Forum Discussion
Excel Calc in Power BI
Hi Rehansaeed2024 ,
Thanks for confirming that the results are matching in the reference PBIX. This confirms that the Excel decayed SUMPRODUCT logic is correctly implemented in Power BI using DAX.The issues you are facing in your actual PBIX are not due to the formula, but mainly because of model structure and data size. The circular dependency error usually comes when a measure like Numerator_Decayed is used directly or indirectly inside a calculated column, or when the Row Index itself is created using measures. In the working PBIX, the Row Index is a simple calculated column, and both Numerator and Denominator are measures only, which avoids this issue.
The “Query exceeded available resources” message happens when the SUMX + FILTER logic runs over a very large or highly detailed dataset. In such cases, the data grain or Row Index scope needs to be reduced to improve performance.
When you see 0.00 values for smaller datasets, it usually means the Row Index is not stable or the table is not sorted in the same order as Excel before assigning the index. Since Excel’s ROW() depends on position, sorting is very important.
Please use the attached PBIX as a reference for how the Row Index and measures should be structured. Once the same separation and sorting are applied in your actual model, the Numerator_Decayed will behave exactly the same as Excel.
Hi Rehansaeed2024 ,
I've recreated your entire scenario in a new sample .pbix file to verify the behavior. After setting up the same tables, adding the Row Index, and applying the decay based numerator and denominator logic, I was able to reproduce the exact expected output on my side. Both the running numerator calculation and the denominator values are working correctly when the DAX measures are created under the LineItems table.
This confirms that the logic you are using is valid, and Power BI calculates the decayed SUMPRODUCT pattern exactly as intended once the Row Index and Alpha value are set up properly. For your reference, I've attached the working .pbix file so you can compare it with your model and follow the steps used to get the final result.
Here is a link to the Excel File.
Here is the link to the sample PBIX file using this EXcel. Denominator Column I am able to get it working struggling with the numerator column.
https://s3-project-rehan.s3.us-east-2.amazonaws.com/Sample_PBIX.pbix
Hi Rehansaeed2024 ,
To reproduce Excel’s decayed SUMPRODUCT logic in Power BI, I first created a stable Row Index column using RANKX in Table A and then brought it into Table B using Row Index_TableB = RELATED('Table A'[Row Index]). This gives Power BI the row-by-row context that Excel gets from ROW(). The denominator was already correct using Alpha = 0.964783. The numerator, which corresponds to SUMPRODUCT(C$2:C3, Alpha^(ROW(C3)-ROW(C$2:C3))), is implemented in DAX by summing all previous Product Cost values with decay based on the row distance:
Numerator_Decayed :=
VAR Alpha = 0.964783
VAR CurrentIndex = SELECTEDVALUE('Table B'[Row Index_TableB])
VAR MinIndex = CALCULATE(MINX(ALLSELECTED('Table B'), 'Table B'[Row Index_TableB]))
RETURN
IF(
ISBLANK(CurrentIndex) || ISBLANK(MinIndex),
BLANK(),
CALCULATE(
SUMX(
FILTER(ALLSELECTED('Table B'), 'Table B'[Row Index_TableB] <= CurrentIndex),
'Table B'[Product Cost] *
POWER(Alpha, CurrentIndex - 'Table B'[Row Index_TableB])
)
)
)
The final weighted output is:Weighted_ProductCost :=
DIVIDE([Numerator_Decayed], [Denominator])
Once the Row Index column and these measures are applied, the Numerator and Weighted values match Excel’s SUMPRODUCT output exactly, fully solving the translation of the Excel formula into Power BI DAX.Getting incorrect Numerator Numbers.
Product ID Order Date Sum of Product Cost Denominator Numerator_Decayed 60055423 1/2/2024 0:00 8868094 60056134 1/2/2024 0:00 7629201 1 66,139,611.38 60058514 1/2/2024 0:00 5618481 1.964783 63,810,372.69 60055845 1/5/2024 0:00 13441245 2.895588 61,563,162.79 60056380 1/5/2024 0:00 2053174 3.793613 59,395,092.89 60059183 1/5/2024 0:00 9238018 4.660012 57,303,375.90 60058377 1/10/2024 0:00 9470897 5.495899 55,285,322.91 60051952 1/11/2024 0:00 9820500 6.302348 53,338,339.69 Hi Rehansaeed2024 ,
I've attached a PBIX file to help you verify the calculation. It includes a synced Row Index so Excel and Power BI follow the same row order, the full NumeratorDecayed measure, and a table showing the correct results. Once you open the file and view the sorted table, you’ll see the Numerator matches Excel because the decay logic depends on row order.
Please find below attached .pbix file for your reference.
- Ashish_Mathur
Super User
Hi,
In cell B2 of Table A worksheet, shouldn't the date be Jan 1, 2025. Also, are all dates in mm/dd/yy format?
