Can you explain the expansion for ID=1 in more detail ? ... View more

I'm a beginner to SAS That is a good justification for not jumping into Macro.  There is huge utility in the DATA Step language without a need for any macro.   Far less confusing would be to use a variable based array to create the variable names amount_00-amount_76.  Your complex formula would compute the index for assignment.   data want; array amts amount00-amount76; index = ((Fdevyr - Flye) * 4 + int((Fdevm - 1) / 3) - int((FM - 1) / 3)); if component = 'Counts' then amts(index) = count; run;   ... View more

Really like tracking the new meds, as they are the core feature of the processing.   You might want to make OLDMEDS longer to accomodate the new meds it is accumulating.   length OLDMEDS $500 NEWMEDS MED $20;   ... View more

What percentages were you expecting to see ?   Is it possible GROUP=TERM G100 is computing percentages different than you expectation ?   The chart numbers do appear correct.  These are the numbers I see Question 1 June 2019 Response N % (fraction) Agree     15 79 15/19 Missing 2 10.5 2/19 Neutral 2 10.5 2/19 Question 2 June 2019 Agree 7 36.8 7/19 Disagree 2 10.5 2/19 Missing 1 5.3 1/19 Neutral 9 47.4 9/19 ... View more

A double transpose can reshape the data into the form you desire.  A double transpose is needed because two variables are being pivoted into columns.   data have; attrib id length=8 startdate enddate informat=ddmmyy10. format=ddmmyy10.; input ID Startdate Enddate; datalines; 1 01/01/2001 31/12/2001 1 01/01/2002 04/05/2002 1 06/07/2002 30/11/2002 2 05/05/2001 05/06/2001 2 07/07/2001 15/10/2001 3 01/04/2002 08/09/2003 ; data have_view / view=have_view; set have; by id; if first.id then seq=1; else seq+1; run; proc transpose data=have_view out=name_value; by id seq; var startdate enddate; run; proc transpose data=name_value out=want(drop=_name_); by id; id _name_ seq; run; A second way is to perform an 'array-based transposed'.  This requires an apriori scan of the data to find the size of the largest id group.   proc sql noprint; select max (subselect.freq) into :across_count trimmed from (select id, count(*) as freq from have group by id) as subselect; %put NOTE: &=across_count; data want2; do _n_ = 1 by 1 until (last.id); set have; by id; array starts startdate1-startdate&across_count; array ends enddate1 -enddate&across_count; starts [_n_] = startdate; ends [_n_] = enddate; end; attrib startdate: enddate: format=ddmmyy10. informat=ddmmyy10.; drop startdate enddate; run; If you want the array based solution to have a column order that is like start1, end1, start2, end2, etc... you will need a macro to generate the variable names in that order.  Example: %macro enum(min=1, max=, template=); %local index; %do index = &min %to &max; %unquote(&template) %end; %mend; data want3; do _n_ = 1 by 1 until (last.id); set have; by id; attrib %enum(max=&across_count, template=%nrstr(startdate&index enddate&index)) format=ddmmyy10. informat=ddmmyy10. ; array starts startdate1-startdate&across_count; array ends enddate1 -enddate&across_count; starts [_n_] = startdate; ends [_n_] = enddate; end; drop startdate enddate; run;       ... View more

Can you clarify the question by showing which of the data is correct and which is deemed incorrect by double checking ? ... View more

Thinking outside the box, you can coerce an input statement to read a concatenation of the flag variables transformed to appear like an csv input line.  Lots of tricks mashed into this answer:   DATA Step with SET and INFILE Dummy DATALINES for INPUT to introduce _INFILE_ INFILE CARDS with TRUNCOVER to prevent LOST CARD Held input with @@ to prevent dataline advancement INPUT @1 to reposition parse start point INPUT with (one:) variable list specifier to read into an array INPUT with (binary32.:) informat list to read 1's as binary data upto next delimiter (,) using : list input modifier MAX of one: to determine largest run of 1's LOG2 of 1+MAX to get number of 1's in largest value that must be 2^p - 1 * faux data with 32 neg flag variables; data have; do patid = 1 to 20; array neg neg1-neg32; do over neg; neg = ranuni(123) < 0.35; end; output; end; run; data want; attrib longest_run length=8; set have; infile cards dlm=',' truncover; if _n_ = 1 then input @1 @@; * activate _infile_; _infile_ = translate(cats(of neg:),',','0'); * repurpose _infile_ value; * put _infile_; array one one1-one32; input @1 (one:) (binary32.:) @@; * read the _infile_ (input buffer); _error_ = 0; longest_run = log2 (1+max(of one:)); drop one:; cards; . run; ... View more