Configure class precision and coverage values Configure how often a solution predicts specific output field values by selecting the class precision and coverage. Before you begin Role required: admin or ml_admin. Train the solution definition whose output field values you want to configure. About this task The system creates a class record for each output field value that it can predict. Each class record includes a list of possible precision and coverage combinations to choose from. By default, solutions use the highest combination of precision and coverage available. You can select another combination to refine predictions based on acceptable precision and coverage values. Procedure Navigate to Agent Intelligence > Solutions. The system displays the list of available solutions. Select the solution whose classes you want to configure. This solution must have a State of Solution Complete. The system displays the Solution record. From the Class Confidence related list, select the class you want to configure. The solution only lists output field values for which it can make predictions. If the output field value is missing from this list, update the solution definition filter to provide more data for this output field value, and retrain the solution. The system displays the Class Confidence record. Review the precision and coverage combinations available from the Precision Coverage Lookups embedded list. Select the check box for the precision and coverage combination you want to use to make predictions of this class. You can only select one check box. Some combinations produce special prediction results. Table 1. Special prediction combinations Prediction result Precision Coverage Never include class in predictions 100 0 Always include class in predictions 0 100 From the Actions on selected rows control, select Apply Values. The system displays a Precision / Coverage Setting confirmation window. Click OK to confirm the change or Cancel to discard it. What to do nextTest predictions of this class to verify the system produces acceptable results.