Viewing data for a deployment
Select a deployment from the dashboard to see monitoring data for that deployment. The heading displays information about the deployed model, such as the Model ID, Description, and Evaluation date fields.
The Evaluation window displays details for the fairness, quality, and drift monitors. Depending on the type of model, either pre-production or production, the Tests run, Tests passed, and Tests failed fields display the following results:
- For a pre-production model deployment, the results of your test run are displayed
- For a production model deployment, the results of the most-recent regularly scheduled hourly fairness and quality, and the most-recent 3-hourly drift evaluations are displayed.
Click one of the tiles, for example the Fairness tile to view the fairness timeline view:
Because the algorithm checks are run only every hour or every three hours, there are also links provided to check fairness, drift, and quality on-demand. From the Schedule pane, you can click the Evaluate fairness now, Evaluate quality now, or Evaluate drift now links to make an immediate check of your data.
Next, click the chart and move the marker across the chart to see statistics for an individual hour.
You can see the following values:
- Fairness: See which groups met their set thresholds for approval.
- Quality: Depending on the type of model, you can view a list of quality metrics. Alerts display when a metric is not within the configured threshold.
- Drift: Drift provides both the drop in accuracy and the drop in data consistency based on the training data.
- Analytics: Click the Predictions by Confidence view to see the confidence percentages. You can also customize the analytics chart by clicking the Chart Builder option.
View transaction details, where you can see how the prediction was determined and also, how you might change the outcome by adjusting values. For more information, see Explaining transactions.