Anomaly detector gives impossible min expected value

[MichaelDoig]

Hi all,

We're enjoying the Anomaly Detector and have added the visual to several assets across our organisation. We noticed that we're seeing some cases where the expected min value is a negative number, which is an impossible scenario in our case. 

When you hover over the visual, you can see:

Your Measure

Expected Value

Expected Min Value

Expected Max Value

Is there any way we can control this? Configure the bounds of the min and max in some way, at least to ensure min can never drop below 0?

Thanks,
Michael 

[v-alq-msft]

Hi, @MichaelDoig 

Based on your description, I assume that you want to hide the data points where the measure is a negative number.

You may create a measure like below.

Measure = 
IF(
    SUM('Table'[Val])>=0,
    1,0
)

Then you need to put the measure in the visual level filter to filter the result.

Best Regards

Allan

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

[Anonymous]

I am facing a similar issue.   Doing anamoly detection on shipped quantity.   It's not possible to have a negative shipped quantity, and I have indeed added a measure to the visual to only show >= 0 but I still get negative values in the "expected min value".    There is no way to set the min/max values in the anamoly field itself.  

[amitchandak]

@MichaelDoig , Try changing Sensitivity and expected range.

https://powerbi.microsoft.com/en-us/blog/anomaly-detection-preview/

But as this is preview mode, Good Idea to suggest the changes/observations . Refer this feedback topic - https://community.powerbi.com/t5/Desktop/Share-your-thoughts-on-the-new-Anomaly-detection-feature-pr...

[MichaelDoig]

Cheers for the reply!

I've hidden the name of our measure, but lets just call it sales. 

I hope this helps. 

[amitchandak]

@MichaelDoig , can you share some screenshot.

Or Can you share a sample pbix after removing sensitive data.