Continuous subtracting of values from previous row in a different column - powerbi

Question

I am working on a Power BI dashboard that is forecasting values up until 2030, depending on projects that are under implementation between 2022 to 2030.
I have one table [Actual] that contains actual values from 2021 and 2022 which I have at a monthly frequency.
I also have another table [Project Impacts] that contains the projects, with the impacts that they have on the actual values, and this is available from 2022 to 2030, at a quarterly frequency.
There are a few important measures relevant to this problem.
From the CO2 table, we have ‘Actual Values’.
From the Projects table, we have a measure named ‘Estimated saving from projects’, which calculates how much savings there are based on the projects each quarter. The table below shows the measures next to each other.
What I want to do is subtract the ‘Estimated savings from projects’ from the ‘Actual Values’, so that I am left with the ‘Forecast Values’. This is the results I get after doing this (image below).
As you can see from the image above, the ‘Forecast Values’ seems to be working properly, up until 2023/Q1. This is because there is no current data for 'Actual Values' after 2022, so the ‘Estimated saving from Projects’ is just subtracting from 0, hence the negative values in the far right column.
What I want to do is keep the calculation the same up until the end of 2022, but after 2022, the ‘Estimated saving from Projects’ should be subtracted from the last quarters value.
So as an example, the 2023/Q1 ‘Estimated saving from Projects’ value of 7.87 should be subtracted from the 53.51 value from the 2022/Q4 value in the ‘Forecast Value’ measure, giving a value of 45.64 instead of the value -7.57. Then the same process should repeat, so the 7.87 from 2023/Q2 should be subtracted from the 45.64 from the previous quarter and so on.
Does anyone know of a DAX formula, calculated column, or any solution to the problem above?
Many thanks

Answer 1

OK, so this isn't DAX, but I think the result works in your situation. I resolved it in Power Query. I recreated your table and added on to it:
+------+---------+--------------+------------------+
| Year | Quarter | ActualValues | EstimatedSavings |
+------+---------+--------------+------------------+
| 2021 | 2021Q1 | 83.88 | |
| 2021 | 2021Q2 | 72.84 | |
| 2021 | 2021Q3 | 72.25 | |
| 2021 | 2021Q4 | 68.13 | |
| 2022 | 2022Q1 | 77.53 | 0.17 |
| 2022 | 2022Q2 | 67.75 | 0.17 |
| 2022 | 2022Q3 | 63.08 | 0.17 |
| 2022 | 2022Q4 | 58.14 | 4.63 |
| 2023 | 2023Q1 | | 7.87 |
| 2023 | 2023Q2 | | 7.87 |
| 2023 | 2023Q3 | 104.33 | 8.94 |
| 2023 | 2023Q4 | | 5.11 |
| 2024 | 2024Q1 | | 10.08 |
| 2024 | 2024Q2 | | 10.08 |
| 2024 | 2024Q3 | | 9.04 |
| 2024 | 2024Q4 | | 8.42 |
+------+---------+--------------+------------------+
For illustration purposes, I then created a column I called "ForeCastValuesOLD". You will see below that instead of negative values, you get a "null" value, because the empty cells in "ActualValues" translate to a "null" value in Power Query (expected behavior for a column formatted as decimal number), so subtracting "EstimatedSavings" from this results in another "null" value. It's still wrong in terms of what you want, but it's a "different wrong".
I then selected the "ActualValues" column, and on the "Transform" tab clicked on the arrow next to "Fill" and then on "Down".
Note: "Fill | Down" will not work if you have zeros in that column, it only works with null values. See also here.
As a last step, I repeated the ForeCastValues calculated column, only with the new "ActualValues" column. The results are what you are looking for, I believe, and it doesn't matter how many blanks you have or where they are in that column.
This is the final table:
+------+---------+--------------+------------------+-------------------+----------------+
| Year | Quarter | ActualValues | EstimatedSavings | ForeCastValuesOLD | ForeCastValues |
+------+---------+--------------+------------------+-------------------+----------------+
| 2021 | 2021Q1 | 83.88 | | | |
| 2021 | 2021Q2 | 72.84 | | | |
| 2021 | 2021Q3 | 72.25 | | | |
| 2021 | 2021Q4 | 68.13 | | | |
| 2022 | 2022Q1 | 77.53 | 0.17 | 77.36 | 77.36 |
| 2022 | 2022Q2 | 67.75 | 0.17 | 67.58 | 67.58 |
| 2022 | 2022Q3 | 63.08 | 0.17 | 62.91 | 62.91 |
| 2022 | 2022Q4 | 58.14 | 4.63 | 53.51 | 53.51 |
| 2023 | 2023Q1 | 58.14 | 7.87 | | 50.27 |
| 2023 | 2023Q2 | 58.14 | 7.87 | | 50.27 |
| 2023 | 2023Q3 | 104.33 | 8.94 | 95.39 | 95.39 |
| 2023 | 2023Q4 | 104.33 | 5.11 | | 99.22 |
| 2024 | 2024Q1 | 104.33 | 10.08 | | 94.25 |
| 2024 | 2024Q2 | 104.33 | 10.08 | | 94.25 |
| 2024 | 2024Q3 | 104.33 | 9.04 | | 95.29 |
| 2024 | 2024Q4 | 104.33 | 8.42 | | 95.91 |
+------+---------+--------------+------------------+-------------------+----------------+
And here is the M code:
let
Source = Table.FromRows(Json.Document(Binary.Decompress(Binary.FromText("fc5LDsMgDATQu7CORvgDdo7RdcT9r1EMjdSWJCuP9MayjyNxZkrbGK8ILnDvM7XtF7kHY7heoUzkcoWxUR0kX8gTedw0QwnMIFsKEarByl0hNqsg+10h7pd+P6aiylmQWZDxQ7wBt8X4wSJQVkgEx65LQ+d2AdFpOk3Pq5Tj9X/kJ5SJO7Iu9rnoUE6tvQE=", BinaryEncoding.Base64), Compression.Deflate)), let _t = ((type nullable text) meta [Serialized.Text = true]) in type table [Year = _t, Quarter = _t, ActualValues = _t, EstimatedSavings = _t]),
#"Changed Type" = Table.TransformColumnTypes(Source,{{"Quarter", type text}, {"ActualValues", type number}, {"EstimatedSavings", type number}}),
#"Added Custom1" = Table.AddColumn(#"Changed Type", "ForeCastValuesOLD", each [ActualValues] - [EstimatedSavings]),
#"Changed Type2" = Table.TransformColumnTypes(#"Added Custom1",{{"ForeCastValuesOLD", type number}}),
#"Filled Down" = Table.FillDown(#"Changed Type2",{"ActualValues"}),
#"Added Custom" = Table.AddColumn(#"Filled Down", "ForeCastValues", each [ActualValues] - [EstimatedSavings]),
#"Changed Type1" = Table.TransformColumnTypes(#"Added Custom",{{"ForeCastValues", type number}})
in
#"Changed Type1"