Edit

Share via


Explore anomaly detection events in Fabric Real-Time hub (Preview)

Anomaly detection in Real-Time hub helps you automatically identify unusual patterns or outliers in your Eventhouse tables by applying recommended models based on the structure of your data. It allows you to visualize anomalies, adjust detection sensitivity, and set up continuous monitoring with automated alerts or actions.

Important

This feature is in preview.

View anomaly detection events detail page

  1. In Real-Time hub, select Fabric events under the Subscribe to category.

    Screenshot that shows the Fabric events page in Real-Time hub.

  2. Select Anomaly detection events from the list.

    Screenshot that shows the selection of anomaly detection events on the Fabric events page.

  3. In the Anomaly detection events page, select an event to view its details or select Set alert to set an alert for the event.

    Screenshot that shows the detail page for OneLake events.

Schemas

An event has the following top-level data:

Property Type Description Example
source string Identifies the context in which an event happened. <tenant-id>
subject string Identifies the subject of the event in the context of the event producer. /workspaces/<workspace-id>/items/<ad-item-id>/configuration/<configuration-id>
type string One of the registered event types for this event source. Microsoft.Fabric.AnomalyEvents.AnomalyDetected
time timestamp The time the event is generated based on the provider's UTC time. 2017-06-26T18:41:00.9584103Z
id string Unique identifier for the event. <Required-GUID>
specversion string The version of the Cloud Event spec. 1.0
dataschemaversion string The version of the data schema. 1.0
data object Event data. See the next table for details.

The data object has the following properties:

Property Type Description Example
analysisType string Type of analysis performed. univariate
confidenceScore number Confidence score of the anomaly detection. 0.95
timeStampAttributeName string Name of the timestamp attribute used in the analysis. StartTime
timeStampAttributeValue string Value of the timestamp attribute at the time of the anomaly detection. 2017-06-26T18:41:00.9584103Z
univariate object Contains details of the univariate analysis. See the next table for details.
customAttributes object Contains any custom attributes associated with the anomaly detection. See the next table for details.

The univariate object has the following properties:

Property Type Description Example
instanceIdAttributeNames string Name of the instance ID attribute used in the analysis. SkuId
instanceIdAttributeValues string Value of the instance ID attribute at the time of the anomaly detection. sku-12345
monitoredAttributeName string Name of the monitored attribute used in the analysis. Temperature
monitoredAttributeValue string Value of the monitored attribute at the time of the anomaly detection. 90

The customAttributes object has the following properties:

Property Type Description Example
SkuId string Unique identifier for the SKU. sku-12345
Temperature number Temperature value at the time of the anomaly detection. 90
Humidity number Humidity value at the time of the anomaly detection. 50
Location string Location where the anomaly was detected. Kitchen
Status string Status of the anomaly detection. Normal