Detecting Outliers without ML?

Is it possible to have Kibana show/highlight documents that have the biggest percentage variance in a given numeric field over the selected time slice? Imagine I'm logging hits containing URLs and response times:

URL, Response Time (ms)
www.site1.com,25
www.site1.com,28
www.site1.com,35
www.site2.com,141
www.site2.com,150
www.site3.com,1320
www.site3.com,1350

...how would I have Kibana flag up the third hit to www.site1.com as being anomalous because of the percentage variance in response time, please?

Hi @dar,

You might be able to use Kibana watchers to do something comparable by querying both the max value of response time and the standard deviation and triggering when these two differ too much. Keep in mind though that outlier detection is a complicated topic with a lot of difficult to tune edge cases when done in production settings.

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.