You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem?
AD currently supports suppression rules at the feature level, allowing users to define threshold-based rules so that anomalies are ignored by minor fluctuations in anomaly detection. These rules can be configured based on either relative percentage absolute threshold value difference between the expected and actual values.
However, suppression rules today apply globally to all entities within a categorical field. For example, if a user configures anomaly detection for a region field, suppression rules will apply across all regions being monitored:
There is no way to apply suppression rules only to specific category values, such as only suppressing anomalies for us-west-* if the categorical field was region while keeping anomaly detection unchanged for other regions.
What solution would you like?
We propose adding support for category-specific suppression rules in anomaly detection, allowing users to define suppression rules for specific entities within a categorical field instead of applying them to all entities.
Entities can be chosen by regex or some other type of criteria.
Users should be able to set multiple regex pattern for the same categorical field
We will need both a solution on the backend to add this capability and a frontend solution as well to make it easy for users to configure category specific rules.
Frontend Solution
In order to make category specific suppression rules easy to configure we will need to adjust our anomaly-detection dashboards plugins.
Options:
Extend the Suppression Rules Section
Add an optional dropdown or text field that allows users to specify which category values the suppression rules should apply to.
Support explicit entity selection (e.g., us-west-1, us-east-1) and regex capabilities (e.g., us-west-*).
Change the categorical field section
When a categorical field is enabled, display an additional selector to specify which category values to apply the suppression rule to.
Users should be able to:
Define suppression rules for all entities (default behavior).
Define suppression rules only for specific entities (new feature).
Define suppression rules using regex
The text was updated successfully, but these errors were encountered:
We define feature first and then categorical field. "Extend the Suppression Rules Section" option would force people to go back and forth in UX. Not good.
Categorical field is for splitting time series. It is not optimal to mix rules and categorical together.
How about a separate rule section? You can preload existing features and categorical field. They can add rules on top of features and categorical fields?
We define feature first and then categorical field. "Extend the Suppression Rules Section" option would force people to go back and forth in UX. Not good.
Categorical field is for splitting time series. It is not optimal to mix rules and categorical together.
How about a separate rule section? You can preload existing features and categorical field. They can add rules on top of features and categorical fields?
That's a fair point, adding an additional section can work also, I see there being two ways for this. One way is to create all the rules there along with the regex for each entities these rules will apply too, the other option is still create the rules as part of the feature definition but show them as read only in this new section of the form along with displaying the categorical fields in order to let users define the regex rules there
Is your feature request related to a problem?
AD currently supports suppression rules at the feature level, allowing users to define threshold-based rules so that anomalies are ignored by minor fluctuations in anomaly detection. These rules can be configured based on either relative percentage absolute threshold value difference between the expected and actual values.
However, suppression rules today apply globally to all entities within a categorical field. For example, if a user configures anomaly detection for a region field, suppression rules will apply across all regions being monitored:
region=us-west-1
region=us-west-2
region=us-east-1
region=eu-west-3
...etc.
There is no way to apply suppression rules only to specific category values, such as only suppressing anomalies for us-west-* if the categorical field was
region
while keeping anomaly detection unchanged for other regions.What solution would you like?
We propose adding support for category-specific suppression rules in anomaly detection, allowing users to define suppression rules for specific entities within a categorical field instead of applying them to all entities.
We will need both a solution on the backend to add this capability and a frontend solution as well to make it easy for users to configure category specific rules.
Frontend Solution
In order to make category specific suppression rules easy to configure we will need to adjust our anomaly-detection dashboards plugins.
Options:
Add an optional dropdown or text field that allows users to specify which category values the suppression rules should apply to.
When a categorical field is enabled, display an additional selector to specify which category values to apply the suppression rule to.
Users should be able to:
The text was updated successfully, but these errors were encountered: