Date:         Tue, 17 Mar 2009 14:51:46 -0400
Reply-To:     Akshaya <akshaya.nathilvar@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Akshaya <akshaya.nathilvar@GMAIL.COM>
Subject:      Re: how to substitute a missing value with previous record?
Comments: To: Ya Huang <ya.huang@amylin.com>
In-Reply-To:  <200903171839.n2HAmaGR026054@malibu.cc.uga.edu>
Content-Type: text/plain; charset=ISO-8859-1

I posted DOW solution(adopted from previous Data _null_'s post) earlier.
That will fill the consecutive missing spots from previous non-missing
value.
Data want;
 do until(eof);
        set have end=eof;
        locf=ifn(missing(score),locf,score);
        output;
 end;
Run;

On Tue, Mar 17, 2009 at 2:39 PM, Ya Huang <ya.huang@amylin.com> wrote:

> This won't work if you have more than one consecutive missing:
>
> data xx;
> input x $ value;
> cards;
> 3/17/2009  2
> 3/18/2009  3
> 3/19/2009  7
> 3/20/2009  .
> 3/21/2009  .
> 3/22/2009  4
> 3/23/2009  9
> ;
>
> data better;
> drop previous;
>  previous = value ;
>   set xx ;
> value1 = coalesce( value, previous);
> run;
>
> proc print;
> run;
>
> value       x        value1
>
>  2      3/17/200       2
>  3      3/18/200       3
>  7      3/19/200       7
>  .      3/20/200       7
>  .      3/21/200       .
>  4      3/22/200       4
>  9      3/23/200       9
>
> the missing in value1 is expected to be 7.
>
> On Tue, 17 Mar 2009 13:53:34 -0400, Peter Crawford
> <peter.crawford@BLUEYONDER.CO.UK> wrote:
>
> >i like the smallest.
> >data better ;
> >  drop previous;
> >  previous = value ;
> >  set original ;
> >  value= coalesce( value, previous);
> >run;
> >
> >
> >On Tue, 17 Mar 2009 08:20:56 -0700, Yi Zhang <zhangyiosu@GMAIL.COM>
> wrote:
> >
> >>hi guys,
> >>I has a time series data with some missing values. I want to
> >>substitute the missing value with the value in the last time. For
> >>example, I have
> >>3/17/2009               2
> >>3/18/2009               3
> >>3/19/2009               7
> >>3/20/2009               3
> >>3/21/2009               missing
> >>3/22/2009               4
> >>3/23/2009               9
> >>
> >>I want to have
> >>3/17/2009               2
> >>3/18/2009               3
> >>3/19/2009               7
> >>3/20/2009               3
> >>3/21/2009               3
> >>3/22/2009               4
> >>3/23/2009               9
> >>
> >>Any idea how to do this in a DATE STEP?
>



--
AkshayA!