Hi @soumri  It's not elegant, but it works (as we've discussed over email).  This is more step-by-step than I would normally do it but I did it this way intentionally so that you (and others) would understand what I was doing.  Looking forward to seeing your paper!  Please reach out should you need anything further. Chris   /* Code written and tested on SAS University Edition, running on a MacBook Pro and VirtualBox. SAS UE is run on Chrome. Requirements: 1) A minimum of 4 and a maximum of 11 observations for each identifier in a given rank (already verified) AND 2) Each group (GRP) for a given YTD must contain at least 5 different IDs. AND 3) Each class of GYS must contain at least 4 observations. Description of the code: 1 - 3: Take the requirements above and create specific temporary tables with the data that meet the criteria 4: Take the temp table that I have created in 1 and create a new table where the IDs from Table1 match those in our original table 5: Take the new table created in 4 and match it with the original data by GRP and YTD 6: Take the new table created in 5 and match it with the original data by ID Results: This ends up with a dataset (n = 525) that match all three of your criteria. Addition: Step 7 - Remerged the 525 observations back with the original dataset and create a flag to indicate a match. Notes: Change "work.import" to whatever library.table_name you have. Once you've made that change, the remainder of the code should work. I've tested this three times and get the same results. */ /* Import code %web_drop_table(WORK.IMPORT); FILENAME REFFILE '/folders/myfolders/dataset.csv'; PROC IMPORT DATAFILE=REFFILE DBMS=CSV OUT=WORK.IMPORT; GETNAMES=YES; RUN; PROC CONTENTS DATA=WORK.IMPORT; RUN; %web_open_table(WORK.IMPORT); */ proc sql; /* 1 */ create table work.table1 as select ID, count(*) as Cnt from work.import group by ID having count(*)>=4 and count(*)<=11 order by ID; /* 2 */ create table work.table2a as select distinct grp, ytd, id from work.import; create table work.table2 as select grp, ytd, count(*) as Cnt from work.table2a group by grp, ytd having count(*)>=5 order by grp, ytd; /* 3 */ create table work.table3 as select gys, count(*) as Cnt from work.import group by gys having count(*)>=4 order by gys; /* 4 */ create table work.table5 as select a.* from work.import a, work.table3 b where a.gys = b.gys; /* 5 */ create table work.table6 as select a.* from work.import a, work.table2 b where a.grp = b.grp and a.ytd = b.ytd; /* 6 */ create table work.table7 as select a.* from work.import a, work.table1 b where a.id = b.id; quit; /* 7 */ proc sql; create table work.final as select a.*, case when b.id is not null then '1' else '0' end as Match from work.import a left join work.table7 b /* LEFT Join = ALL records from the original table and the records that match from TABLE7 */ on a.grp = b.grp and a.gys = b.gys and a.id = b.id; select * from work.final order by Match desc; quit;   ... View more