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,
In Direct Query mode, is there a way to calculate median numbers such that if the number of values selected is odd then the middle one will become the median. If the number of values chosen are odd, then the median would become the value of the "upper" median value instead of being an average. For instance, if the numbers chosen are : 1, 3, 4, 6. The median of this set of numbers is 4.
If the numbers chosen were 1, 3, and 4, then the 3 would be chosen/displayed as the median.
In this case, the number that should be chosen is 2.333.
Solved! Go to Solution.
Hi again,
It turns out you can do something like this, assuming you have enabled all functions in DirectQuery.
I have taken this code basically verbatim from
https://www.daxpatterns.com/statistical-patterns/#median22
Median Unorthodox =
VAR NumValuesHalved = COUNT ( YourTable[Latency] ) / 2
RETURN
MINX (
FILTER (
VALUES ( YourTable[Latency] ),
CALCULATE (
COUNT ( YourTable[Latency] ),
YourTable[Latency] <= EARLIER ( YourTable[Latency] )
)
> NumValuesHalved
),
YourTable[Latency]
)The code produces the "upper" median value in the case of an even number of items, otherwise produces the middle value.
Regards,
Owen 🙂
Hi again,
It turns out you can do something like this, assuming you have enabled all functions in DirectQuery.
I have taken this code basically verbatim from
https://www.daxpatterns.com/statistical-patterns/#median22
Median Unorthodox =
VAR NumValuesHalved = COUNT ( YourTable[Latency] ) / 2
RETURN
MINX (
FILTER (
VALUES ( YourTable[Latency] ),
CALCULATE (
COUNT ( YourTable[Latency] ),
YourTable[Latency] <= EARLIER ( YourTable[Latency] )
)
> NumValuesHalved
),
YourTable[Latency]
)The code produces the "upper" median value in the case of an even number of items, otherwise produces the middle value.
Regards,
Owen 🙂
Thanks so much for your help. I can't even begin to tell you how much time I spent trying to google the answer and find out. Cheers
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 |
|---|---|
| 24 | |
| 22 | |
| 21 | |
| 20 | |
| 14 |
| User | Count |
|---|---|
| 58 | |
| 52 | |
| 41 | |
| 33 | |
| 32 |