You manage an Azure Machine Learning workspace. You must set up an event-driven process to trigger a retraining pipeline. You need to configure an Azure service that will trigger a retraining pipeline

Azure Machine Learning publishes data drift detection results as events directly to Azure Event Grid. Event Grid then routes these events to a configured handler - such as a pipeline trigger - making it the correct service to connect data drift detection to a retraining pipeline in a fully event-driven way.

Azure Functions can serve as an event handler that runs pipeline code, but it depends on Event Grid to detect and deliver the drift event - it is not itself the triggering service.

Event Hubs is designed for high-throughput telemetry and data streaming ingestion, not for routing discrete Azure ML workspace lifecycle events like data drift notifications.

Logic Apps can orchestrate multi-step workflows but is not the primary service that receives and routes Azure ML data drift events; it would still require Event Grid as the event source.