Would you prefer to match under or over? Wouldn't sorting descending and using a running total until you hit the desired amount work here?
Assuming your Target is in the same data set, if it's not merge it in first and ensure your data is sorted by region and descending weight.
proc sort data=have; by region descending weight; run;
data want;
set have;
by region descending weight;
*holds value across rows;
retain running_total;
*if first region re-set values to prevent prior data from affecting calcuations;
if first.region then do;
running_total = weight;
inclusion = 1;
end;
else if inclusion = 1 then do;
running_total = running_total + weight;
if running_total > target_weight then inclusion = 0;
end;
run;
That's the brute force basic logic way. A slight improvement would be picking the value before the max and then finding the least amount after that would allow the sum to be closest to the target.
Or try PROC OPTMODEL in SAS/OR if you have the correct license, but beyond my knowledge.
https://communities.sas.com/t5/SAS-Programming/Summing-to-a-target-value/td-p/707928
