Please try the below code   data have; input id date:ddmmyy8. event&:$20.; format date date9.; cards; 1 01012000 Kindergarten 1 01012001 High school 1 01012003 Primary 1 01012010 High School 2 01012000 University 2 01012010 High School 3 01012009 Kindergarten 3 01012010 Preschool 3 01012012 Kindergarten 3 01012019 High School 3 01012022 Kindergarten 3 01012024 University ; proc format; invalue sch 'Kindergarten'=1 'High school'=3 'High School'=3 'Primary'=2 'University'=4 'Preschool'=1 ; proc sort data=have; by id date; run; data have; set have; schnum=input(event,sch.); by id date; run; proc sort data=have; by id schnum; run; data want; set have; by id schnum; retain schnum2 date2; if first.id then schnum2=schnum; if first.id then date2=date; format date2 date9.; if schnum>schnum2 and date<date2 then delete; run; ... View more

Hi @Bounce    If you have overlapping ranges, you can try the following code, which will retrieve the min(start_date) and the max(end_date) if the lag is strictly more than 1. PS: I have deliberately modified the entry data to show different cases.   My best,   data have; input ID Start_Date End_Date; informat Start_Date End_Date ddmmyy10.; format Start_Date End_Date date9.; cards; 1 01/01/2019 04/01/2019 1 07/01/2019 12/01/2019 1 10/01/2019 18/01/2019 1 14/01/2019 16/01/2019 1 16/01/2019 22/01/2019 2 01/01/2019 04/01/2019 2 07/01/2019 12/01/2019 2 10/01/2019 18/01/2019 ; run; proc sort data=have; by ID Start_Date End_Date; run; data have2; set have; format _lag date9.; by ID; _lag = lag(End_Date) ; if first.ID then do; _lag = .; flag = 0; end; if _lag + 1 < Start_Date then flag + 1; run; proc sql; create table want as select ID, min(Start_Date) as Start_date format=date9., max(End_date) as End_date format=date9. from have2 group by ID, flag; quit; ... View more

There is a one-step solution, which  benefits from the CALL SORTC subroutine:   DATA Example; format ID 1. VAR1 VAR2 VAR3 $1.; input ID VAR1 VAR2 VAR3; datalines; 1 A A A 2 A B C 3 A A A 4 A B B run; data want (keep=id var); set example; array _v {*} dummy var1-var3; call sortc(of _v{*}); do i=2 to dim(_v); if missing(_v{i}) then continue; var=_v{i}; if var^=_v{i-1} then output; end; run;   The "if missing(_v{i}) then continue statement skips to the next iteration of the do loop, skipping the later output stateement. I put  a dummy variable at the lower bound  of the array, to smiplify the  "if var^=_v{i-1}" test. ... View more

data have; input Dog $ Meal $ 3-9 Day; datalines; A Fish 1 A Fish 2 A Chicken 2 A Lamb 3 B Fish 1 B Chicken 2 B Chicken 3 B Chicken 4 B Steak 5 B Fish 5 B Chicken 5 C Fish 1 C Steak 2 C Steak 3 C Steak 4 C Fish 4 C Lamb 5 C Steak 6 ; data want; if _n_=1 then do; dcl hash h(); h.defineKey ("Dog", "Meal"); h.defineData ("X"); h.defineDone (); end; set have; if h.find() ne 0 then X = 1; else X + 1; h.replace(); run; Result:     Dog Meal Day X A Fish 1 1 A Fish 2 2 A Chicken 2 1 A Lamb 3 1 B Fish 1 1 B Chicken 2 1 B Chicken 3 2 B Chicken 4 3 B Steak 5 1 B Fish 5 2 B Chicken 5 4 C Fish 1 1 C Steak 2 1 C Steak 3 2 C Steak 4 3 C Fish 4 2 C Lamb 5 1 C Steak 6 4 ... View more