Anomoly detection setup question

Hello there,
We’ve recently upgraded our ES cluster and now have the ability to use anomoly detection. I’ve read the Open Distro documentation but I’m not sure how to get the anomoly detection working for my use case.

Is it possible to configure a detector so that when it runs it checks the same time for the last two or three weeks? Is that part of the Windows size when setting up the model configuration?

I want to set up a detector which checks the number of invocations to see if we have an service interruption or not (# of invocations). Can I set up an anomoly detector so when it scans it looks 10 minutes in the past and also the same window for the past 2 to 4 weeks?

@jberto78 Thanks for your interest in anomaly detection (AD) plugin. Currently AD plugin supports detecting time series data in realtime way. The model will learn from historical data and it doesn’t support checking the same window for past weeks. But we are happy to help build some new feature to solve your problem. Can you elaborate your use case? Seems you want to analyze trend/weekly pattern?