PowerBI calculate overflow backlog in measure - powerbi

Question

My data is as follows:
Factory ID
Week
CAPACITY
Request
1
21
1000
500
1
22
1000
1200
1
23
1500
1600
1
24
1500
1100
2
21
1000
500
2
22
2000
1900
2
23
2000
1000
2
24
2000
2500
3
21
1000
200
3
22
1000
900
3
23
1000
1300
3
24
1000
800
I want to calculate backlog in a measure or any other way to have backlog be dynamic based on the factories I select. Backlog is calculated as follows:
Backlog = Capacity - (Request + Previous week backlog); where we have backlog when requests + pre. week backlog exceeds capacity or else it is 0. I cannot move capacity from future weeks, so the backlog would always accumulate going forward
Eg. If I select Factory 1, my backlog should look as follows:
Factory Selected: 1
Week
Backlog
21
0
22
-200
23
-300
24
-100
Factory Selected: 1,2
Week
Backlog
21
0
22
-100
23
0
24
-400
Factory Selected: 1,3
Week
Backlog
21
0
22
-100
23
-500
24
0
I have been trying to find a solution since the last 2 days. Let me know if you need any additional details. Any help will be greatly appreciated.

Answer 1

In DAX there are no loops and no recursion, this means that we must write some very ugly loop unrolling dax code. This is a minimal implementation of a working measure, but to make it working in the general case the BacklogWeekNN variables must be added until reaching the maximum possible week depth in the model.
Backlog =
VAR MinWeek = CALCULATE(MIN( T[Week] ), REMOVEFILTERS( T ) )
VAR MaxWeek = MAX( T[Week] )
VAR TAggregated =
ADDCOLUMNS(
CALCULATETABLE( VALUES( T[Week] ), T[Week] <= MaxWeek, REMOVEFILTERS( T ) ),
"#Capacity", CALCULATE( SUM( T[CAPACITY] ), ALLEXCEPT( T, T[Week], T[Factory ID] ) ),
"#Request", CALCULATE( SUM( T[Request] ), ALLEXCEPT( T, T[Week], T[Factory ID] ) )
)
VAR BacklogWeek00 = SUMX( FILTER( TAggregated, T[Week] = MinWeek ), [#Capacity] - [#Request] ) + 0
VAR BacklogWeek01 = SUMX( FILTER( TAggregated, T[Week] = MinWeek + 1), [#Capacity] - [#Request] ) + IF(BacklogWeek00 > 0, 0, BacklogWeek00)
VAR BacklogWeek02 = SUMX( FILTER( TAggregated, T[Week] = MinWeek + 2), [#Capacity] - [#Request] ) + IF(BacklogWeek01 > 0, 0, BacklogWeek01)
VAR BacklogWeek03 = SUMX( FILTER( TAggregated, T[Week] = MinWeek + 3), [#Capacity] - [#Request] ) + IF(BacklogWeek02 > 0, 0, BacklogWeek02)
VAR Result = IF(BacklogWeek03 > 0, 0, BacklogWeek03)
RETURN Result
This way we can obtain the desired resulting matrix using a slicer to select the factories (I'm afraid there are a few errors in the expected result samples int the question)
Edit: I used 'FILTER( TAggregated, T[Week] = MinWeek )' instead of the equivalent CALCULATE/CALCULATTABLE DAX code to avoid context transition happening and because the weeks table has very few rows to be iterated (I can imagine a maximum of a few tens or at maximum hundreds of rows if keeping a few years history)