Max function help

Question

I am trying to pull 1 row for each document_id, it should be the max version but im getting duplicate rows for some reason.
create table max_test
"PROJECT" VARCHAR2(50 BYTE),
"DOCUMENT_ID" VARCHAR2(50 BYTE),
"VERSION" VARCHAR2(5 BYTE),
"CREATE_DATE" VARCHAR2(DATE),
"UPDATE_DATE" VARCHAR2(DATE),
"MANAGER_NAME" VARCAHR2(100 BYTE)
INSERT ALL
INTO max_test (PROJECT,DOCUMENT_ID, VERSION, CREATE_DATE, UDPATE_DATE,MANAGER_NAME) VALUES (500,1, 18, '01-MAR-12','01-MAR-12','BOB JONES')
INTO max_test (PROJECT,DOCUMENT_ID, VERSION, CREATE_DATE, UDPATE_DATE,MANAGER_NAME) VALUES (500,2, 6, '01-MAR-12','01-MAR-12','STEVE JONES')
INTO max_test (PROJECT,DOCUMENT_ID, VERSION, CREATE_DATE, UDPATE_DATE,MANAGER_NAME) VALUES (500,3, 2, '01-MAR-12','01-MAR-12','MIKE JONES')
INTO max_test (PROJECT,DOCUMENT_ID, VERSION, CREATE_DATE, UDPATE_DATE,MANAGER_NAME) VALUES (501,4, 5, '01-MAR-12','01-MAR-12','MARK JONES')
INTO max_test (PROJECT,DOCUMENT_ID, VERSION, CREATE_DATE, UDPATE_DATE,MANAGER_NAME) VALUES (501,5, 1, '01-MAR-12','01-MAR-12','TOM JONES')
INTO max_test (PROJECT,DOCUMENT_ID, VERSION, CREATE_DATE, UDPATE_DATE,MANAGER_NAME) VALUES (502,6, 20, '01-MAR-12','01-MAR-12','EVE JONES')
INTO max_test (PROJECT,DOCUMENT_ID, VERSION, CREATE_DATE, UDPATE_DATE,MANAGER_NAME) VALUES (502,7, 3, '01-MAR-12','01-MAR-12','CHRIS JONES')
INTO max_test (PROJECT,DOCUMENT_ID, VERSION, CREATE_DATE, UDPATE_DATE,MANAGER_NAME) VALUES (503,8, 1, '01-MAR-12','01-MAR-12','LARRY JONES')
INTO max_test (PROJECT,DOCUMENT_ID, VERSION, CREATE_DATE, UDPATE_DATE,MANAGER_NAME) VALUES (503,9, 2, '01-MAR-12','01-MAR-12','JOE JONES')
select distinct(project),max(version),document_id,create_date,update_date,manager_name
from max_test
group by project,document_id,create_date,update_date,manager_name
I get multiple rows for each project and only want 1 row per project, the 1 row that has the max version

Answer 1

Hi Jay,
first of all when you post statements I suggest you to test them before posting.
Your statements had some syntax error.
Additionally when you post your code put it between two lines starting with {noformat}{noformat} to have your code properly formatted.
I have modified your statements in the way I suppose they should be:CREATE TABLE max_test
project VARCHAR2 (50 BYTE)
, document_id VARCHAR2 (50 BYTE)
, version VARCHAR2 (5 BYTE)
, create_date DATE
, update_date DATE
, manager_name VARCHAR2 (100 BYTE)
INSERT INTO max_test (PROJECT,DOCUMENT_ID, VERSION, CREATE_DATE, UPDATE_DATE,MANAGER_NAME) VALUES (500,1, 18, TO_DATE('01-MAR-12','DD-MON-YYYY'), TO_DATE('01-MAR-12','DD-MON-YYYY'),'BOB JONES');
INSERT INTO max_test (PROJECT,DOCUMENT_ID, VERSION, CREATE_DATE, UPDATE_DATE,MANAGER_NAME) VALUES (500,2, 6, TO_DATE('01-MAR-12','DD-MON-YYYY'), TO_DATE('01-MAR-12','DD-MON-YYYY'),'STEVE JONES');
INSERT INTO max_test (PROJECT,DOCUMENT_ID, VERSION, CREATE_DATE, UPDATE_DATE,MANAGER_NAME) VALUES (500,3, 2, TO_DATE('01-MAR-12','DD-MON-YYYY'), TO_DATE('01-MAR-12','DD-MON-YYYY'),'MIKE JONES');
INSERT INTO max_test (PROJECT,DOCUMENT_ID, VERSION, CREATE_DATE, UPDATE_DATE,MANAGER_NAME) VALUES (501,4, 5, TO_DATE('01-MAR-12','DD-MON-YYYY'), TO_DATE('01-MAR-12','DD-MON-YYYY'),'MARK JONES');
INSERT INTO max_test (PROJECT,DOCUMENT_ID, VERSION, CREATE_DATE, UPDATE_DATE,MANAGER_NAME) VALUES (501,5, 1, TO_DATE('01-MAR-12','DD-MON-YYYY'), TO_DATE('01-MAR-12','DD-MON-YYYY'),'TOM JONES');
INSERT INTO max_test (PROJECT,DOCUMENT_ID, VERSION, CREATE_DATE, UPDATE_DATE,MANAGER_NAME) VALUES (502,6, 20, TO_DATE('01-MAR-12','DD-MON-YYYY'), TO_DATE('01-MAR-12','DD-MON-YYYY'),'EVE JONES');
INSERT INTO max_test (PROJECT,DOCUMENT_ID, VERSION, CREATE_DATE, UPDATE_DATE,MANAGER_NAME) VALUES (502,7, 3, TO_DATE('01-MAR-12','DD-MON-YYYY'), TO_DATE('01-MAR-12','DD-MON-YYYY'),'CHRIS JONES');
INSERT INTO max_test (PROJECT,DOCUMENT_ID, VERSION, CREATE_DATE, UPDATE_DATE,MANAGER_NAME) VALUES (503,8, 1, TO_DATE('01-MAR-12','DD-MON-YYYY'), TO_DATE('01-MAR-12','DD-MON-YYYY'),'LARRY JONES');
INSERT INTO max_test (PROJECT,DOCUMENT_ID, VERSION, CREATE_DATE, UPDATE_DATE,MANAGER_NAME) VALUES (503,9, 2, TO_DATE('01-MAR-12','DD-MON-YYYY'), TO_DATE('01-MAR-12','DD-MON-YYYY'),'JOE JONES');
Now, coming to your problem, it looks that you have no idea of how the group by clause and distinct keywords are affecting your query.
Using DISTINCT before the list of your column will ensure that the output does not have duplicate rows (having the same value). Note that distinct does not affect a single column but all the columns of the query. So writingselect distinct(project),max(version),document_id,create_date,update_date,manager_name
ORselect distinct project,max(version),document_id,create_date,update_date,manager_name
has no difference.
Group by is listing the columns you want to use to group your output.
I.e.:SELECT project, MAX (version) version
FROM max_test
GROUP BY project;
PROJECT VERSION
503 2
500 6
501 5
502 3
If you add more colums in your group by then the grouping functions (MAX, MIN, AVG, SUM etc.) will be applied to groups represented by unique combination of all columns in your GROUP BY clause.
In your case your are grouping by  project,document_id,create_date,update_date,manager_name
So you will get the maximum version of each combination of project,document_id,create_date,update_date,manager_name. If document_id, create_date, update_date and manager_name are different for a specific project, then you will have several rows for the same project.SELECT DISTINCT (project), MAX (version), document_id, create_date
, update_date, manager_name
FROM max_test
GROUP BY project, document_id, create_date, update_date
, manager_name;
PROJECT MAX(VERSION) DOCUMENT_ID CREATE_DATE UPDATE_DATE MANAGER_NAME
503 1 8 01-MAR-12 01-MAR-12 LARRY JONES
503 2 9 01-MAR-12 01-MAR-12 JOE JONES
500 18 1 01-MAR-12 01-MAR-12 BOB JONES
502 20 6 01-MAR-12 01-MAR-12 EVE JONES
502 3 7 01-MAR-12 01-MAR-12 CHRIS JONES
500 6 2 01-MAR-12 01-MAR-12 STEVE JONES
501 1 5 01-MAR-12 01-MAR-12 TOM JONES
501 5 4 01-MAR-12 01-MAR-12 MARK JONES
500 2 3 01-MAR-12 01-MAR-12 MIKE JONES
I hope it is clear.
Regards.
Al