@AshleyBright: Forget Notepad. When you copy-and-paste into it, what you see may not be what you have in your SAS variable. Instead, locate the observation number with the variable value you want to investigate. Suppose this number is 123, your variable in question is called S, and the data set is called HAVE. Run the step: data _null_ ; obs_num = 123 ; set have point = obs_num ; do pos = 1 to length (s) ; b = char (s, pos) ; put b=$hex2. b= pos= ; end ; stop ; run ; and look at the log. For the first 14 positions of the string you've exemplified here, the step will print:   b=22 b=" pos=1 b=22 b=" pos=2 b=22 b=" pos=3 b=74 b=t pos=4 b=65 b=e pos=5 b=73 b=s pos=6 b=74 b=t pos=7 b=3A b=: pos=8 b=7B b={ pos=9 b=22 b=" pos=10 b=22 b=" pos=11 b=22 b=" pos=12 b=22 b=" pos=13 b=74 b=t pos=14   This will represent the true character content of your string, and you'll be able to determine whether the "quotes" you see in the viewer are really quotes or something else and/or if there're some "hidden" characters you can't see. If some characters that look to you like quotes are in reality other characters, such as not-printables, note their hex values in the log and use them as hex literals with the COMPRESS function to kill them.    Now if you determine from the log that they are real double quotes, only multiplied for some reason, and you want to collapse any number of consecutive double-quotes to a single double-quote, run: data want ; set have ; do while (find (s, '""')) ; s = tranwrd (s, '""', '"') ; end ; run ; Or, if you want to leave no more than 2 consecutive double-quotes in the value of your variable, change the DO loop to: do while (find (s, '"""')) ; s = tranwrd (s, '"""', '""') ; end ; Anything is possible if you really know what's in your string and what and in which manner you want to change (as long as there's a pattern). But looking at the SAS data after moving them out of the SAS system, especially by copying and pasting, is error-prone.   Kind regards Paul D.           ... View more

Assuming that the 3rd row in your want (I named this as ActualWant) where num = 3 is actually supposed to be num=2 then this might work for you.   It would have been more helpful to have an idea of what exactly you wanted the code to do and had a 'data have' that contains enough examples (such as id 102) for your 'data want'.   If this doesn't work as intended please explain why and it can then be fixed.   data have; infile datalines dsd truncover; input id:32. num:32. std:DATE9. endt:DATE9.; format std endt yymmddd10.; datalines4; 101,2,10MAY2016,15MAY2016 101,3,16MAY2016,22MAY2016 101,4,23MAY2016,05JUN2016 101,5,06JUN2016,19JUN2016 101,6,20JUN2016,20JUN2016 ;;;; run; data want; set have (rename=(std=_std endt=_endt)); by id; format std endt yymmddd10.; if first.id then do; /*Potential problem if _std and _endt are on the same day here. Just modify the if condition to if first.id and _endt - _std GE 1 then do; which also checks that the enddate is after the startdate */ std = _std; endt = std + 1; output; Iterator = 0; do until (endt ge _endt or Iterator ge 2); Iterator = sum(Iterator,1); std = std + 2; endt = min(_endt,endt + 2); if Iterator eq 2 then endt = _endt; *Only want 2 iterations and then put the solution; output; end; *end until; end; *end if; else do; std = _std; endt = _endt; output; end; *end else if; drop Iterator _std _endt; run; data Actualwant; infile datalines dsd truncover; input id:32. num:32. std: DATE9. endt: DATE9.; format std endt yymmddd10.; datalines4; 101,2,10MAY2016,11MAY2016 101,2,12MAY2016,13MAY2016 101,3,14MAY2016,15MAY2016 101,3,16MAY2016,22MAY2016 101,4,23MAY2016,05JUN2016 101,5,06JUN2016,19JUN2016 101,6,20JUN2016,20JUN2016 102,2,12JAN2018,13JAN2018 102,2,14JAN2018,15JAN2018 102,2,16JAN2018,19JAN2018 102,3,20JAN2018,28JAN2018 102,4,29JAN2018,12FEB2018 102,5,13FEB2018,18FEB2018 102,6,19FEB2018,28FEB2018 ;;;; run;   ... View more

This is what I was able to come up with.  It matches your WANTED.   data wanted; infile datalines dsd; input id num (start end) (:date9.); format start end date9.; datalines4; 101,2,12MAR2018,15MAR2018 101,3,16MAR2018,18MAR2018 101,3,19MAR2018,19MAR2018 101,4,20MAR2018,29MAR2018 102,2,12JAN2018,15JAN2018 102,2,16JAN2018,18JAN2018 102,2,19JAN2018,19JAN2018 102,3,20JAN2018,28JAN2018 102,4,29JAN2018,12FEB2018 102,5,13FEB2018,18FEB2018 102,6,19FEB2018,28FEB2018 ;;;; run; data have; infile datalines dsd; input id num (start end) (:date9.); format start end date9.; datalines4; 101,2,12MAR2018,14MAR2018 101,3,15MAR2018,19MAR2018 101,4,20MAR2018,29MAR2018 102,2,12JAN2018,19JAN2018 102,3,20JAN2018,28JAN2018 102,4,29JAN2018,12FEB2018 102,5,13FEB2018,18FEB2018 102,6,19FEB2018,28FEB2018 ;;;; run; data testV / view=testv; d=0; do until(last.id); set have; by id start; do date=start to end; if d lt 8 then d + 1; if 1 le d le 4 then g=1; else if d le 7 then g=2; else g=3; if g eq 3 then do; output; leave; end; else output; end; end; format date date9.; run; data test; do until(last.id); set testV; by id g start; if g in(1,2) then do; if first.g then do; nstart=date; nnum=num; end; if last.g then do; end=date; start=nstart; num=nnum; output; end; end; else do; start=date; output; end; end; format nstart date9.; drop g d nstart nnum date; run; proc print; run; proc compare base=wanted compare=test listequalvars; run; ... View more

Hi @vraj1,   There are at least two possible interpretations of your criteria and hence different solutions:   The "first 2 days" are days 1 and 2 (according to variable DAY), the "next 2 days" are days 3 and 4 and so on. data want; set test(drop=exd); if period='A' then do; if ext='pine' then exd=2*min(int((day+1)/2),3); else if ext='ozi' then exd=5*min(int((day+1)/2),3); end; run; The "first 2 days" with EXT='pine' correspond to the first two observations with PERIOD='A' and EXT='pine' within an ID regardless of DAY (but assuming chronological sort order), etc. data want; set test(drop=exd); by id day; if first.id then call missing(_cp, _co); if period='A' then do; if ext='pine' then do; _cp+1; exd=2*min(int((_cp+1)/2),3); end; else if ext='ozi' then do; _co+1; exd=5*min(int((_co+1)/2),3); end; end; drop _:; run; ... View more