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

Earn a 50% discount on the DP-600 certification exam by completing the Fabric 30 Days to Learn It challenge.

Reply
Anonymous
Not applicable

Dynamic outlier detection based on standard deviation

I have a scatterplot where the x-axis is a paid dollar amount and the y-axis is a count and each dot represents a different person.  Both fields are calculated measures.  I have used the scatterplot visual to create lines on the chart for the mean, 68th percentile (1SD), and 95th percentile (2SD) for each axis.  I have multiple filters at the top such as date range, location, etc.  The mean and standard deviations would change based on the filter selections.  I want to color code the points on the scatterplot to be a different color if both axis' are >= 2SD (how we are defining an outlier).  I've seen outlier detection methods for PowerBI but they seem to be based on more static thesholds, while mine would not be.

 

I was able to create another measure field for the SD's (since the visual automatically calculated it) and I thought I needed a calculated field so I could just compare that in a formula.  But I want to compare each dot/person's count to the total standard deviation of the count (and also each person's paid dollar amount to the standard deviation of that), which I'm unable to compute.  Referencing the table below for example, I'd want to compare the count field for each person to the total of the std_of_count field and that would be an outlier, but instead I'm only able to figure out how to do it for each row.  I believe this would be a start to figuring it out so I can label this as an outlier (1) or not (0), then I can use the data colors/markers field within the visual to color code this as a rule.

 

PERSON COUNT (y-axis)STD_OF_COUNT
Person 155
Person 233
.........
Person 5011
TOTAL2001
1 REPLY 1
v-binbinyu-msft
Community Support
Community Support

Hi @Anonymous ,

In order to better understand your demands and give the right solution, could you please provide some more specific information? such as your desensitized example data and a screenshot of your desired results?

Thanks for your efforts & time in advance.

 

Best regards,
Community Support Team_ Binbin Yu

Helpful resources

Announcements
RTI Forums Carousel3

New forum boards available in Real-Time Intelligence.

Ask questions in Eventhouse and KQL, Eventstream, and Reflex.

MayPowerBICarousel

Fabric Monthly Update - May 2024

Check out the May 2024 Fabric update to learn about new features.

LearnSurvey

Fabric certifications survey

Certification feedback opportunity for the community.