Forum Discussion

meiwah's avatar
meiwah
Resolver I
3 years ago
Solved

How reference row context in addcolumn

I've a table as below which contains the CaseID of a case, the division of a case and the predicted division of a case. This predicted value is computed by calling a ML model. 

 

CaseID Division Division Prediction
12345AA
12346AB
12347BB
12348BC
12349CC
12350CC
12351CA

 

I would like to compute a table to summarise the results. I've tried to use ADDCOLUMN but can't get it right. Anyone, pls help to provide the correct dax syntax. Thanks! 🙂

 

Division No Cases No Predictions 
A22
B22
C33
  • Thank you for all help! I've managed to find the solution.

     

    ADDCOLUMNS(
       values(Table[Division]),
       "No Cases", CALCULATE(countrows(Table)),
       "No Predictions", CALCULATE(COUNTROWS(Table), Table[Division Prediction]=earlier([Division]), All(Table[Division]))
    )

     

    I have used the first column in AddColumns as the table I intend to loop through, which is Division.

     

    Then use Calculate to do a context transition, bringing Table[Division] the row context into the filter context. Hence, "No Cases" will have the correct figure.

     

    Then use Earlier to reference the filter context from outside of Calculate, to use it on Table[Division Prediction]. Remove the filter on Table[Division]

6 Replies

  • tamerj1's avatar
    tamerj1
    Community Champion

    Hi meiwah 

    please try

    NeawTable =
    SUMMARIZE (
    'Table',
    'Table'[Division],
    "No Cases", COUNTROWS ( 'Table' ),
    "No Predictions", DISTINCTCOUNT ( 'Table'[Division Prediction] )
    )

    • meiwah's avatar
      meiwah
      Resolver I

      Thanks Tamerj1! But the measure didn't give the right output 🙂

  • I would create a dimension table for your divisions, simply DISTINCT('Table'[Division]), create an active relationship from this to the division column in your fact table and create an inactive relationship from it to your predicted division column. You could then create measures like

    No Cases = COUNTROWS( 'Table' )
    
    No Predicted Cases = CALCULATE( COUNTROWS('Table'), USERELATIONSHIP( 'Table'[Predicted Division], Division[Division]) )
  • divpredic =
    CALCULATE(COUNTROWS('Tablename'),DISTINCT(Division),DISTINCT(Prediction))

    • meiwah's avatar
      meiwah
      Resolver I

      Thanks Devanshi! But the measure didn't give the right output 🙂

  • Thank you for all help! I've managed to find the solution.

     

    ADDCOLUMNS(
       values(Table[Division]),
       "No Cases", CALCULATE(countrows(Table)),
       "No Predictions", CALCULATE(COUNTROWS(Table), Table[Division Prediction]=earlier([Division]), All(Table[Division]))
    )

     

    I have used the first column in AddColumns as the table I intend to loop through, which is Division.

     

    Then use Calculate to do a context transition, bringing Table[Division] the row context into the filter context. Hence, "No Cases" will have the correct figure.

     

    Then use Earlier to reference the filter context from outside of Calculate, to use it on Table[Division Prediction]. Remove the filter on Table[Division]