creating a time window for event occurences

[jins]

Trying to get this manipulation going and any thoughts/help on this is appreciated.

Dataset Sample

id    status   date

1     OK           02-MAY-09

1    OK          10-MAY-09

1    DEL        01-JUL-22

1    OK          01-DEC-22

1    OK          04-DEC-22

1    DEL        06-DEC-22

2   DEL        01-JAN-23

3   OK          03-JAN-23

4   DEL        04-JAN-23

4  DEL         05-JAN-23

4  OK         06-JAN23

Looking for the following output which can be long or wide format

1     OK          02-MAY-09

1    DEL        01-JUL-22

1    OK          01-DEC-22

1    DEL        06-DEC-22

2   DEL        01-JAN-23

3   OK          03-JAN-23

4   DEL        04-JAN-23

4  OK          06-JAN23

Thanks in advance.

[SASJedi]

Can you explain the logic for choosing the rows to output?

[jins]

I should have mentioned that oops....

To obtain the date of the first occurrence between the status change by ID. For example ID-1 has two OK status before the DEL status, so we keep the first one with date 02-May-2009. For ID's with no status changes, it is output as is (ID-2).

[jins]

Is the notsorted option in the by variable a good idea to filter them out if the data is already sorted prior by id, date and status?

data check;

by id status notsorted;

if first.status=1;

run;

[Patrick · Accepted solution]

Same selection logic as already suggested by @jins 

data have;
  infile datalines truncover;
  input id status $ dt:date11.;
  format dt date9.;
  datalines;
1 OK 02-MAY-09
1 OK 10-MAY-09
1 DEL 01-JUL-22
1 OK 01-DEC-22
1 OK 04-DEC-22
1 DEL 06-DEC-22
2 DEL 01-JAN-23
3 OK 03-JAN-23
4 DEL 04-JAN-23
4 DEL 05-JAN-23
4 OK 06-JAN23
;

/* ensure source data is sorted as required */
proc sort data=have presorted;
  by id dt;
run;

data want;
  set have;
  by id status notsorted;
  if first.status;
run;

proc print data=want;
run;

[jins]

thanks.