Manage model monitor

This page describes how to update and delete model monitors.

Update a model monitor

Update a model monitor to modify the default configuration for monitoring jobs. You can't change the model version that a model monitor is associated with. Instead, create a new model monitor.

Console

In the Google Cloud console, go to the Monitoringpage.

Go to Monitoring
Click the model monitor to go to the Monitor detailspage.
Click Edit configurationto update the model monitor.
Modify your schema, notifications, training data source, and objectives.
Click Save.

Python SDK

You can update the following specifications:

Display name
Training dataset
Model schema
Default monitoring objectives
Default monitoring output specification
Default monitoring notification specification
Default explanation specification for feature attribution

The following example updates the model schema:

 monitor_2 
 . 
 update 
 ( 
 model_monitoring_schema 
 = 
 MODEL_MONITORING_SCHEMA 
 )

Delete a model monitor

Remove model monitors to prevent authorized users from inadvertently running monitoring jobs on unused model monitors. When you delete a model monitor, any associated monitoring data is deleted except for metrics and statistics exported by Monitoring, which remain in Cloud Monitoring and in your Cloud Storage.

Console

In the Google Cloud console, go to the Monitoringpage.

Go to Monitoring
Click the model monitor to go to the Monitor detailspage.
Click Edit configurationto update the model monitor.

Python SDK

 from 
  
 google.cloud.aiplatform.private_preview.centralized_model_monitoring 
 import 
 model_monitor 
 my_model_monitor 
 . 
 delete 
 ( 
 force 
 = 
 true 
 )