paid;
if type ne type_1 then type_1 = 'mixed';
if last.claim;

proc print;

run;

OBS    CLAIM    TYPE_1      PAID_SUM

 1       1      body           200
 2       2      material        12
 3       3      mixed          111
 4       4      mixed           32
 5       5      body           120

> -----Original Message-----
> From: Paolo ORIFICI [SMTP:Orifici@LACAJA.COM.AR]
> Sent: Thursday, July 26, 2001 8:44 AM
> To:   SAS-L@LISTSERV.UGA.EDU
> Subject:      Data restructuring
>
> HI list members,
>
> I'm new to SAS. I want to restructure my SAS dataset in order to obtain a
> single observation from multiple records.
>
> The original data set has the following structure:
>
> Claim       Type                 Paid
> 1              body                  100
> 1              body                  70
> 1              body                  30
> 2              material              5
> 2              material              7
> 3              body                  70
> 3              material              3
> 3              material              8
> 3              body                  30
> 4              material              12
> 4              body                  20
> 5              body                  120
>
>
> The new dataset should have this new structure:
>
> Claim      Type_1            Paid_sum
>
> 1             Body                  200
> 2             Material              12
> 3             Mixed                 111
> 4             Mixed                 32
> 5             Body                   120
>
>
>
> Thank you in advance.
>
> Paolo Orifici