We configured smart detector in our alarm template.
Now we see a Variance Change (decrease) alarm with a root time of “Yesterday 15:00UTC”, however the alarm only appeared in the alarm list after 00:00UTC. We would like to better understand how smart detector works, does the variance change needs a full day of mean to detect a variance change?
Hi Dennis,
As explained in this tutorial (Automatically detect anomalies with DataMiner), it is recommended that at least one week of data is present, and that both real time and average trending are enabled on the target parameters for the most accurate anomaly detection results.
It appears that in this case, the engine was fast enough to already understand the variance as an anomaly after only a few hours.
You can find more details in the official documentation.
- Advanced analytics in trend graphs | DataMiner Docs
- Advanced analytics features in the Alarm Console | DataMiner Docs
Regards,