Splunk Machine Learning Toolkit contains Detect Categorical Outliers method. Splunk documentation mentions that The Detect Categorical Outliers assistant uses the probabilistic measures algorithm.
I am trying to understand how it works in more detail. Where can i find more information about internals of this method?
Hi @Kiril123,
The main command being used in the "Detect Catgorical Outliers" assistant is the anomalydetection command.
You can read more about it on the docs page, but to summarize, it uses log probabilities, interquartile ranges, as well as gaussian assumptions (depending on the mode you use).
Hello,
I am still looking for a technical solution to this. The given links in the answers above give some gist around it but I am not pretty clear what is happening to calculate the categorical outliers.
In one of the answers: "it uses log probabilities, interquartile ranges, as well as gaussian assumptions". If I want to implement by my self, how should I go forward? How the log probabilities are applied?
Will anyone please help?
Thank you very much.
Hi @Kiril123,
The main command being used in the "Detect Catgorical Outliers" assistant is the anomalydetection command.
You can read more about it on the docs page, but to summarize, it uses log probabilities, interquartile ranges, as well as gaussian assumptions (depending on the mode you use).
Hello,
Detect Categorical Outliers assistant is based on "anomalydetection" command.
Its documentation can be found here:
http://docs.splunk.com/Documentation/SplunkCloud/6.6.3/SearchReference/Anomalydetection
Hopefully, that answers your question.
the Detect Categorical Outliers uses "anomalydetection" which is a splunk search command. You can find details on this page:
http://docs.splunk.com/Documentation/Splunk/7.0.0/SearchReference/Anomalydetection.
You can also click the "Show SPL" button to find out the underlying SPL that constructs the query.