SAS finding overlapping dates and add a flag - date

Question

I have a dataset at id level with some overlapping dates. All I need to find those rows and add an identifier to count the number overlapping records.
Data:
ID ITEM StrDate EndDate
1001 A121 02/01/2022 02/15/2022
1001 B121 03/10/2022 03/10/2022
1002 C121 02/01/2022 02/10/2022
1002 D121 02/05/2022 02/15/2022
1003 E121 03/10/2022 03/21/2022
1003 F121 03/12/2022 03/21/2022
1004 G121 01/12/2022 01/14/2022
Below is the Result that I am expecting
Want:
ID ITEM StrDate EndDate Indicator
1001 A121 02/01/2022 02/15/2022 N
1001 B121 03/10/2022 03/10/2022 N
1002 C121 02/01/2022 02/10/2022 Y
1002 D121 02/05/2022 02/15/2022 Y
1003 E121 03/10/2022 03/21/2022 Y
1003 F121 03/12/2022 03/21/2022 Y
1004 G121 01/12/2022 01/14/2022 N
I tried sorting the data first on StrDate and EndDate
Proc sort data=Data; by ID StrDate EndDate;run;
Then I tried using lag function to find the same id and subtract the dates but I figured that's not the correct way of doing.
I appreciate your help here. thanks

Answer 1

SAS Date values are integers that can be used as an index into a tracking array. This technique is called a direct-index search.
Example:
A double DOW solution can be coded to find the overlapping records. The first loop flags dates in use and the second loop evaluates the range for an overlap by finding a flag via direct-index.
data have;
input ID ITEM $ StrDate EndDate;
attrib strdate enddate format=mmddyy10. informat=mmddyy10.;
datalines;
1001 A121 02/01/2022 02/15/2022
1001 B121 03/10/2022 03/10/2022
1002 C121 02/01/2022 02/10/2022
1002 D121 02/05/2022 02/15/2022
1003 E121 03/10/2022 03/21/2022
1003 F121 03/12/2022 03/21/2022
1004 G121 01/12/2022 01/14/2022
;
data want;
array tracker(100000) _temporary_ ;
do _n_ = 1 by 1 until (last.id);
set have;
by id;
do _i_ = strdate to enddate;
tracker(_i_) + 1; /* flag date using direct-index */
end;
end;
do _n_ = 1 to _n_;
set have;
/* no overlap would mean no dates in range would find a flag set */
/* and loop would exit with _i_ > enddate */
do _i_ = strdate to enddate while (tracker(_i_) = 1);
end;
length overlap_indicator $1;
overlap_indicator = ifc (_i_ > enddate, 'N', 'Y');
output;
end;
call missing (of tracker(*));
drop _: ;
run;

Answer 2

Extend, count and remerge, this is my thought.
*An extra observation added to ID 1002;
data have;
input ID $ ITEM $ StrDate mmddyy10. +1 EndDate mmddyy10.;
format StrDate EndDate mmddyy10.;
cards;
1001 A121 02/01/2022 02/15/2022
1001 B121 03/10/2022 03/10/2022
1002 C121 02/01/2022 02/10/2022
1002 D121 02/05/2022 02/15/2022
1002 D121 03/05/2022 03/15/2022
1003 E121 03/10/2022 03/21/2022
1003 F121 03/12/2022 03/21/2022
1004 G121 01/12/2022 01/14/2022
;
run;
*Extend;
data middle;
set have;
do date=StrDate to EndDate;
output;
end;
run;
*Count and remerge;
proc sql noprint;
create table want as
select distinct a.*, ifc(b.count and a.StrDate<=b.date<=a.EndDate,'Y','N') as Indicator
from have as a
left join (
select id, date, count(date) as count from middle
group by id, date
having count>1
) as b on a.id=b.id
;
quit;
By the way, if not all records overlapping in dates of one ID but you want to flag all of them out, you need to modify table lookup condition by removing the a.StrDate<=b.date<=a.EndDate.

Answer 3

Simple overlap logic:
proc sql;
create table want as
select
a.*,
/* simple overlap logic */
case
when a.strdate <= b.strdate & a.enddate >= b.strdate then 'Y'
when b.strdate < a.strdate & b.enddate >= a.strdate then 'Y'
else 'N'
end as overlap
from
have a
left join
have b
on a.id = b.id /* join on same ids */
and a.item <> b.item /* but not the same item */
;
quit;
Result:
ID ITEM StrDate EndDate overlap
1001 B121 03/10/2022 03/10/2022 N
1001 A121 02/01/2022 02/15/2022 N
1002 D121 02/05/2022 02/15/2022 Y
1002 C121 02/01/2022 02/10/2022 Y
1003 E121 03/10/2022 03/21/2022 Y
1003 F121 03/12/2022 03/21/2022 Y
1004 G121 01/12/2022 01/14/2022 N
Overlap occurs if StartA <= StartB when:
StartA EndA>=StartB
|-------------|
|---------
StartB