Date:         Wed, 21 Apr 2010 09:35:37 -0700
Reply-To:     mlhoward@avalon.net
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Mary <mlhoward@AVALON.NET>
Subject:      Re: Filling in Gaps?
Comments: To: Gerhard Hellriegel <gerhard.hellriegel@T-ONLINE.DE>
Content-Type: text/plain; charset="UTF-8"

Gerhard,

Thanks very much- these are excellent ideas to think about below, and the real issue is more complicated (adherence).

Sorry about the "want" data not being too good-maybe I'm trying to simulate the actual consumer in my "wanted" data, where they sometimes drop a pill in the sink and thus don't really have an extra on hand :-)

-Mary



--- gerhard.hellriegel@T-ONLINE.DE wrote:

From: Gerhard Hellriegel <gerhard.hellriegel@T-ONLINE.DE>
To: SAS-L@LISTSERV.UGA.EDU
Subject:      Re: Filling in Gaps?
Date:         Wed, 21 Apr 2010 12:19:50 -0400

you should define a array of days, something like

  array pill(*) day1-day10;

then you can scroll through the array and add the extra pills in a pill-
container. If you detect a missing day, take a pill from container.

container=0;
do i=1 to dim(pill);
  if pill(i)>1 then container = sum(container, pill(i)-1);
  if i>1 and pill(i)=. then do;
     if container>0 then do;
       pill(i)=1;
       container=container-1;
     end;
  end;
end;

what is not clear: what happens if there are gaps at start and extra pills
later?

. . 1 2 2 1 1 1

Result?

1 1 1 1 1 1 1 1  ?

Or you need pills (gaps) which come later (container is not full enough).

In that case you might scroll 2 times through the days. First time for
collecting pills, second for distributing them.

Other question is: what if there are not all pills needed? (at the end the
container has still extra-pills). Or what is if you don't have empty days,
still fill the container?

If the container contains pills at the end, I think it depends on the kind
of pills if you can throw them away....

2 1 1 1 1 1 1 1 1 1

result?

1 1 1 1 1 1 1 1 1 1  and one left?

or

1 1 1 1 1 1 1 1 1 2

or ???

Gerhard




On Wed, 21 Apr 2010 09:10:58 -0700, Mary <mlhoward@AVALON.NET> wrote:

>Hi,
>
>I've got a data set like this:
>
>data test2;
>infile cards missover;
>input id day1 day2 day3 day4 day5 day6 day7 day8 day9 day10;
>cards;
>001 1 1 1 1 2 2 . . . . 1 1
>002 2 2 2 . 1 . . . . . 1 .
>;
>run;
>
>These are the number of pills someone has on hand each day.  What I want
is to fill in the gaps with the extra pills on hand, so I want a result
like this:
>
>001 1 1 1 1 1 1 1 1 . . 1 1
>002 1 1 1 1 1 1 . . . . 1 .
>
>Ideas would be appreciated; I'd like to keep this wide rather than
transpose it long if possible.
>
>-Mary