Running sum last 6 months measure

[ChristianRHouen]

Currently trying to make a running sum of Number1 & Number2 for last 6 months. Something weird is happening at the end, and the logic does not follow the intended calculations.

I used following formula:

Number1_L6M_sum = 
CALCULATE (
    SUM ( Table1[Number1] ),
    filter(all(Table1[Date]), Table1[Date] < min(Table1[Date]) && DATEADD(Table1[Date], 6, MONTH) >= min(table1[Date]))
)

But as you can numbers start to not add up as they approach current date. (Green on left should match green on right and red ditto)

Why is this happening - And what is a potential solution?

[v-frfei-msft · Accepted solution]

Hi @ChristianRHouen ,

I have created a sample for your reference, Please create a measure as below to work on it.

Measure = 
VAR pre6 =
    EDATE ( MAX ( 'Table 1'[date] ), -6 )
VAR cure =
    MAX ( 'Table 1'[date] )
RETURN
    CALCULATE (
        SUM ( 'Table 1'[Number1] ),
        FILTER ( ALL ( 'Table 1' ), 'Table 1'[date] >= pre6 && 'Table 1'[date] <= cure )
    )

For more details, please check the pbix as attached.

[ChristianRHouen]

I stand corrected - it's actually quite flexible. The only reason I had issues was that the code should be 

'Table 1'[date] <= cure )

Instead of

'Table 1'[date] < cure )

 in order to include the last day of the month.

Would you correct your answer, so I can mark it as correct? 🙂  

[v-frfei-msft]

Hi @ChristianRHouen ,

I have updated my previous post:)

[ChristianRHouen]

Thanks, Frank!

Unfortunately, this solution doesn't seem to scale well, as the underlying data model actually has many rows for each month, and Number1 is actually a Number1 = sum(underlying_number1).

This means that for the sum, I can easily view it in different date buckets, be they year, month, week. 

Your solution is definitely more correct than mine, but even if I change the date in your file to years it becomes non-sensical.

Do you have a different solution maybe?