SQL Rank() function excluding rows - sql

Question

Consider I have the following table.
ID value
1 100
2 200
3 200
5 250
6 1
I have the following query which gives the result as follows. I want to exclude the value 200 from rank function, but still that row has to be returned.
SELECT
CASE WHEN Value = 200 THEN 0
ELSE DENSE_RANK() OVER ( ORDER BY VALUE DESC)
END AS RANK,
ID,
VALUE
FROM #table
RANK ID VALUE
1 5 250
0 2 200
0 3 200
4 1 100
5 6 1
But I want the result as follows. How to achieve it?
RANK ID VALUE
1 5 250
0 2 200
0 3 200
2 1 100
3 6 1

Answer 1

If VAL column is not nullable, taking into account NULL is the last value in ORDER BY .. DESC
select *, dense_rank() over (order by nullif(val,200) desc) * case val when 200 then 0 else 1 end
from myTable
order by val desc;

Answer 2

There is no way to exclude Val in Dense Rank currently ,unless you filter in where clause..that is the reason ,you get below result
RANK ID VALUE
1 5 250
0 2 200
0 3 200
4 1 100
5 6 1
You will need to filter once and then do a union all
;with cte(id,val)
as
(
select 1, 100 union all
select 2, 200 union all
select 3, 200 union all
select 5, 250 union all
select 6, 1 )
select *, dense_rank() over (order by val desc)
from cte
where val<>200
union all
select 0,id,val from cte where val=200

Answer 3

You could split the ranking in to separate queries for the values you want to include/exclude from the ranking and UNION ALL the results like so:
Standalone executable example:
CREATE TABLE #temp ( [ID] INT, [value] INT );
INSERT INTO #temp
( [ID], [value] )
VALUES ( 1, 100 ),
( 2, 200 ),
( 3, 200 ),
( 5, 250 ),
( 6, 1 );
SELECT *
FROM ( SELECT 0 RANK ,
ID ,
value
FROM #temp
WHERE value = 200 -- set rank to 0 for value = 200
UNION ALL
SELECT DENSE_RANK() OVER ( ORDER BY value DESC ) AS RANK ,
ID ,
value
FROM #temp
WHERE value != 200 -- perform ranking on records != 200
) t
ORDER BY value DESC ,
t.ID
DROP TABLE #temp
Produces:
RANK ID value
1 5 250
0 2 200
0 3 200
2 1 100
3 6 1
You can modify the ordering at the end of the statement if required, I set it to produce your desired results.

Answer 4

You can also try this, too:
SELECT ISNULL(R, 0) AS Rank ,t.id ,t.value
FROM tbl1 AS t
LEFT JOIN ( SELECT id ,DENSE_RANK() OVER ( ORDER BY value DESC ) AS R
FROM dbo.tbl1 WHERE value <> 200
) AS K
ON t.id = K.id
ORDER BY t.value DESC

Answer 5

The solution in the original question was actually pretty close. Just adding a partition clause to the dense_rank can do the trick.
SELECT CASE
WHEN VALUE = 200 THEN 0
ELSE DENSE_RANK() OVER(
PARTITION BY CASE WHEN VALUE = 200 THEN 0 ELSE 1 END
ORDER BY VALUE DESC
)
END AS RANK
,ID
,VALUE
FROM #table
ORDER BY VALUE DESC;
The 'partition by' creates separate groups for the dense_rank such that the order is performed on these groups individually. This essentially means you create two ranks at the same time, one for the group without the 200 value and one for the group with only the 200 value. The latter one to be set to 0 in the 'case when'.
Standalone executable example:
DECLARE #table TABLE
(
ID INT NOT NULL PRIMARY KEY
,VALUE INT NULL
)
INSERT INTO #table
(
ID
,VALUE
)
SELECT 1, 100
UNION SELECT 2, 200
UNION SELECT 3, 200
UNION SELECT 5, 250
UNION SELECT 6, 1;
SELECT CASE
WHEN VALUE = 200 THEN 0
ELSE DENSE_RANK() OVER(
PARTITION BY CASE WHEN VALUE = 200 THEN 0 ELSE 1 END
ORDER BY VALUE DESC
)
END AS RANK
,ID
,VALUE
FROM #table
ORDER BY VALUE DESC;
RANK ID VALUE
1 5 250
0 2 200
0 3 200
2 1 100
3 6 1