Anomaly Detection Settings (Metric)
A powerful feature that InsightFinder has, and other anomaly detection systems do not, is InsightFinder does not detect anomalies based on thresholds. Therefore, when adjusting the anomaly detection, the system does not need to adjust the threshold of different metrics, but only needs to adjust a sensitivity parameter, which is used to adjust sensitivity of the neural network in anomaly detection. The settings include two sensitivity settings. One for sending mail alerts and the sensitivity for exceptions defaults to “Low”. The other is the sensitivity of the anomaly shown on the UI interface. The default is “Medium Low”.
- Sensitivity: The higher the sensitivity is, the more anomalies will be found.
- Anomaly Duration Threshold: Used to set the duration of the anomaly which will be treated as anomaly events, events of the duration less than this threshold will be ignored.
- Maximum Hint Number: Joyce doesn’t know – Get with Banji
- Enable Dynamic Baseline Detection: Whether dynamic baselining is enabled or not (Default: Checked)
- Dynamic Baseline Violation Factor: The amount that is needed to be calculated to create a baseline threshold anomaly. These can be either lower, higher, or both depending upon the “Detection Type” from the Metric Configuration settings page.
- Enable Period Detection Filter: Filter out metric results that are occur regularly