This can be a common occurance for data is too volatile.

Volitile data can be data that is not predictable or has too little data to draw conclusions.

An example of volatile data

But! There are some things you can do to help:

Tweak the AI thresholds

By editing an alert (you can do this from the alerts navigation page), you can sweak the AI sensitivity to reduce the number of alerts firing.

How to lessen the alert sensitivity

For data that is particulary volatile, you can even move the slider to 0 for a maximum leniency when evaluating if an anomaly has occured.

This will increase the upper and lower bounds that your data can exist in before an alert is fired.



Set Silent times

On the alert editing page, you can click 'Global settings' to update settings for all alerts



You can then set silent times. This means, even if an anomaly is detected, we will not send alerts within the times are are silenced.
The following is a typical configuration:



Switch to manual

Sometimes the data is still too volitile. In this case, turning off our AI alerting, and moving to a manual threshold can be better.

This is a typical setup, where we will only fire an alert if the upper and lower red lines are crossed.

Was this article helpful?
Cancel
Thank you!