You should have placed a SAMPLE Data Set that represent your real Data SET. I have given a solution based on your sample Data Set.
Now you say there will be varying number of records - 2 records instead of 3. I hope you don't come with newer issues with your original Data Set.
The solution is based on first creating triplets with adding "NONE" when "Charity" is missing. Next the same program without using LAST.ID. We can simply check for either "NONE" or "Charity" to output a record at a time.
data source;
INFILE DATALINES ;
input ID VAR & $19. CLASSIFICATION $ ;
DATALINES;
201 PepsiCo XYZ
201 111 main street XYZ
201 Charity XYZ
331 CocaCola XYZ
331 1823 unicorn street XYZ
331 Charity XYZ
331 Nike XYZ
331 123 brock avenue XYZ
331 Charity XYZ
331 Adidas XYZ
331 14 Market street XYZ
331 Charity XYZ
354 McDonalds XYZ
354 111 Obesity road XYZ
354 Charity XYZ
354 KFC XYZ
354 111 Fat Avenue XYZ
441 Burger King XYZ
441 112 Skinny ROAD XYZ
441 Adidas XYZ
441 Socks Blvd XYZ
441 Tiege Henley XYZ
441 Skincare Road XYZ
441 Charity XYZ
;
RUN;
proc print data = source;
run;
data have;
set source;
if Var = "Charity" then i = 0;
else i+1;
if i = 3 and Var ^= "Charity" then do;
temp = Var;
Var = "NONE";
output;
Var = temp;
output;
temp =' ';
i = 1;
end;
else output;
drop temp i;
run;
proc print data = have;
run;
The Data Set, HAVE, will have triplets either with "NONE" or "Charity".
The next the old program without LAST.ID checks.
data clean;
length ID 8 Name $10 Address $19 Classification $8 Var $19;
do i = 1 by 1 until(VAR in ("NONE","Charity"));
set have;
select(mod(i,3));
when(1) Name=Var;
when(2) Address = Var;
when(0) Classification = Var;
otherwise;
end;
end;
drop i Var;
run;
proc print data = clean;
run;
