Creating a custom attribute to filter the underlying data with a 95% distribution | The place for Zendesk users to come together and share
Skip to main content
R
Robert37
June 25, 2021
Question

Creating a custom attribute to filter the underlying data with a 95% distribution

  • June 25, 2021
  • 1 reply
  • 28 views

This is the custom calculated attribute that I'm trying to use to filter out the data set within a range of 2 standard deviations from the mean (in other words, a 95% distribution):

IF VALUE(First reply time - Business hours (min)) > (AVG(First reply time - Business hours (min)) - (2*STD_DEV(First reply time - Business hours (min))))
AND VALUE(First reply time - Business hours (min)) > (AVG(First reply time - Business hours (min)) - (2*STD_DEV(First reply time - Business hours (min))))
THEN [Ticket ID] ENDIF

 

Out of this specific context, how do we use statistical distributions to slice our data?

    1 reply

    G
    Gab12Employee
    Employee
    July 28, 2021

    Hi Robert,

    It appears that, unfortunately, it is currently not possible to build a single attribute or a single metric that could pull up tickets within the standard deviation. I made some tests with custom metrics and attributes using a similar formula as yours, and I found that the results do not seem to align when you drill into the values. You'll be able to show the value of standard deviation from the pool of data if you use the aggregator STD_DEV with the metric, but it looks like STD_DEV has limited applications within Explore at this time compared to other aggregators.

    Have you tried exploring using fixed calculated metrics to show the range of normal distribution? Maybe you can try using two fixed metrics to calculate: 1) Average FRT minus 2 standard deviations; and, 2) Average FRT plus 2 standard deviations. You can then reference these fixed metrics in another custom metric or a standard calculated attribute in the query. We could not guarantee that this would work for your use case, but it may be worth exploring.

    Sorry again if Explore falls short when it comes to calculations involving standard deviations. Hopefully, in future releases, Explore would allow for more applications for STD_DEV in queries.

    Thank you, Robert!

    Terms & ConditionsAccessibility statement