Get certified for free when you join Fabric Data Days 2026 and dive into Fabric, Power BI, SQL, AI, and other essential data skills.
Join nowTry your skills in the Power BI Dataviz World Championship! Round one ends June 26. Join now
Hi,
I am trying to calculate the average between two dates/times using DAX. I am using the following formula and it is returning a negative number even though the difference is the same date and should be -0 day 1 hr 25 mins for the first highlighted example. What am i doing wrong?
Thanks!
Solved! Go to Solution.
hi @Bryanna
try this:
duration_average =
VAR _seconds=
AVERAGEX(
TableName,
DATEDIFF( TableName[DateRequested], TableName[DateClosed], SECOND)
)
VAR _secondsABS = ABS(_seconds)
VAR _day = TRUNC(_secondsABS/(24*60*60))
VAR _hour = TRUNC( MOD(_secondsABS, 24*60*60)/(60*60))
VAR _minute = TRUNC( MOD(_secondsABS, 60*60)/60)
VAR _second = MOD(_secondsABS, 60)
RETURN
IF(
_seconds<=0,
_day&" Day "& _hour&" Hour "& _minute&" Minute "&_second&" Second Ahead",
_day&" Day "& _hour&" Hour "& _minute&" Minute "&_second&" Second Delay"
)
i tried with such data:
| DateClosed | DateRequested |
| 9/10/2022 11:00:08 | 9/10/2022 15:00:00 |
| 9/12/2022 11:00:08 | 9/12/2022 13:00:00 |
| 9/14/2022 19:00:08 | 9/14/2022 13:00:00 |
it worked like this:
The tricky part is the behaviour of TRUNC/INT/MOD for negative values. Lemme try to depict it in Excel:
Hi,
This looks much better thank you! I just removed the seconds portion since it had a very large number of decimal places.
Please see this article for a better way to deal with dates/times/durations in DAX.
Calculate and Format Durations in DAX – Hoosier BI
Pat
Hi!
I tried swapping as well but some still look odd. Still shows as a negative but they shipped it sooner than the requested ship date.
hi @Bryanna
try this:
duration_average =
VAR _seconds=
AVERAGEX(
TableName,
DATEDIFF( TableName[DateRequested], TableName[DateClosed], SECOND)
)
VAR _secondsABS = ABS(_seconds)
VAR _day = TRUNC(_secondsABS/(24*60*60))
VAR _hour = TRUNC( MOD(_secondsABS, 24*60*60)/(60*60))
VAR _minute = TRUNC( MOD(_secondsABS, 60*60)/60)
VAR _second = MOD(_secondsABS, 60)
RETURN
IF(
_seconds<=0,
_day&" Day "& _hour&" Hour "& _minute&" Minute "&_second&" Second Ahead",
_day&" Day "& _hour&" Hour "& _minute&" Minute "&_second&" Second Delay"
)
i tried with such data:
| DateClosed | DateRequested |
| 9/10/2022 11:00:08 | 9/10/2022 15:00:00 |
| 9/12/2022 11:00:08 | 9/12/2022 13:00:00 |
| 9/14/2022 19:00:08 | 9/14/2022 13:00:00 |
it worked like this:
The tricky part is the behaviour of TRUNC/INT/MOD for negative values. Lemme try to depict it in Excel:
Don't miss out on Data Days, June 15 through August 7. Learn Fabric, Power BI, SQL, AI and more.
Check out the May 2026 Power BI update to learn about new features.
| User | Count |
|---|---|
| 3 | |
| 2 | |
| 2 | |
| 1 | |
| 1 |
| User | Count |
|---|---|
| 6 | |
| 4 | |
| 3 | |
| 3 | |
| 3 |