Conditionally selecting the next two records

[PDevi]

Edited  to add additional requirements. Sorry for the confusion.

I have a data set that looks like below:

Sample data:

Id	Term	Audit_type	Prog	Level	Load
1	Fall 2015	1	ABCD	01	F
1	Fall 2015	2	ABCD	01	F
1	Winter 2016	1	ABCD	02	F
1	Winter 2016	2	ABCD	02	F
1	Fall 2016	1	ABCD	03	F
1	Fall 2016	2	ABCD	03	F
1	Winter 2017	1	ABCD	04	F
1	Winter 2017	2	ABCD	04	F
2	Fall 2012	1	LMNO	01	F
2	Fall 2012	2	LMNO	01	F
2	Winter 2013	1	LMNO	02	F
2	Winter 2013	2	LMNO	02	F
2	Spring 2013	1	LMNO	03	P
2	Fall 2013	1	LMNO	03	F
2	Fall 2013	2	LMNO	03	F
2	Winter2014	1	WXYZ	01	F
2	Spring 2014	1	PQRS	01	P
2	Spring 2014	2	PQRS	01	P
2	Fall 2014	1	PQRS	01	P
2	Winter 2015	1	PQRS	01	F
2	Winter 2015	2	PQRS	01	F
3	Fall 2016	1	EFGH	01	F
3	Fall 2016	2	EFGH	01	F
3	Winter 2017	1	EFGH	02	F
3	Spring 2017	1	EFGH	02	P
3	Spring 2017	2	EFGH	02	P
3	Fall 2017	1	EFGH	02	F
3	Fall 2017	2	EFGH	02	F

What I need is to select the next two records, for when the student comes back after level '01' in the first term, such that the audit_type is always '2'. And this has to be for every id, term, program combination.

For id #2, the student continues and moves forward in the first program. The second program is discontinued and s/he changes it to a third program, which is continued. I need to bring in information for the first and the third programs both.

So basically, my selection needs to be as below:

Desired Solution:

Id	Term	Audit_type	Prog	Level	Load
1	Winter 2016	2	ABCD	02	F
1	Fall 2016	2	ABCD	03	F
2	Winter 2013	2	LMNO	02	F
2	Fall 2013	2	LMNO	03	F
2	Spring 2014	2	PQRS	01	P
2	Winter 2015	2	PQRS	01	F
3	Spring 2017	2	EFGH	02	P
3	Fall 2017	2	EFGH	02	F

I would appreciate it if someone could please help. Thanks in advance.

[Astounding]

This might look easier in the morning, but it should work:

data want;

set have;

by id term notsorted;

if first.id then do;

   term_count=1;

   output_count=0;

end;

else if first.term then term_count + 1;

if term_count > 1 and output_count < 2 and audit_type=2 then do;

   output;

   output_count + 1;

end;

drop term_count output_count;

run;

[PDevi]

@Astounding It does work, thank you very much. But I have an added layer of requirement that I had forgotten to include in the sample data. When I tested your code, that's when I realized my mistake. The sample data and the desired outcome have been edited. I am sorry for the over sight.

[Astounding]

If I understand the new requirements properly, a very similar program can do the trick.

proc sort data=have;

   by id prog;

run;

data want;

set have;

by id prog term notsorted;

if first.prog then do;

   term_count=1;

   output_count=0;

end;

else if first.term then term_count + 1;

if term_count > 1 and output_count < 2 and audit_type=2 then do;

   output;

   output_count + 1;

end;

drop term_count output_count;

run;

It's untested at this point, so see what you get from it.

[ChrisNZ]

Like this?

data WANT;
  set HAVE;
  by ID;
  if first.ID then call missing(AFTER_FIRST,OUTPUT_NO);
  if ID=lag(ID) and TERM ne lag(TERM) then AFTER_FIRST+1;
  if AUDIT_TYPE=2 & AFTER_FIRST then do;
    OUTPUT_NO+1;
    if OUTPUT_NO in (1,2) then output;
  end;
run;

id	Term	y	Audit_type	Prog	Level	Load
1	Winter	2016	2	ABCD	2	F
1	Fall	2016	2	ABCD	3	F
2	Spring	2014	2	PQRS	1	P
2	Winter	2015	2	PQRS	1	F
3	Spring	2017	2	EFGH	2	P
3	Fall	2017	2	EFGH	2	F