Anomaly Detection

Home > Select Project > Alert > Event Configuration > Anomaly detection tab

Anomaly detection identifies abnormal rising or falling patterns in metric data and triggers events when movements that differ from the usual behavior are detected.

Screen layout

The Anomaly detection tab provides two sections, stacked vertically, that use different judgment methods. Previously the two features were separated into tabs with the same name, but they have been merged into one.

Section	Description
Automatic Baseline	The collection server calculates the threshold for each metric on its own. You only set the target and the duration condition
Adaptive Range	You directly specify the category, field, filter, and the sensitivity of rising/falling patterns

The two section titles are feature names defined by the product, so the same notation is used regardless of language.

Each section can be collapsed and expanded. Collapsing a section keeps the search term and the table sort state. The count next to the section title shows the number of rules remaining after the search filter.

Default Options

• Search: Search for specific events by selecting an event name or an event recipient tag.

• Add Alert Policy: Add a new anomaly detection event.

The wording used in both sections has been unified as follows.

• The button that adds an event is labeled Add Alert Policy in both sections.

• The Adaptive Range section also has a search box and uses the same guidance text as the Automatic Baseline section.

• When you delete a rule, both sections show the "Are you sure you want to delete?" confirmation window.

Table | Anomaly Detection Event List Layout

Item	Description
Event Message	Message sent when an event is triggered
Category	Unit used to classify metric data and serve as the basis for metric event configuration
Field	Metric field used to detect anomalies
Filter	Conditions that limit the data range subject to anomaly detection
Rising/Falling Pattern	Sensitivity of rising or falling patterns
Event Recipient Tag	Users or groups that receive notifications for the event
	Edit event settings
	Delete event
	Event enablement status - Enabled: Event is triggered when conditions are met - Disabled: Event does not run and conditions are not evaluated

The rule list in the Adaptive Range section has changed from cards to a table. In addition to the items above, the following columns are shown.

Item	Description
No.	Sequence number in the list
Level	Severity of the event (Critical, Warning, Info)
Enabled	Toggle that turns the rule on or off

The table in the Automatic Baseline section uses the same structure; only the middle column differs with that section's items.

Event Configuration

You can configure alert notifications to be sent when metric values rise or fall in a way that differs from their usual behavior.

Add Event Rules

1. Click the [ Add Alert Policy ] button at the top right of the Adaptive Range section.

2. Configure Event Basic Information, Metrics, Anomaly Detection, and Additional Information in order.

Event Basic Information

1. Select the event level (Critical, Warning, Info).

   • This determines the event priority and alert level.

2. Enter the alert message to be delivered when the event is triggered.

   • You can use variables in the message to include actual values at the time the event occurs.
   Note

   Available Variables

   • $pcode: Project code
   • $category: Category
   • $field: Field
   • $value: Triggered value
   • $oname: Event target

Metrics

1. Select a Category.

   • This category serves as the baseline for the metric data used to detect anomalies.

2. Select the Field to be used for the event trigger conditions. Multiple selections are supported.

3. Select filters that define the event condition targets used to narrow the event scope. Multiple selections are supported.

Anomaly Detection

1. Select the sensitivity levels for Rising Pattern and Falling Pattern.

2. Use the toggle buttons to enable or disable each pattern.

   • To detect only rising patterns: Rising Pattern ON, Falling Pattern OFF

   • To detect only falling patterns: Falling Pattern ON, Rising Pattern OFF

Additional Information

Event Active Time

Configure tags to control when an event runs, such as working hours, non-working hours, or maintenance windows. If no tag is set, the event runs continuously for 24 hours while it is enabled.

1. Click + Add.

2. Under Event Active Time, click + Create New Tag.

3. In Create Event Active Time Tag, select a Tag Name, Days, Time, and Color, then click the [ Apply ] button.

4. Created tags appear in the tag list and can be applied by selecting the checkbox.

5. To edit or delete a tag, click the icon and modify or remove it in Edit Event Active Time Tag.

Caution

When a tag is deleted, it is removed from all users to whom it is applied. However, tags that are currently in use by event rules cannot be deleted.

Event Recipient Tags

When event recipient tags are configured, notifications are sent to project members who have those tags. If no recipient tag is set, notifications are sent to all users in the project.

1. Click + Add Tag or the + icon.

2. Click + Create New Tag at the bottom of the Event Recipient Tags window.

3. In the Create Tag window, enter a tag name, select a color, and click the [ Create Tag ] button.

4. To edit or delete a created tag, click the icon in the tag list.

5. Select the desired tag from the Tag List in the Event Recipient Tags window to apply it.

Tip

When configuring events, selecting event recipient tags allows notifications to be sent to project members and third-party plugins associated with those tags. You can assign tags to project members and third-party plugins separately via Alert > Notifications.

Edit Event

1. Go to the Alert > Event Configuration > Anomaly detection tab.

2. In the Anomaly Detection list, click the icon on the right of the event you want to edit.

3. In the Anomaly Detection window, update the options and click the [ Save ] button.

Delete Event

1. Go to the Alert > Event Configuration > Anomaly detection tab.

2. In the Anomaly Detection list, click the icon on the right of the event you want to delete.

3. In the delete confirmation window, click the [ Delete ] button.