We're all guessing, but this might be what you want:  compute today's opening.   If so, the opening for the first day can't be computed.  But here's one approach to get as much as possible from the data:   proc sort data=have; by company date; run;   data want; set have; by company; opening = lag(closing); if first.company then opening=.; run; ... View more

can you please write down the sample code ... View more

I developed a code to do the work - see ttached file. The code generates a program in a directory as defined in first line - FILENAME statement. Adapt the filename to your instalation. To run the generated program add at end the line: %include pgm;   For test I looped from FEB 2014 to FEB 2015; Addapt required dates on lines: 4, 5, 13. It can be done dynamicaly if you put the code into a macro:   %macro create(y_start, m_start, y_end, m_end);      /* copy the code here and change lines: */     /* line 4 */  y1 = &y_start;     /* line 5 */ m1 = &m_start;     /* line 13 */  do until(y1 = &y_end and m1 = &m_end); %mend create; %create (2014, 02, 2016, 09); %include pgm;   ... View more

IFN and IFC are often the best way to avoid the most commong pitfalls of lags:   data want; set have; new_smo_flags=ifc(policy_no=lag(policy_no) and cust_id=lag(cust_id),lag(smo_flag),smo_flag); run;   the benefit of ifn (for numeric results) and ifc (for char results) is tha both possible options are evaluated (i.e.the lag queue  is updated) no matter which option is returned.  These functions are syntactically like the IF function in excel.   regards, Mark; ... View more

data want; set have;  by cust_id;       retain flag1; drop flag1;      if first.cust_id or flag = 'Y' then flag_1 = flag;      else flag = flag1; run; ... View more