Cummulative Prev Change

[Anonymous]

Good afternoon 

I am struggling with a DAX formula, and will appreciate some help

Want to create a formula for "Cummulative Prev Change" as per screenshot and following conditions

1) First two rows have to be zero

2) from 3rd row, the formula should get PreviousVersions's "Requested Value" + PreviousVersions's "Cummulative Prev Change" if

I tried following formula, to no effect 😞  

CummulativePrevChange2 = IF(PA_KEY[Version]>1,LOOKUPVALUE(PA_KEY[RequestedValue],PA_KEY[Version],PA_KEY[Version]-1,PA_KEY[PA#],PA_KEY[PA#]),0)

Attached are the sample file and screenshot of how the column should calculate values...

PowerBI File 

How it should look like 

Thank you in advance for any advise 🙂

[Nathaniel_C · Accepted solution]

Hi @Anonymous 
The solution at this point is two columns.


Let me know if you have any questions.

If this solves your issues, please mark it as the solution, so that others can find it easily. Kudos are nice too.
Nathaniel

Cumulative = 

var _curIndex =PA_KEY[Index]
var _minIndex = CALCULATE(MIN(PA_KEY[Index]),ALLEXCEPT(PA_KEY,PA_KEY[PA#]),PA_KEY[Index]<_curIndex)

return

IF(_curIndex = _minIndex || _curIndex = _minIndex+1,BLANK(), CALCULATE(sum(PA_KEY[RequestedValue]),ALLEXCEPT(PA_KEY,PA_KEY[PA#]),PA_KEY[Index]=_curIndex -1))

Cumulative and previous = 
VAR _curIndex = PA_KEY[Index]   //Get current Index from this row
VAR _minIndex = CALCULATE ( MIN ( PA_KEY[Index] ), ALLEXCEPT ( PA_KEY, PA_KEY[PA#] ), PA_KEY[Index] < _curIndex )    //Get minimum index for this PA#                                
                              
VAR _prev3 = CALCULATE ( SUM ( PA_KEY[Cumulative] ),  ALLEXCEPT ( PA_KEY, PA_KEY[PA#] ), PA_KEY[Index] = _curIndex - 1 )   // Get the previous row amount from the Cumulative column                                

RETURN
    IF (                                        //Return a blank if the current index is equal to the first two rows of this PA# else do the calc and add the previous row from the Cumulative column
        _curIndex = _minIndex
            || _curIndex = _minIndex + 1,
        BLANK (),
        CALCULATE (
            SUM ( PA_KEY[RequestedValue] ),
            ALLEXCEPT (
                PA_KEY,
                PA_KEY[PA#]
            ),
            PA_KEY[Index] = _curIndex - 1
        ) + _prev3
    )

[jdbuchanan71 · Accepted solution]

@Anonymous 

Just this portion is the measure and it relies on the index column @Nathaniel_C  added to your sample table.  

Column = 
VAR _MinIndex = CALCULATE ( MIN ( PA_KEY[Index] ), ALLEXCEPT ( PA_KEY, PA_KEY[PA#] ) )
VAR _RowIndex = [Index]
RETURN CALCULATE ( SUM ( PA_KEY[RequestedValue] ), ALLEXCEPT ( PA_KEY, PA_KEY[PA#] ), PA_KEY[Index] > _MinIndex && PA_KEY[Index] < _RowIndex )

 Are you able to add an index column to your source data that will properly order the lines?

[Nathaniel_C]

[Anonymous]

thank you Nathan,

Kindly can you share the DAX for your column?

thank you

-Usman

[Nathaniel_C · Accepted solution]

Hi @Anonymous 
The solution at this point is two columns.


Let me know if you have any questions.

If this solves your issues, please mark it as the solution, so that others can find it easily. Kudos are nice too.
Nathaniel

Cumulative = 

var _curIndex =PA_KEY[Index]
var _minIndex = CALCULATE(MIN(PA_KEY[Index]),ALLEXCEPT(PA_KEY,PA_KEY[PA#]),PA_KEY[Index]<_curIndex)

return

IF(_curIndex = _minIndex || _curIndex = _minIndex+1,BLANK(), CALCULATE(sum(PA_KEY[RequestedValue]),ALLEXCEPT(PA_KEY,PA_KEY[PA#]),PA_KEY[Index]=_curIndex -1))

Cumulative and previous = 
VAR _curIndex = PA_KEY[Index]   //Get current Index from this row
VAR _minIndex = CALCULATE ( MIN ( PA_KEY[Index] ), ALLEXCEPT ( PA_KEY, PA_KEY[PA#] ), PA_KEY[Index] < _curIndex )    //Get minimum index for this PA#                                
                              
VAR _prev3 = CALCULATE ( SUM ( PA_KEY[Cumulative] ),  ALLEXCEPT ( PA_KEY, PA_KEY[PA#] ), PA_KEY[Index] = _curIndex - 1 )   // Get the previous row amount from the Cumulative column                                

RETURN
    IF (                                        //Return a blank if the current index is equal to the first two rows of this PA# else do the calc and add the previous row from the Cumulative column
        _curIndex = _minIndex
            || _curIndex = _minIndex + 1,
        BLANK (),
        CALCULATE (
            SUM ( PA_KEY[RequestedValue] ),
            ALLEXCEPT (
                PA_KEY,
                PA_KEY[PA#]
            ),
            PA_KEY[Index] = _curIndex - 1
        ) + _prev3
    )

[Nathaniel_C]

Hi @Anonymous , @jdbuchanan71 ,
You are in luck! This is from @jdbuchanan71 , one column and much simpler code.

Hi Nathaniel,

The problem I see is how to determine the order of the lines.  I know you added an index but will the rows always come out of the source in the right order?  I would verify that with the original poster.

If the index can be added safely then this code should work.

Column = 
VAR _MinIndex = CALCULATE ( MIN ( PA_KEY[Index] ), ALLEXCEPT ( PA_KEY, PA_KEY[PA#] ) )
VAR _RowIndex = [Index]
RETURN CALCULATE ( SUM ( PA_KEY[RequestedValue] ), ALLEXCEPT ( PA_KEY, PA_KEY[PA#] ), PA_KEY[Index] > _MinIndex && PA_KEY[Index] < _RowIndex )


=========================

We don't need to do any checking for the first 2 rows because the filters  PA_KEY[Index] > _MinIndex && PA_KEY[Index] < _RowIndex take care of it.

On index line 5 for example.

_MinIndex = 4

_RowIndex = 5




We ask for the sum of PA_KEY[RequestedValue] where the Index is > 4 AND < 5.  No such lines so we get nothing.
On Index line 7 we as for the sum of [RequestedValue] where the Index > 4 AND < 7.  We get the sum of Index 5 and 6

Nathaniel

[jdbuchanan71 · Accepted solution]

@Anonymous 

Just this portion is the measure and it relies on the index column @Nathaniel_C  added to your sample table.  

Column = 
VAR _MinIndex = CALCULATE ( MIN ( PA_KEY[Index] ), ALLEXCEPT ( PA_KEY, PA_KEY[PA#] ) )
VAR _RowIndex = [Index]
RETURN CALCULATE ( SUM ( PA_KEY[RequestedValue] ), ALLEXCEPT ( PA_KEY, PA_KEY[PA#] ), PA_KEY[Index] > _MinIndex && PA_KEY[Index] < _RowIndex )

 Are you able to add an index column to your source data that will properly order the lines?

[Anonymous]

thank you JDBuchanon, and Nathan for your replies

Adding an index (order by) will make query quite slow.

Is there any other suggestions without adding index column please ?

thank you again

-Usman

[jdbuchanan71]

@Anonymous 

If the data always come into the model in the correct order you can just add an index row in the query editor the way @Nathaniel_C did.

Without an index column, what it the logic to determine "previous row"?  Keep in mind that PowerBI does not understand "this row is the row before the row I am on" the way excel does.  That is why we need an index column or some other logic to tell PowerBI how to search the entire table and find the previous rows.  Also, what is your data source for the information?

[Anonymous]

thanks for your reply

"PA_Key" and "Invoice Line" is the unique key/index; e.g. "1043-1-1" which shows "Purchase Order-Version#-Invoice Line#" respectively

Can we somehow use the original table to do it, or would we have to add a new index and use order by clause, to always order them by Purchase Order, Invoice Line#, Version # ?

thank you

[jdbuchanan71 · Accepted solution]

The problem with using the original table is you don't know what order the lines are going to come in unless you give it an order by.  You could add the index on the SQL side as well, something like this.

SELECT
	PA_Key,
	Version,
	[Invoice LINE],
	CONTRACTLINEVALUE,
	REQUESTEDVALUE,
	PA#,
	ROW_NUMBER() OVER( ORDER BY ( SELECT 0 ) ) AS RowIndex
FROM Table
ORDER BY
	PA_Key,
	Version,
	[Invoice LINE]

[Anonymous]

Thank you JDBuchanan and Nathan, for all you help.

Much appreciated 🙂