Automated Retraining with Azure Logic App and Event Grid for ML Model Data Drift Detection
Question
You have deployed your ML model and you want to trigger the retraining of it automatically when data drift is detected by an automatic dataset monitor.
In this case, an existing Azure Data Factory pipeline must be launched in order to retrain the model.
Azure Event Grid, in combination with Azure Logic App should be used.
Create a Logic App Select trigger: When an Event Grid resource event occurs Select trigger: When a message is received in a Service Bus Queue Set the Event Type to DatasetDriftDetected Set the Event Type to RunStatusChanged Create an Azure Data Factory pipeline run step Which of the above actions should be executed to achieve your goal?
Answers
Explanations
Click on the arrows to vote for the correct answer
A. B. C. D.Correct Answer: A.
Option A is CORRECT because a Logic App must be created, triggered by an Event Grid event of type DatesetDriftDetected, the another step to launch the existing AFF pipeline must be added to the Logic App flow.
Option B is incorrect because the trigger of the Logic App is an Event Grid event, rather than a message in a Service Bus Queue.
The event type is incorrect, too.
Option C is incorrect because the event type DatasetDriftDetected must be set.
Option D is incorrect because it is an Event Grid event which triggers the execution of the Logic App.
Diagram:
/answer/imgckeditor_6.png)
Reference:
The correct answer is B. 1, 3, 4, 5.
To trigger the retraining of a machine learning model when data drift is detected, we need to set up an automatic dataset monitor that continuously checks the data and detects any changes in the data distribution, and triggers a retraining of the machine learning model when necessary.
Azure Event Grid is a service that enables the creation of event-driven architectures by providing a centralized event routing service that allows different Azure services to publish and subscribe to events. Azure Logic App is a workflow automation tool that helps us to create automated workflows and integrate with different Azure services.
The steps to achieve the goal are as follows:
Create a Logic App: This is the first step. We need to create a new Logic App in the Azure portal.
Select trigger: When an Event Grid resource event occurs: We need to add a trigger to our Logic App that listens to events from Event Grid. This trigger will be used to start our Logic App workflow whenever an Event Grid resource event occurs.
Set the Event Type to DatasetDriftDetected: We need to configure the Event Grid trigger to listen to the DatasetDriftDetected event type. This event will be fired whenever a dataset drift is detected by our automatic dataset monitor.
Create an Azure Data Factory pipeline run step: After the Event Grid trigger is fired, we need to create a step that will start an Azure Data Factory pipeline run. This pipeline will retrain our machine learning model using the updated data.
Set the Event Type to RunStatusChanged: We also need to configure the Event Grid trigger to listen to the RunStatusChanged event type. This event will be fired whenever the status of the pipeline run changes.
By combining Azure Event Grid and Azure Logic App, we can create a workflow that listens to dataset drift events, triggers the retraining of our machine learning model, and monitors the status of the pipeline run.
Option A is incorrect because it includes steps 2 and 6, which are not required.
Option C is incorrect because it includes steps 2 and 6, which are not required.
Option D is incorrect because it includes steps 2 and 5, which are not sufficient to achieve the goal. We also need to create an Azure Data Factory pipeline run step.