Hi,   Thank you very much. It worked and helped me a lot!   Thanks,   Moraes ... View more

It is weird that I got nothing in log.  or Try this one :   data have; input ID $ (Start End) (:yymmdd10.); format Start End yymmdd10.; cards; 1 20190519 20190601 2 20190306 20190324 3 20190822 20190830 ; data have2; input Date :yymmdd10.; format Date yymmdd10.; cards; 20190522 20190528 20190615 20190820 20190823 20190827 ; data want; if _n_=1 then do; if 0 then set have2; declare hash h(dataset:'have2'); h.definekey('date'); h.definedone(); end; set have; count=0; do i=start to end; if h.check(key:i)=0 then count+1; end; drop i date; run; ... View more

I suspect you want to know more than just A and I here? ... View more

Sounds like you just want to collapse overlapping (or close to overlapping) periods into one period.  Not sure why you want two datasets out when one dataset can contain all of the periods with just a separate variable to indicate which period it is. So first make sure your dates are actual date variables (and not strings like you example). data have; input SUBJECT (START_DATE END_DATE) (:date.) seq; format start_date end_date date9.; cards; 3008 01OCT2020 01OCT2020 1 3008 02OCT2020 04OCT2020 2 3008 05OCT2020 06OCT2020 3 3008 08OCT2020 10OCT2020 4 3009 03OCT2020 03OCT2020 1 3009 04OCT2020 05OCT2020 2 3009 05OCT2020 05OCT2020 3 3009 06OCT2020 07OCT2020 4 3010 13OCT2020 15OCT2020 1 3010 17OCT2020 18OCT2020 2 ; Then generate a PERIOD variable to flag the overlapping periods.  Adjust the IF condition if you want to allow different size gap between start and end to be considered part of the same large period. Including START_DATE and END_DATE in the BY statement will cause the data step to insure the observations are sorted properly.   data periods ; set have ; by subject start_date end_date ; lag_end = lag(end_date); if first.subject then period=1; else if start_date > lag_end+1 then period+1; drop lag_end; run; Now collapse to one observation per derived period. data want; set periods ; by subject period; if first.period then start=start_date ; if last.period then do; start_date=start; output; end; retain start; drop start seq; run; Results: START_ Obs SUBJECT DATE END_DATE period 1 3008 01OCT2020 06OCT2020 1 2 3008 08OCT2020 10OCT2020 2 3 3009 03OCT2020 07OCT2020 1 4 3010 13OCT2020 15OCT2020 1 5 3010 17OCT2020 18OCT2020 2   Note if the original start/end date data is messier with overlapping or nested time periods you will need more complex logic to determine the overlapping periods. ... View more

Just for fun, version with arrays: data have; input LOC $ 3. x1 x2 develop; x3 = 2 * x1; x4 = 2 * x2; cards; AUS 10 20 1 AUT 15 25 1 BEL 8 21 1 ARG 3 5 0 BGR 5 6 0 BRA 1 4 0 ; run; proc print; run; %let NoX= 4; data want; length LOC $ 12; set have; by descending develop; array X(_i_) x1-x&NoX.; array S(_i_) _sumX1-_sumX&NoX.; array C(_i_) _countX1-_countX&NoX.; if first.develop then do over S; S = 0; C = 0; end; do over X; S + X; C + (X > .z); end; output; if last.develop then do; LOC = catx(" ", "Average", develop); do over X; X = S / C; end; output; end; drop develop _:; run; proc print; run;   B-) ... View more

Look closely at codes posted. Your code misses the explicit OUTPUT statement in the DO loop, so you only have one implicit output at the end of the data step iteration, outputting the last record read from a file. ... View more