Forum Discussion

Skilled Sharer

8 years ago

Solved

How to reference columns in virtual tables?

Hi, I am creating a virtual table usingVAR. How can I refer to the columns of this newly created virtual table in the same table creation DAX? New Table = VAR JointTable = NATURALLEFTOUTERJOIN(...

Need Help

OwenAuger
8 years ago
Hi again anandav

Thanks for the additional info.
From what you've provided, could I suggest a different approach, as doing the fill-down in DAX is possible but a little awkward.

I'm making an assumption that you are really interested in SeatNum and Booked Customer from your screenshot below. If so, I would propose this:

DAX Table = VAR SeatNumbersPlusBookings = GENERATEALL ( SeatNumbers, GENERATE ( SeatBookings, INTERSECT ( GENERATESERIES ( SeatBookings[Seat Start], SeatBookings[Seat End] ), { SeatNumbers[SeatNum] } ) ) ) VAR FinalTable = SELECTCOLUMNS ( SeatNumbersPlusBookings, "SeatNum", SeatNumbers[SeatNum], "Booked Customer", VAR CurrentCustomer = SeatBookings[Customer] RETURN IF ( ISBLANK ( CurrentCustomer ), "Empty", CurrentCustomer ) ) RETURN FinalTable
I'm using GENERATEALL to do the join rather than NATURALLEFTOUTERJOIN since your physical SeatBookings table don't have all the required values (since it contains Seat Start and Seat End), and GENERATE to convert the start/end values to a range.

You could of course include additional columns on top of the two output by the above.

Regards,
Owen

anandav

Skilled Sharer

8 years ago

Hi OwenAuger,

Thanks a lot for the detail reply. That is a very clear explanation.

I was trying to create a table and fill down the missing values.

Presently I have done it in two steps.

1. Create the table.

DAX Table =
VAR JointTable = NATURALLEFTOUTERJOIN(SeatNumbers,SeatBookings)
RETURN
JointTable

2. Then fill down the missing value in a new column.

Customer Fill Down =
VAR LstNoBlankCustomer =
    CALCULATE (
        LASTNONBLANK ( 'DAX Table'[SeatNum], 1 ),
        FILTER (
            ALL ( 'DAX Table' ),
            'DAX Table'[SeatNum] <= EARLIER ( 'DAX Table'[SeatNum] )
                && NOT ( ISBLANK ( 'DAX Table'[Customer] ) )
        )
    )
RETURN
    CALCULATE (
        MAX ( 'DAX Table'[Customer] ),
        FILTER ( ALL ( 'DAX Table' ), 'DAX Table'[SeatNum] = LstNoBlankCustomer )
    )

Repeat the new column step for Seat Start and Seat End.

So the final table is :

What I was trying to achieve is instead of creating the virtual table and then adding columns to it do it in a single dax create table step.

OwenAuger

Super User

8 years ago

Hi again anandav

Thanks for the additional info.

From what you've provided, could I suggest a different approach, as doing the fill-down in DAX is possible but a little awkward.

I'm making an assumption that you are really interested in SeatNum and Booked Customer from your screenshot below. If so, I would propose this:

DAX Table = 
VAR SeatNumbersPlusBookings = 
    GENERATEALL (
        SeatNumbers,
        GENERATE (
            SeatBookings,
            INTERSECT (
                GENERATESERIES ( SeatBookings[Seat Start], SeatBookings[Seat End] ),
                { SeatNumbers[SeatNum] }
            )
        )
    )
VAR FinalTable =
    SELECTCOLUMNS (
        SeatNumbersPlusBookings,
        "SeatNum", SeatNumbers[SeatNum],
        "Booked Customer",
        VAR CurrentCustomer = SeatBookings[Customer]
        RETURN
            IF ( ISBLANK ( CurrentCustomer ), "Empty", CurrentCustomer )
    )
RETURN
    FinalTable

I'm using GENERATEALL to do the join rather than NATURALLEFTOUTERJOIN since your physical SeatBookings table don't have all the required values (since it contains Seat Start and Seat End), and GENERATE to convert the start/end values to a range.

You could of course include additional columns on top of the two output by the above.

Regards,

Owen

anandav
Skilled Sharer
8 years ago
Hi OwenAuger,
Really elegant solution. My solution will not be memory efficient for large tables. Your solution is really good.
Thanks a lot for taking time to help solve this.

PS. I realised I have a lot more to learn/understand on using DAX.
anandav
Skilled Sharer
8 years ago
Hi OwenAuger,

I was trying to understand the solution but ran into some problems. I tried to seperate each part of the DAX into VARs but it gives different results compared to using all in one DAX statement.

Table =
VAR SeatsINBookedRange = GENERATESERIES ( MIN(SeatBookings[Seat Start]), MAX(SeatBookings[Seat End]) )
VAR AllSeats = SeatNumbers
VAR BookedAndEmptySeats = INTERSECT(SeatsINBookedRange, AllSeats)
VAR Test1 = GENERATE(SeatBookings, BookedAndEmptySeats)
VAR CustomerSeatBookings =
        GENERATE (
            SeatBookings,
            INTERSECT (
                GENERATESERIES ( SeatBookings[Seat Start], SeatBookings[Seat End] ),
                SeatNumbers
            )
        )
//VAR Test2 = GENERATEALL(SeatBookings, CustomerSeatBookings)
VAR SeatNumbersPlusBookings =
    GENERATEALL (
        SeatNumbers,
        GENERATE (
            SeatBookings,
            INTERSECT (
                GENERATESERIES ( SeatBookings[Seat Start], SeatBookings[Seat End] ),
                { SeatNumbers[SeatNum] }
            )
        )
    )
RETURN
SeatNumbersPlusBookings

Questions:
1.  VAR SeatsINBookedRange = GENERATESERIES ( MIN(SeatBookings[Seat Start]), MAX(SeatBookings[Seat End]) )
Thoug in the compsite DAX statement SeatBookings[Seat Start] can be referenced when in the VAR I had to use MIN and MAX functions).
Why is thsi?

2. VAR AllSeats = SeatNumbers
Couldn't create a table with {} as it is not allowed.
"A single value for column 'SeatNum' in table 'SeatNumbers' cannot be determined. This can happen when a measure formula refers to a column that contains many values without specifying an aggregation such as min, max, count, or sum to get a single result."

3.  VAR BookedAndEmptySeats = INTERSECT(SeatsINBookedRange, AllSeats)
Returns the same table as in Step#1. Value from 1 to 19

4.   VAR Test1 = GENERATE(SeatBookings, BookedAndEmptySeats)
Return wrong results which is a cartisian product of tables.

5.  VAR CustomerSeatBookings
Combination of steps 1,2,3,4 in single DAX statement gives correct esults.

6. VAR Test2 = GENERATEALL(SeatBookings, CustomerSeatBookings)
Gives an error "Function GENERATEALL does not allow two columns with the same name 'SeatBookings'[Customer]."

7. VAR SeatNumbersPlusBookings
From your solution, gives the correct results.

If I change { SeatNumbers[SeatNum] } to SeatNumbers table reference the results are wrong. What is { SeatNumbers[SeatNum] } means?
Will you be able to help explain why the compsite DAX statement works and the individual statements don't?
Thanks in advance.
- OwenAuger
  Super User
  8 years ago
  Hi again anandav
  
  Thanks for the follow-up questions!
  
  I may have to give you a more detailed answer later, but the general explanation is that the value returned by each expression is dependent on the context it is evaluated in. In particular, GENERATEALL and GENERATE take the table supplied as the first argument, and evaluate the second argument (a table expression) in the row context of each row of the first argument.
  
  When you evaluate the various expressions independently, you get strange results because they were intended to be evaluated within a particular (row) context. Also, in a row context, you can refer to the values of columns in the current row with a "naked" column reference, such as SeatBookings[Seat Start].
  
  Diagram below:
  
  Also the curly-braces syntax is a table constructor. In this case, { SeatNumbers[SeatNum] } creates a 1x1 table containing the SeatNum value from the current row of SeatNumbers. I am using this to filter the series of seat numbers created by GENERATESERIES. Looking at this again, I could just as well have used FILTER, as in something like FILTER ( GENERATESERIES(...), [Value] = SeatNumbers[SeatNum] )
  
  Let me know if that helps - I will try to get back and reply on your specific questions later though.
  
  Best regards,
  Owen
  - anandav
    Skilled Sharer
    8 years ago
    Thanks a lot for the detail explanation Owen.
    DAX gets confusing at times since some functions like clauclate we have to work from outer function to inner fucntion and others from inner to oueter (as I understand). But your diagram helps a lot!
    
    Is it ok if I use your explanation in the blog I have done with credit to you?
    I have added the solution with credit to you but also wanted to include this explanation iof it is ok with you.

Forum Discussion

How to reference columns in virtual tables?

Related content

Fabric virtual table

Reference a column in virtual table

Variable Virtual Table column reference

UDF with variable table reference and fixed column reference

DAXUDF Reference a column