Skip to main content
cancel
Showing results for 
Search instead for 
Did you mean: 

Earn the coveted Fabric Analytics Engineer certification. 100% off your exam for a limited time only!

Reply
Van22
Regular Visitor

Outliers in DAX

Hi, I'm working on a dataset that has 2 (or more) months worth of data. We mark the outliers manually in Excel by sorting the resolution hrs and then getting 5% of the total rows per month and then marking the bottom and top 5% with "Outlier" and the middle values with "Normal". In the final measure, we want the value to exclude all those marked with "Outlier" in a monthly basis. 

 

I need a calculated column that will do this for me in PowerBI so that when I use the slicer for month, the marking for Outlier and Normal will automatically change as well. 

 

Do you have any suggestions? Here's the sample data set. 

 

IDRestored HoursMonth RestoredOutlier Flag
C-0010.012021-05Outlier
C-0121.1288783432021-05Outlier
C-0151.2014204982021-05Normal
C-0101.2491101172021-05Normal
C-0161.3004064082021-05Normal
C-0091.4278883192021-05Normal
C-0111.6230734332021-05Normal
C-0131.6429827132021-05Normal
C-0081.7080778482021-05Normal
C-0141.8032076682021-05Normal
C-0061.813980352021-05Normal
C-0071.966169622021-05Normal
C-00232021-05Normal
C-0175.0897301672021-05Normal
C-0195.2567098852021-05Normal
C-0225.3100292022021-05Normal
C-0245.3150798532021-05Normal
C-0185.4692671532021-05Normal
C-0235.5873053292021-05Normal
C-0215.7423168622021-05Normal
C-0205.9939406092021-05Normal
C-003512021-05Normal
C-0052122021-05Outlier
C-00457752021-05Outlier
C-0463.0430738782021-06Outlier
C-0423.186849552021-06Outlier
C-0433.6052380512021-06Normal
C-0443.6853298032021-06Normal
C-0453.9505394662021-06Normal
C-0384.3540582262021-06Normal
C-0414.6488599572021-06Normal
C-0404.6673963482021-06Normal
C-0374.8742793532021-06Normal
C-0394.9727515752021-06Normal
C-0275.042435552021-06Normal
C-0265.3849489042021-06Normal
C-0255.4397205672021-06Normal
C-0285.9724869352021-06Normal
C-0496.0747145732021-06Normal
C-0506.4715259932021-06Normal
C-0486.679320562021-06Normal
C-0476.8531024932021-06Normal
C-03511.23757772021-06Normal
C-02911.336805212021-06Normal
C-03011.390725332021-06Normal
C-03611.393329492021-06Normal
C-03111.628650162021-06Normal
C-03411.669616072021-06Normal
C-03211.830122772021-06Outlier
C-03311.92019542021-06Outlier

 

Thanks!

 

1 ACCEPTED SOLUTION
V-lianl-msft
Community Support
Community Support

Hi @Van22 ,

 

Try to create a new column like below:

outlier_test = 
var rank_asc = RANKX(FILTER('Table',EARLIER('Table'[Month Restored])='Table'[Month Restored]),'Table'[Restored Hours],,ASC,Dense)
var rank_desc = RANKX(FILTER('Table',EARLIER('Table'[Month Restored])='Table'[Month Restored]),'Table'[Restored Hours],,DESC,Dense)
var count_row = COUNTROWS(FILTER('Table',EARLIER('Table'[Month Restored])='Table'[Month Restored]))*0.05
return IF(rank_asc<=count_row||rank_desc<=count_row,"Outlier","Normal")

Vlianlmsft_0-1639547689805.png

 


Best Regards,
Liang
If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.

View solution in original post

4 REPLIES 4
V-lianl-msft
Community Support
Community Support

Hi @Van22 ,

 

Try to create a new column like below:

outlier_test = 
var rank_asc = RANKX(FILTER('Table',EARLIER('Table'[Month Restored])='Table'[Month Restored]),'Table'[Restored Hours],,ASC,Dense)
var rank_desc = RANKX(FILTER('Table',EARLIER('Table'[Month Restored])='Table'[Month Restored]),'Table'[Restored Hours],,DESC,Dense)
var count_row = COUNTROWS(FILTER('Table',EARLIER('Table'[Month Restored])='Table'[Month Restored]))*0.05
return IF(rank_asc<=count_row||rank_desc<=count_row,"Outlier","Normal")

Vlianlmsft_0-1639547689805.png

 


Best Regards,
Liang
If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.

Thank you for this! Worked like a charm. 🙂

Van22
Regular Visitor

Ah, essentially, on my sample, the column Outlier should be changing and recalculated to show outlier and normal depending on the month that the user will select. 

Sorry, just getting back into pbi and i'm a bit lost. 

lbendlin
Super User
Super User

"I need a calculated column that will do this for me in PowerBI so that when I use the slicer for month, the marking for Outlier and Normal will automatically change as well. "

 

Calculated columns do not care about user interactions or slicers. Did you mean to say you need a measure?

Helpful resources

Announcements
April AMA free

Microsoft Fabric AMA Livestream

Join us Tuesday, April 09, 9:00 – 10:00 AM PST for a live, expert-led Q&A session on all things Microsoft Fabric!

March Fabric Community Update

Fabric Community Update - March 2024

Find out what's new and trending in the Fabric Community.