Help Calculating T=0, T=1, T=2...

[Anonymous]

Hi everyone,

Need some assistance calculating T values. I thought about using Exponential Regression but maybe that's not the right way. You guys may have a simpler solution.

Scenario:

I have a list of different hadware equipment that have been failling over time. I'm trying to plot these failures based on their actual date but converted to T=X instead of using the date itself.

Chart example:

NOTE: T=0 is given to me as a fixed dataset.

Here is a table example of the data:

Serial Number	Failure Date	Failure Month	Count failures per month	T=0	T=X (not sure how to calculate this)
12345	02-Dec-2017	Dec-2017	1	Dec-2016	T=1
12345	18-Sep-2019	Sep-2019	2	Dec-2016	T=2
1378Gb49	09-Sep-2019	Sep-2019	2	Jul-2017	T=1
2894GP20	27-Aug-2018	Aug-2018	1	Apr-2018	T=1
12345	26-Feb-2020	Feb-2020	1	Dec-2016	T=3

You will notice that from the table above, the serial number 12345 failed 3 times in a period of 3 years (T=0 being Dec-2016). Therefore each failure gains a +1 to T=X. However, the T=X calculation should always start from the base T=0 date.

NOTE: Each individual serial number will have their own set of T=X. I have a lot of different serial numbers, the plan is to group all failures by month (independent of serial number). The serial number is only used to properly calculate T=X.

I'm preparing this in Power BI, I'm assuming I will have to use Power Query due to the complexity. I'm not the greatest mathematician out there so I'm probably stuck on something simple to resolve.

Please let me know if this is something you guys can assist. Also, let me know if you require more information and I'll add to this post.

Thank you in advance!

[v-eachen-msft]

Hi @Anonymous ,

Just a supplement to the case of T=0:

T =
VAR _Rank =
    RANKX (
        FILTER ( 'Table', 'Table'[Serial Number] = EARLIER ( 'Table'[Serial Number] ) ),
        'Table'[Failure Date],
        ,
        ASC
    )
RETURN
    IF (
        FORMAT ( 'Table'[T=0], "yyyymm" )
            = FORMAT (
                CALCULATE (
                    MIN ( 'Table'[Failure Month] ),
                    ALLEXCEPT ( 'Table', 'Table'[Serial Number] )
                ),
                "yyyymm"
            ),
        "T = " & _Rank - 1,
        "T = " & _Rank
    )

[az38 · Accepted solution]

Hi @Anonymous 

it looks like a good task for RANKX column:

T = 
var _Rank = 
    RANKX(
        FILTER(
                'Table',
                'Table'[Serial Number]=EARLIER('Table'[Serial Number])
                ),
        'Table'[Failure Date], , ASC
        )

RETURN

"T = " & _Rank 

[Anonymous]

Hi @az38 ,

My preliminary tests have been successful with your formula.

I'll keep working on it but I think that you got it.

Thank you very much for the great solution.

Cheers!