Date:         Tue, 7 Dec 2010 23:59:40 -0500
Reply-To:     Arthur Tabachneck <art297@ROGERS.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Arthur Tabachneck <art297@ROGERS.COM>
Subject:      Re: Groupings
Comments: To: Yu Zhang <zhangyu05@GMAIL.COM>

Same idea as Yu, but with a different suggested ordering:

proc sort data=have;
  by id date vara;
run;

data want;
  set have;
  by id date vara;
  if first.date then flagA=0;
  if not(first.vara and last.vara) then do;
    flag=1;
    if first.vara then flagA+1;
  end;
  else flag=0;
run;

HTH,
Art
---------
On Tue, 7 Dec 2010 22:40:09 -0600, Yu Zhang <zhangyu05@GMAIL.COM> wrote:

>Randy,
>
>I Assumed that the data is pre-sorted by date Id and VarA. otherwise you
>need to add a proc sort to sort the data.
>
>HTH
>
>Yu
>
>data have;
>input
>Date   :date9.          ID      VarA;
>cards;
>01FEB2010        1        24
>01FEB2010        1        51
>01FEB2010        1        51
>01FEB2010        1        54
>01FEB2010        1        62
>01FEB2010        1        62
>01FEB2010        1        62
>02FEB2010        1        5
>02FEB2010        1        7
>02FEB2010        1        9
>02FEB2010        1        9
>02FEB2010        1        9
>01FEB2010        2        1
>01FEB2010        2        1
>01FEB2010        2        1
>01FEB2010        2        5
>01FEB2010        2        7
>01FEB2010        2        7
>02FEB2010        2        8
>02FEB2010        2        10
>02FEB2010        2        12
>02FEB2010        2        14
>02FEB2010        2        14
>;
>run;
>
>data want;
>set have;
>by Date ID  VarA notsorted;
>if first.varA and last.varA  then flag=0;
>else flag=1;
>if first.id then temp=0;
>if first.varA and not last.varA then temp+1;
>if flag=0 then flagA=0;
>else flagA=temp;
>drop temp;
>run;
>
>On Tue, Dec 7, 2010 at 9:43 PM, Randy <randistan69@hotmail.com> wrote:
>
>> The rule is that on a day for a ID, if there is more than one VarA then
the
>> flag is 1 else the flag is 0.
>> For FlagA I am grouping the cases where there is more than one case of
>> VarA,
>> on a day for an ID
>>
>> Randy
>>
>> On Tue, 7 Dec 2010 19:18:57 -0800, Jeremy Miles <jeremy.miles@GMAIL.COM>
>> wrote:
>>
>> >I don't see the rule that means that 24 becomes 0, 0 , 51 becomes 1, 1,
62
>> >becomes 1, 2.
>> >
>> >But I might be missing something.  Could you explain?
>> >
>> >J
>> >
>> >On 7 December 2010 19:04, Randy <randistan69@hotmail.com> wrote:
>> >
>> >> My Data is as follows:
>> >>
>> >>
>> >> Date             ID      VarA
>> >> 01FEB2010        1        24
>> >> 01FEB2010        1        51
>> >> 01FEB2010        1        51
>> >> 01FEB2010        1        62
>> >> 01FEB2010        1        62
>> >> 01FEB2010        1        62
>> >> 02FEB2010        1        5
>> >> 02FEB2010        1        7
>> >> 02FEB2010        1        9
>> >> 02FEB2010        1        9
>> >> 02FEB2010        1        9
>> >> 01FEB2010        2        1
>> >> 01FEB2010        2        1
>> >> 01FEB2010        2        1
>> >> 01FEB2010        2        5
>> >> 01FEB2010        2        7
>> >> 01FEB2010        2        7
>> >> 02FEB2010        2        8
>> >> 02FEB2010        2        10
>> >> 02FEB2010        2        12
>> >> 02FEB2010        2        14
>> >> 02FEB2010        2        14
>> >>
>> >> I want to group VarA by adding two flags:
>> >>
>> >> Date             ID      Var      Flag     FlagA
>> >> 01FEB2010        1        24       0        0
>> >> 01FEB2010        1        51       1        1
>> >> 01FEB2010        1        51       1        1
>> >> 01FEB2010        1        62       1        2
>> >> 01FEB2010        1        62       1        2
>> >> 01FEB2010        1        62       1        2
>> >> 02FEB2010        1        5        0        0
>> >> 02FEB2010        1        7        0        0
>> >> 02FEB2010        1        9        1        1
>> >> 02FEB2010        1        9        1        1
>> >> 02FEB2010        1        9        1        1
>> >> 01FEB2010        2        1        1        1
>> >> 01FEB2010        2        1        1        1
>> >> 01FEB2010        2        1        1        1
>> >> 01FEB2010        2        5        0        0
>> >> 01FEB2010        2        7        1        2
>> >> 01FEB2010        2        7        1        2
>> >> 02FEB2010        2        8        0        0
>> >> 02FEB2010        2        10       0        0
>> >> 02FEB2010        2        12       0        0
>> >> 02FEB2010        2        14       1        1
>> >> 02FEB2010        2        14       1        1
>> >>
>> >> Please help
>> >>
>> >>   Randy
>> >>
>> >
>> >
>> >
>> >--
>> >Jeremy Miles
>> >Psychology Research Methods Wiki: www.researchmethodsinpsychology.com
>>