Does vertica support derivative column concept in case statement.

hdk0310

HI,
Please help me out as I am stuck with a Sql Case statement in vertica.

I have a a data set where
ID | rank | Columnname
1 | 1 | A
1 | 2 | B
1 | 3 | C
1 | 4 | D
My Desired Output :
ID | rank | Columnname | Aggregate
1 | 1 | A | A
1 | 2 | B | A,B
1 | 3 | C | A,B,C
1 | 4 | D | A,B,C,D

The aggregate column should concatenate the columnname values based on the rank.
For example For rank =2 Aggregate value should be Columnname in rank 1+ Columnname in Seq 2.
For rank = 3 Aggregate value should be Columnname in rank 1+ Columnname in rank 2 (Or derivative obtain from last statement )+ Columnname in rank 2. and so on

I have written a CASE statement to handle this . But This is not giving the result as expected since for every rank it is re-evaluating the case instead of taking result from first iteration.

SELECT
ID,rank,columnname
(CASE WHEN rank = 1 THEN ColumnName END) ||
(CASE WHEN rank = 2 THEN ', ' || ColumnName ELSE '' END) ||
(CASE WHEN rank = 3 THEN ', ' || ColumnName ELSE '' END) AS Aggregate
FROM Table;

The problem with this is from rank =2 I GET NULL VALUES FOR AGGREGATE.
NOTE= Rank can go upto 25.

Please let me know if anyone has a solution to this.

Thanks

Jim_Knicely

What version of Vertica are you running?

It's relatively easy to accomplish your goal using the LISTAGG function which is available in Vertica 9.1.1-4 and above.

Example:

dbadmin=> SELECT * FROM tab1;
 id | rank | columnname
----+------+------------
  1 |    1 | A
  1 |    2 | B
  1 |    3 | C
  1 |    4 | D
(4 rows)

dbadmin=> SELECT a.id, a.rank, a.columnname, listagg(b.columnname)
dbadmin->  FROM tab1 a
dbadmin->  JOIN tab1 b
dbadmin->    ON a.id = b.id
dbadmin->   AND b.rank <= a.rank
dbadmin->  GROUP BY a.id, a.rank, a.columnname;
 id | rank | columnname | listagg
----+------+------------+---------
  1 |    1 | A          | A
  1 |    2 | B          | A,B
  1 |    3 | C          | A,B,C
  1 |    4 | D          | A,B,C,D
(4 rows)

See:
https://www.vertica.com/docs/9.2.x/HTML/Content/Authoring/SQLReferenceManual/Functions/Aggregate/LISTAGG.htm

Jim_Knicely

Another option is to use the LAG analytic function:

dbadmin=> SELECT * FROM tab1;
 id | rank | columnname
----+------+------------
  1 |    1 | A
  1 |    2 | B
  1 |    3 | C
  1 |    4 | D
  2 |    1 | X
  2 |    2 | Y
  2 |    3 | Z
(7 rows)

dbadmin=> SELECT id, rank, columnname,
dbadmin->        CASE
dbadmin->        WHEN rank = 1
dbadmin->          THEN columnName
dbadmin->        WHEN rank = 2
dbadmin->          THEN LAG(columnName) OVER (PARTITION BY id ORDER BY rank) || ',' || columnName
dbadmin->        WHEN rank = 3
dbadmin->          THEN LAG(columnName, 2) OVER (PARTITION BY id ORDER BY rank) || ',' ||
dbadmin->            LAG(columnName) OVER (PARTITION BY id ORDER BY rank) || ',' || columnName
dbadmin->        WHEN rank = 4
dbadmin->          THEN LAG(columnName, 3) OVER (PARTITION BY id ORDER BY rank) || ',' ||
dbadmin->            LAG(columnName, 2) OVER (PARTITION BY id ORDER BY rank) || ',' || LAG(columnName) OVER (PARTITION BY id ORDER BY rank) || ',' || columnName
dbadmin->        END as aggregate
dbadmin->   FROM tab1
dbadmin->   ORDER BY id, rank;
 id | rank | columnname | aggregate
----+------+------------+-----------
  1 |    1 | A          | A
  1 |    2 | B          | A,B
  1 |    3 | C          | A,B,C
  1 |    4 | D          | A,B,C,D
  2 |    1 | X          | X
  2 |    2 | Y          | X,Y
  2 |    3 | Z          | X,Y,Z
(7 rows)

hdk0310

Thanks Jim
I will try both the approaches .