Review classification solution statistics

Use the Solution Statistics dashboard to determine if a classification solution has sufficient precision and coverage for each class. Identify classes that require configuration or retraining with a new solution definition filter.

Before you begin

• To ensure optimal dashboard display, enable responsive dashboards or change the default dashboard layout.
• Role required: admin, ml_admin, or ml_report_user

About this task

The Solution Statistics dashboard lists the precision, coverage, and distribution for each class of active solutions. The system uses the classes with the highest number of records when it builds a solution. Note that the number of classes predicted may be less than 50, and can skip a class if there is not enough historical data to build a solution that can predict the class confidently.

Procedure

1. Navigate to All > Predictive Intelligence > Classification > Solution Statistics.
2. From Filter by solution, select the solution whose statistics you want to review.
3. From Filter by version, select the solution version whose statistics you want to review.
4. Click Apply.
   The system updates the dashboard based on the filters selected.

   

5. Identify classes with unwanted combinations of precision, coverage, and distribution values.
   For example, identify classes that have low precision or coverage but a high distribution.
6. Identify any missing classes you want the model to include.
   For example, identify any missing incident categories from the Incident classification solution.

What to do next

If you're satisfied with the solution you've reviewed, it will already be active and ready to use if it's the latest version. If you're not satisfied, you can choose a different version of the solution and make it active. You can also tune and retune the solution by configuring the class precision and coverage to use acceptable values.