Theoretically speaking, it may be best to create a table of the top 1000 opportunity IDs first, and then select all records from your original table that have those ids. As such: data have;
input emp_name $ opp_id month subj_prob;
datalines;
John 1234567 1 20
John 1234567 2 40
John 1234567 3 60
John 324783 1 80
John 324783 2 100
Sol 1234567 4 100
Sol 489932 1 20
;
run;
PROC SQL;
CREATE TABLE TOP1K AS
SELECT OPP_ID, COUNT(*) AS TOTAL FROM HAVE GROUP BY OPP_ID;
QUIT;
PROC SQL OUTOBS=1000;
CREATE TABLE TOP1K AS
SELECT * FROM TOP1K;
QUIT;
PROC SQL;
CREATE TABLE ALLIN1K AS
SELECT * FROM HAVE
INNER JOIN TOP1K ON TOP1K.OPP_ID = HAVE.OPP_ID;
QUIT;
... View more
