Date:         Wed, 27 Jan 2010 10:53:10 -0600
Reply-To:     "Data _null_;" <iebupdte@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "Data _null_;" <iebupdte@GMAIL.COM>
Subject:      Re: Deleting next observation in SAS
Comments: To: Sri <subhadrasri@gmail.com>
In-Reply-To:  <dd45c67e-9898-48fb-bd93-d68684425721@w12g2000vbj.googlegroups.com>
Content-Type: text/plain; charset=ISO-8859-1

This can be address easily with FIRST and LAST but you will also need NOTSORTED.

data want;
   set test;
   by NOTSORTED pt tmp_up;
   _obs_ + 1; * to see which obs are kept/deleted;
   if tmp_up eq 0 and not(first.tmp_up and last.tmp_up) then delete;
   run;


 pt           dtc    tmp_up    _obs_

1001    01JAN2009       1         1
        20MAR2009       1         5

1002    05JAN2009       1         6
        03FEB2009       0         7
        18MAR2009       1         8
        20APR2009       0         9
        22APR2009       1        10

1003    13JAN2009       0        11
        19FEB2009       1        12
        21FEB2009       1        13
        15APR2009       1        16
        23MAY2009       1        17



On 1/27/10, Sri <subhadrasri@gmail.com> wrote:
> Hello All,
> I'm trying to delete an observation based upon the next occurrence of
> the data. I tried to use lag function and retain statement (may be I
> did not use them in a right way) but none of them gave me the results
> I wanted.
>
> Here is the sample data creation program:
>
> data test;
> input pt $4. dtc date9. tmp_up ;
> format dtc date9.;
> cards;
> 100101JAN20091
> 100109JAN20090
> 100111JAN20090
> 100118FEB20090
> 100120MAR20091
> 100205JAN20091
> 100203FEB20090
> 100218MAR20091
> 100220APR20090
> 100222APR20091
> 100313JAN20090
> 100319FEB20091
> 100321FEB20091
> 100323FEB20090
> 100313MAR20090
> 100315APR20091
> 100323MAY20091
> ;
>
> Data looks like this:
>
> obs     pt      dtc     tmp_up
> 1       1001    1-Jan-09        1
> 2       1001    9-Jan-09        0
> 3       1001    11-Jan-09       0
> 4       1001    18-Feb-09       0
> 5       1001    20-Mar-09       1
> 6       1002    5-Jan-09        1
> 7       1002    3-Feb-09        0
> 8       1002    18-Mar-09       1
> 9       1002    20-Apr-09       0
> 10      1002    22-Apr-09       1
> 11      1003    13-Jan-09       0
> 12      1003    19-Feb-09       1
> 13      1003    21-Feb-09       1
> 14      1003    23-Feb-09       0
> 15      1003    13-Mar-09       0
> 16      1003    15-Apr-09       1
> 17      1003    23-May-09       1
>
> My logic should check if the variable tmp_up has 0 or 1 for each pt.
> If it has 0, then it should check the value of tmp_up in the next
> record. If it finds consecutive zeros, then it should delete all those
> records until it comes across value 1 in the variable tmp_up for the
> same pt. In the above data, it should delete obs=2,3,4 for pt=1001 and
> obs=14,15 for pt=1003.
>
> Any help in this matter is highly appreciated.
>
> Thanks in advance.
>