Anomaly Detection and Pattern Analysis with Categorical Data in OpenSearch

General Feedback
1

Hi everyone,

I’ve been exploring anomaly detection in OpenSearch and noticed that it seems to only support numerical values for anomaly detection. In Elasticsearch, it’s possible to detect anomalies using categorical data.

I have a couple of questions:

  • Is it possible to detect anomalies using categorical fields in OpenSearch?
  • Is there a way to apply pattern analysis similar to how it’s done in Elasticsearch?

Thanks in advance for your help!

2

For “detect anomalies using categorical data”, do you mean the following in Detect anomalies | Documentation

“Categorization: Creates jobs that group log messages into categories and use count or rare functions to detect anomalies within them.”

For “Is there a way to apply pattern analysis similar to how it’s done in Elasticsearch?”, are you still talking about the categorization feature above?

3

Hello, yes I’m referring to this categorization and for pattern analysis it in both categorization and it’s described in: AIOps Labs | Kibana Guide [8.15] | Elastic too.

4

Cool. I’ll include this feature in our roadmap discussion. If you are managed service user, please also submit a PFR request.