基于列2 SQL Server的列1的不同值

Question

I have a table with columns ID and Val . For each value of ID we can have either same or different values of Val .

ID  Val
1   A
1   NULL
2   00
2   00
2   00
2   00
3   00
3   A
4   A
5   00
5   00
5   A
6   A
6   A
6   NULL
6   00

From above table, I am looking for IDs which has different values in Val column. If for any given ID all values of Val column are same then it should not come in result.

So result would be something like.

D   Val
1   A
1   NULL
3   00
3   A
5   00
5   00
5   A
6   A
6   A
6   NULL
6   00

Id 2 should not come in result because for Id 2, Val column has same data. Similarly ID 4 will not come in result as ID 4 has only one row.

For each ID if we have more than one value in Val column then is it should show in result.

Thanks for the Help!

Answer 1

For the ids that meet the condition of having different values:

select id
from t
group by id
having min(id) <> max(id);

You can then incorporate this into a query as:

select t.*
from t join
     (select id
      from t
      group by id
      having min(id) <> max(id)
     ) tt
     on t.id = tt.id;

Or, you can use window functions:

select t.id, t.val
from (select t.*,
             min(val) over (partition by id) as minval,
             max(val) over (partition by id) as maxval
      from t
     ) t
where minval <> maxval;

Answer 2

Try this:

SELECT ID, Val
FROM mytable 
WHERE ID IN (SELECT ID
             FROM mytable
             GROUP BY ID
             HAVING COUNT(DISTINCT CASE 
                                     WHEN Val IS NULL THEN '' 
                                     ELSE Val
                                   END) > 1

I've made the assumption that Val field is of type VARCHAR and that it can be either NULL or <> '' .

Answer 3

Build your query at three steps:

1. Select distinct values id, val (to ensure you will get the null safe count)
2. Count distinct values for each id
3. Show results from source table

Try to use inner select instead of subselect to speed up the query.

The solution is written in the query below:

SELECT
    t.*
FROM
    -- select only ids with distinct count > 1
    (
        SELECT
            id
        FROM
            --  select distinct values to ensure your count of null values is real
            (
                SELECT DISTINCT
                    id, val
                FROM
                    t
            ) AS td
        GROUP BY
            id
        HAVING
            COUNT(*) > 1
    ) AS tc
-- join the source table
INNER JOIN
    t
ON
    t.id = tc.id