Difference in Row Values Based on Other Columns

Hi Ncf5031 ,

I created a sample pbix file(see attachment) for you, please check whether that is what you want. You can create a measure as below:

Hrs Diff = 
VAR _selsn =
    SELECTEDVALUE ( 'Failures'[S/N] )
VAR _curdate =
    SELECTEDVALUE ( 'Failures'[date] )
VAR _predate =
    CALCULATE (
        MAX ( 'Failures'[date] ),
        FILTER ( ALLSELECTED ( 'Failures' ), 'Failures'[date] < _curdate )
    )
VAR _prehours =
    CALCULATE (
        SUM ( 'Failures'[Hours] ),
        FILTER ( ALLSELECTED ( 'Failures' ), 'Failures'[date] = _predate )
    )
RETURN
    IF ( ISBLANK ( _prehours ), BLANK (), SUM ( 'Failures'[Hours] ) - _prehours )

If the above one can't help you get the desired result, please provide some sample data in your tables (exclude sensitive data) with Text format and your expected result with backend logic and special examples. It is better if you can share a simplified pbix file. Thank you.

How to upload PBI in Community

Best Regards

Thank you all for your help! I ended up using techniques from the suggestions of both amitchandak and Anonymous.

Below is the code snippet that results in a ~mostly~ correct solution. I say mostly because Ideally I would eliminate the hours for the first failure, but it's not the end of the world if they show up. Perhaps this could be solved by using EVALUATE and START AT

Time Between Failures = 
VAR SerialNumber = 'FAILURES'[SERIAL_NUM]
VAR DT = 'FAILURES'[DATET]
VAR LAST_HOURS = 'FAILURES'[HRS]

RETURN
LAST_HOURS - (CALCULATE(MAX('FAILURES'[HRS]), TOPN(1,FILTER('FAILURES', 'FAILURES'[SERIAL_NUM] = SerialNumber && 'FAILURES'[DATE] < DT), 'FAILURES'[DATE], DESC)))