Blimey that was an ask :smileyshocked:  Well, what I would suggest is getting the same range on all data, in this instance I have just assumed the 12 months, and I would also suggest to get rid of the full date setup, and just work on months/years as that is the real aim, the 01 doesn't help.  Once you have all the data across on a the same range, you can then loop across each element in this range and output the mags for that month.  Then a simple sort (note, its not the nicest coding in the world, you could for instance drop the last two statements if you can stick with months). data subscriptions;    infile datalines;    input @01 id 1.          @03 magazine $4.          @08 date_from date9.          @18 date_to date9.;    datalines;                      1 mag1 01jan2014 01mar2014 1 mag2 01feb2014 01apr2014 1 mag3 01feb2014 01apr2014 1 mag4 01mar2014 01may2014 2 mag1 01jan2014 01jul2014 2 mag2 01feb2014 01apr2014 ; run; data have;   set subscriptions;   array mnths {12} $200.;   do i=1 to 12;     if date_from <= mdy(i,1,2014) < date_to then mnths{i}=magazine;   end; run; proc sql;   create table RESULTS   (     ID num,     MAGAZINE char(200),     MONTH char(200)   ); quit; data _null_;   do i=1 to 12;     call execute(%nrstr(                  'proc sql;                     insert into RESULTS                     select ID,MAGAZINE,"'||strip(put(i,best.))||'" as MONTH                     from WORK.HAVE                      where MNTHS'||strip(put(i,best.))||' is not null;                   quit;'));   end; run; data results;   set results;   attrib date_from date_to format=date9.;   select(month);     when("1") do; date_from='01JAN2014'd; date_to='01FEB2014'd; end;     when("2") do; date_from='01FEB2014'd; date_to='01MAR2014'd; end;     when("3") do; date_from='01MAR2014'd; date_to='01APR2014'd; end;     when("4") do; date_from='01APR2014'd; date_to='01MAY2014'd; end;     when("5") do; date_from='01MAY2014'd; date_to='01JUN2014'd; end;     when("6") do; date_from='01JUN2014'd; date_to='01JUL2014'd; end;     when("7") do; date_from='01JUL2014'd; date_to='01AUG2014'd; end;     when("8") do; date_from='01AUG2014'd; date_to='01SEP2014'd; end;     when("9") do; date_from='01SEP2014'd; date_to='01OCT2014'd; end;     when("10") do; date_from='01OCT2014'd; date_to='01NOV2014'd; end;     when("11") do; date_from='01NOV2014'd; date_to='01DEC2014'd; end;     when("12") do; date_from='01DEC2014'd; date_to='01JAN2015'd; end;     otherwise;   end; run; proc sort data=results;   by id magazine; run; ... View more

Mmm, nice, but it does have leading blanks if your string isn't > 5 in the first place.  Also, if you have a long program (and maybe if length isn't even setup there), its not easy to see what is happening as the code just reverses twice.  Substr or trimming make the statement very explicit. ... View more

IMO defining global macro variables in a macro is not a good idea.  It will probably be ok if your one person in a local install of SAS where your unlikely to use it much or overwrite, assume a certain value.  Why not just add a macro variable in your call to define what you want the output dataset to be called, then there is no need for global variables as the program calling the macro knows exactly what the output will be created as: %macro GetData(mpfile=,Output_dataset=);   filename TempFile temp lrecl=4000;   data _null_;     infile "&mpfile" lrecl=4000;     file TempFile;     input; if lengthc(_infile_) > 0 then do;     if substr(_infile_,1,1) in ("$","!","*") then put _infile_;     end;   run;   proc import datafile = TempFile out = &Output_dataset. dbms = csv replace;     getnames=yes;     guessingrows=5000;   run; %mend; ... View more

I think you would be better served then with either creating a compiled function, or an inline macro. Compiled functions are under fcmp, and I wont go into that. Inline macros though do help.  The idea is to have the macro resolve as a piece of code within your datastep.  For instance in you conversion: %macro Convert (Value=,Conv_Rate=);      &Value * &Conv_Rate. %mend; data temp;      new_value=%Convert (Value=123, Conv_Rate=0.56); run; Here is an old favourite example from some time back, it basically resolves itself out to a number: %macro nobs(ds); %local nobs dsid rc; %let dsid=%sysfunc(open(&ds)); %let nobs=%sysfunc(attrn(&dsid,NLOBSF)); %let rc=%sysfunc(close(&dsid)); &nobs %mend nobs; You can use it in: data test;      set sashelp.cars;      number_of_obs=%nobs(sashelp.cars); run; ... View more

Well, I don't know about that course or its notes. The %local is correct. The call symputx is also correct, but you have to remember that the default operation for that function, i.e. the value for table if you don't specify otherwise, will be G for global give or take some rules on the SYMPUT function. As for the %let and select into, I don't know off the top of my head, would need to check. TBH I don't think many actually care about scope as long as they can access it when wanted. ... View more