Forum Discussion

Anonymous's avatar
Anonymous
Not applicable
4 years ago
Solved

Conditional deduplication based on multiple columns

Hi everyone,   I need to write a very specific deduplication query and I am wondering if you can help me out. We have an ecommerce site and I am getting a daily CSV file that has transactional info...
Need Help
  • Anonymous's avatar
    Anonymous
    4 years ago

    Hi Anonymous 

    Try this code.

    Remove Duplicate =
SUMMARIZE (
    FILTER (
        'Table',
        IF (
            VAR _COUNTROW =
                CALCULATE (
                    COUNTROWS ( 'Table' ),
                    FILTER (
                        ALL ( 'Table' ),
                        'Table'[Master Order Number] = EARLIER ( 'Table'[Master Order Number] )
                            && 'Table'[Item ID] = EARLIER ( 'Table'[Item ID] )
                    )
                )
            RETURN
                _COUNTROW > 1,
            'Table'[RANK] <> 1
                || 'Table'[Invoice Date] <> BLANK (),
            'Table'[RANK] = 1
        )
    ),
    'Table'[Order Number],
    'Table'[Master Order Number],
    'Table'[Order Date],
    'Table'[Invoice Date],
    'Table'[Item ID]
)

    Result is as below.

    Best Regards,
    Rico Zhou

     

    If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.

Anonymous's avatar
Anonymous
Not applicable
4 years ago

lbendlin thanks for the reply.

 

Table 1: Sample of data before filter showing all the order IDs within Master Order # 10034858.

 

Order NumberMaster Order NumberOrder DateInvoice DateItem ID
10034858100348587/12/2021 CA-ST-EA
10034858100348587/12/2021 CA-BB-CS
10034858100348587/12/2021 CC-USR-04
10034858100348587/12/2021 CC-UPTP-04
20405998100348587/12/2021 FP-20EHH-CS
20405999100348587/12/2021 3M-SGCUP-01
10034858100348587/12/20217/13/2021CC-USR-04
10034858100348587/12/20217/13/2021CC-UPTP-04
10034858100348587/12/20217/13/2021CA-ST-EA
10034858100348587/12/20217/13/2021CA-BB-CS
20405998100348587/12/20217/13/2021FP-20EHH-CS
20405999100348587/12/20217/13/20213M-SGCUP-01

 

Table 2: Data after filter applied; duplicate order IDs within Master Order # 10034858 WHERE Invoice Date = NULL have been removed.

(Remember I want this logic to look within all Master Order #s. I have just filtered down to one Master Order # to simplify the example output.)

 

Order NumberMaster Order NumberOrder DateInvoice DateItem ID
10034858100348587/12/20217/13/2021CC-USR-04
10034858100348587/12/20217/13/2021CC-UPTP-04
10034858100348587/12/20217/13/2021CA-ST-EA
10034858100348587/12/20217/13/2021CA-BB-CS
20405998100348587/12/20217/13/2021FP-20EHH-CS
20405999100348587/12/20217/13/20213M-SGCUP-01

 

Thanks again.

Anonymous's avatar
Anonymous
Not applicable
4 years ago

Hi Anonymous 

Try this rank code.

 

RANK = 
RANKX (
    FILTER (
        'Table',
        'Table'[Order Number] = EARLIER ( 'Table'[Order Number] )
            && 'Table'[Master Order Number] = EARLIER ( 'Table'[Master Order Number] )
            && 'Table'[Item ID] = EARLIER ( 'Table'[Item ID] )
            && 'Table'[Order Date] = EARLIER ( 'Table'[Order Date] )
    ),
    'Table'[Invoice Date],
    ,
    ASC
)

 

This code will create a rank based on each [Order Number],[Master Order Number],[Order Date] and [Item ID] group. 

Then create a new dax table to remove rank =1.

 

Remove Duplicate = 
SUMMARIZE(FILTER('Table','Table'[RANK]<>1),'Table'[Order Number],'Table'[Master Order Number],'Table'[Order Date],'Table'[Invoice Date],'Table'[Item ID])

 

Result is as below.

Or you can calcualte sum directly by add rank<>1 in your table filter. You don't need to create a new table.

Such as:

Measure = CALCULATE(SUM([Sales],Filter('Table',[Rank]<>1))

 

Best Regards,
Rico Zhou

 

If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.

  • Anonymous's avatar
    Anonymous
    Not applicable
    4 years ago

    Hi Anonymous, thank you so much for the reply and for taking the time to help me out. I think this gets me closer but it's not quite what I need. I may not have explained it very clearly, but I only want to suppress rows when the following conditions are met:

     

    • There are duplicate Item IDs within a Master Order # AND those duplicate Item IDs contain a blank/NULL invoice date.

    So consider the screenshot below. The ones highlighted in green should be removed, and the white ones left alone, which your script did perfectly.

     

    But the yellow ones I highlighted, which are part of a different Master Order #, should not be removed because they do not have duplicate Item IDs nor do they have a blank Invoice Date. So I don't want to remove the yellow rows. Does this help clarify what I am asking for?

     

    • Anonymous's avatar
      Anonymous
      Not applicable
      4 years ago

      Hi Anonymous 

      Add a filter in calculated table code.

      Remove Duplicate =
SUMMARIZE (
    FILTER ( 'Table', OR ( 'Table'[RANK] <> 1, 'Table'[Invoice Date] <> BLANK () ) ),
    'Table'[Order Number],
    'Table'[Master Order Number],
    'Table'[Order Date],
    'Table'[Invoice Date],
    'Table'[Item ID]
)

      Result is as below.

      Best Regards,
      Rico Zhou

       

      If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.

      • Anonymous's avatar
        Anonymous
        Not applicable
        4 years ago

        Anonymous -- even closer but one thing is still missing. I still want to count orders that were placed but not yet invoiced. Only when there were duplicate Item IDs within a Master Order # that had blank Invoice Dates did I want to discard them from the total.

         

        Consider the Master Order # below, #10007106. It does not have an invoice date (yet) but does not have any duplicate Item IDs. In other words, it has not yet been invoiced but has not been double counted. It should still be counted in total sales (I created a separate metric for invoiced sales).

         

        So how could I modify either the deduplication table or the rank column to account for this? Again, thanks for all your help and for sticking with me on this one.