Configure Anomaly Detection Model Training

To adjust anomaly detection settings, edit the ad-anomaly.yml configuration file on the Pulse node.

File path: /data01/acceldata/config/docker/addons/ad-anomaly.yml

Example Configuration:

Bash
    
 
THRESHOLD=5DAYS_OF_DATA_TO_FETCH=30INSTANT_QUERY_STEP=5mMONGO_URI=<uri>MONGO_ENCRYPTED=trueMONGO_SECRET=<secret>
Copy

The configuration details are as follows.

Parameter	Description	Default Value	Notes
THRESHOLD	The Threshold defines how much deviation from the upper and lower thresholds is acceptable.	5 %	The default value is used if not explicitly set.
DAYS_OF_DATA_TO_FETCH	Specifies the time range for which data is retrieved.	30 days	The default value ensures 30 days of historical data is fetched.
INSTANT_QUERY_STEP	Defines the time interval (resolution) between consecutive data points in the query result.	5m	Default interval between data points.
MONGO_URI	The connection URI for MongoDB.	No default	Must be explicitly configured with your MongoDB connection string.
MONGO_ENCRYPTED	Indicates whether the MongoDB connection is encrypted.	true	Default assumes secure (SSL/TLS) connection.
MONGO_SECRET	Stores the encryption secret used to connect securely to MongoDB.	No default	Must be provided when encryption is enabled.

After updating the configuration:

By default, Pulse runs a range query that fetches data for the past 30 days, returning data points at 5-minute intervals.

Each data point represents a snapshot taken every 5 minutes across the 30-day time range.

Last updated on

Was this page helpful?