Notifications Providers
Last updated
Last updated
Copyright © 2023 Arize AI, Inc
Once you've set your monitors, learn how to decipher their status and set up an alerting system.
Monitors are continuously evaluated according to each monitor's configurations -- an evaluation will result in one of the following statuses for a monitor, ordered by severity:
When viewing the health of your model for a monitor type (drift, performance, or data quality), the status shown will be based on the highest severity of any of the model's monitors for that type. For example, if you are monitoring your model with several performance monitors, such as Accuracy and F1-score, and at least one of them has a "Triggered" status, the status shown for the model's performance will be "Triggered".
Once you set up your monitors, you’ll want to be alerted when your model deviates from its expected ranges.
Arize offers alerting integrations for alerting tools and methods. Send an alert via email, slack, OpsGenie, and PagerDuty. Within these tools, you can add configurations to edit your alerting cadence, severity, and alert grouping.
Schedule downtime to temporarily pause monitor evaluations for a given timeframe. During downtime, your monitor will not assess the metric or create new alerts. Use downtime to silence your monitors during expected intervals when your metric or alerts would not provide meaningful insights, such as during joins, expected low activity times, or weekends.
You can schedule downtime in the monitor configuration panel. By default, downtime will be off:
To enable downtime, edit your monitor and turn on the Schedule downtime toggle under Define the Alerting. In this section, you can customize the time window during which the monitor should not be evaluated. You also need to specify the start date and set the cadence for the downtime schedule.
In the example above, our monitor will not evaluate from 8PM to 9PM daily starting on 10/11.
Healthy
No action is needed!
No Data
This shows when the monitor does not have recent data in the evaluation window. This can be resolved by extending the evaluation window or uploading new data to Arize. All live models should be configured to continuously send your model’s latest data in batch or real-time operations to keep monitors up-to-date.
Triggered
The Arize platform offers drift and performance troubleshooting to root cause model issues and pinpoint specific feature values or cohorts of data where the model performing poorly.
