Replacement for Union in Direct Query

[Anonymous]

Hello everyone,

If I have this kind of table:

It is confirmed, all rows will have same value on each columns.

Initially I want to Transpose this data To be come simply like this :

With an addition of 1 column, so probably like this :

The ABC in column [Field] is actually can be the column name of 1st table, I just want to make it easier.

Is this possible with Direct Query Mode ?

At first I thought a UNION like this :

UNION( 			
SELECTCOLUMNS('Table', "Type", "A", "Description", 'Table'[Column1]),			
SELECTCOLUMNS('Table', "Type", "B", "Description", 'Table'[Column2]),			
SELECTCOLUMNS('Table', "Type", "C", "Description", 'Table'[Column3]),			

But it turns out, this UNION forced me to switch to Import Mode. Any replacement trick to make this possible in Direct Query ?

Thanks

[v-yalanwu-msft · Accepted solution]

Hi @Anonymous  , 

According to the official document , Calculated tables aren't supported:

So,  you could create the A, B, C three columns  as three measures, and then use “union() “to achieve , as follows:

First: create three measures

Ameasure = MAX('Table'[A])

BMeasure = MAX('Table'[B])

Cmeasure = MAX('Table'[C])

Then create new table:

newTable =
DISTINCT (
    UNION (
        SELECTCOLUMNS ( 'Table', "Type", "A", "Description", [Ameasure] ),
        SELECTCOLUMNS ( 'Table', "Type", "B", "Description", [BMeasure] ),
        SELECTCOLUMNS ( 'Table', "Type", "C", "Description", [Cmeasure] )
    )
)

The final output is shown below:

 

Last, If we create a virtual table with union(), the storage mode will change to mixed mode.

Best Regards,
Community Support Team_ Yalan Wu
If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.  

[v-yalanwu-msft · Accepted solution]

Hi @Anonymous  , 

According to the official document , Calculated tables aren't supported:

So,  you could create the A, B, C three columns  as three measures, and then use “union() “to achieve , as follows:

First: create three measures

Ameasure = MAX('Table'[A])

BMeasure = MAX('Table'[B])

Cmeasure = MAX('Table'[C])

Then create new table:

newTable =
DISTINCT (
    UNION (
        SELECTCOLUMNS ( 'Table', "Type", "A", "Description", [Ameasure] ),
        SELECTCOLUMNS ( 'Table', "Type", "B", "Description", [BMeasure] ),
        SELECTCOLUMNS ( 'Table', "Type", "C", "Description", [Cmeasure] )
    )
)

The final output is shown below:

 

Last, If we create a virtual table with union(), the storage mode will change to mixed mode.

Best Regards,
Community Support Team_ Yalan Wu
If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.  

[Anonymous]

Hi,

Thanks, but unfortunately I must use pure DIrect Query.

rgds,

[v-yalanwu-msft · Accepted solution]

Hi, @Anonymous 

You could see my pbix file. Your data source is still in DQ mode. Since you created a virtual table with union(), it is mix(), which does not affect your data source itself.

Best Regards,
Community Support Team_ Yalan Wu
If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.  

[amitchandak]

@Anonymous , As far as I understood this transformation is only possible in import mode

[Anonymous]

So, no TRANSPOSE.. and also no UNION...

No other tricks ? 😅

[amitchandak]

@Anonymous , I was think of usin userealtionship with a table having distinct names (that can be as import mode table if needed)

But that depend of requirements  https://radacad.com/userelationship-or-role-playing-dimension-dealing-with-inactive-relationships-in-power-bi

In this manner we can take total from three columns

also run time union in a measure, depend on need