Need help with nested IF/ELSE conditions and previous observations

[alaxman]

This is my code:

data MS_AL.pers;
set MS_AL.PERSISTENCE;
by enrolid svcdate;

if first.enrolid then do;
svcdate_new=indexdt;
retain svcdate_new daysupp;
end;

else if svcdate_new+daysupp > svcdate then do;
svcdate_new = svcdate_new+daysupp;
end;

else do; svcdate_new = svcdate; end;

format svcdate_new mmddyy10.;
run;

This gives the output like this:

Essentially what I'm trying to do is to create a new column with shifted dates - svcdate_new. The condition for shifting is that:

1. First row is always indexdt

2. For all subsequent rows, if svcdate (for example in the 2nd row it is 8/22/16) < previous svcdate_new (in this case 7/25/16) + daysupp from previous row (30 in this case), then svcdate_new (for second row) = svcdate_new from previous row (7/25/16) + daysupp from previous row (30 in this case) which it computes correctly as 8/25/16.

However, for the 3rd row, svcdate_new should be 8/24/16 +30 = 9/23/2016. However, it is incorrectly computing as svcdate. 

Can someone help?

[Jagadishkatam]

Could you please post the sample data to copy and paste, in picture format it is not possible to copy and test.

[alaxman]

SVCDATE DAYSUPP INDEXDT ENROLID SVCDATE_NEW
7/25/16 30 7/25/16 128682503 7/25/16
8/22/16 30 7/25/16 128682503 8/24/16
9/13/16 6 7/25/16 128682503 9/13/16
9/15/16 2 7/25/16 128682503 9/15/16
9/20/16 30 7/25/16 128682503 10/15/16

Here you go. Thanks!
10/19/16 30 7/25/16 128682503 11/14/16
11/17/16 30 7/25/16 128682503 12/14/16

[alaxman]

Posting the output data snapshot:

SVCDATE	DAYSUPP	INDEXDT	ENROLID	SVCDATE_NEW
7/25/16	30	7/25/16	128682503	7/25/16
8/22/16	30	7/25/16	128682503	8/24/16
9/13/16	6	7/25/16	128682503	9/13/16
9/15/16	2	7/25/16	128682503	9/15/16
9/20/16	30	7/25/16	128682503	10/15/16
10/19/16	30	7/25/16	128682503	11/14/16
11/17/16	30	7/25/16	128682503	12/14/16

[Shmuel]

Try next not tested code.

To add previous row daysupp to need to save it (remain) otherwise

the current row overrides it.

data MS_AL.pers;
set MS_AL.PERSISTENCE;
by enrolid svcdate;
retain svcdate_new svdaysupp;

if first.enrolid then do;
   svcdate_new=indexdt;
   svdaysupp = daysupp;
end;

else if svcdate_new+svdaysupp > svcdate then do;
   svcdate_new = svcdate_new+svdaysupp;
end;

else do; svcdate_new = svcdate; end;
svdaysupp = daysupp;

drop svdaysupp; /* ? */
format svcdate_new mmddyy10.;
run;

[andreas_lds]

@alaxman: Posting data in usable form, not in screenshots or tables helps us to provide working solutions.

The retain-statement is non-conditional, meaning that it is always executed, even if it is part of an if- or else-block. The retain-statement marks variables, so that in during the automatic iterations values are not reset to missing, but this does not affect variables loaded from a dataset or read with input-statement. So retaining daysup is not possible, you have to create a new variable to be retained (see the solution posted by @Shmuel)

[mkeintz · Accepted solution]

You can take advantage of the fact that variables read by a SET statement are automatically retained until that SET statement is encountered again.  Usually a set statement executes once per incoming record.  But below there are two sets.  The second is executed the same number of times as the first, but it executes in batches - once per group.  As a result the variables svcdate_new and _svdaysupp are retained until over the course of each ENROLID group:

data want (drop=_:);
  set have;
  by enrolid;

  if first.enrolid then do until (first.enrolid);
    set have (rename=(indexdt=svcdate_new daysupp=_svdaysupp));
	by enrolid ;
  end;
  else if svcdate_new + _svdaysupp>svcdate then svcdate_new=svcdte_new + _svdaysupp;
  else svcdate_new=svcdate;

  _svdaysupp=daysupp;
  format svcdate_new mmddyy10. ;
run;

I think the interesting feature here is the

  if first.enrolid then do until (first.enrolid);

statement.  The "if first.enrolid" condition refers to the PRECEDING set ... by statements.  But the "do until (first.enrolid)" refers to the FOLLOWING  set ... by statements. 

[alaxman]

Thanks mkeintz! This worked. I definitely learned about the do until. That indeed was the interesting part here. Appreciate the help