Which term describes the fraction of primary-target cases with a predicted score that exceeds the predicted score of secondary-target cases?

Concordance refers to a situation in model evaluation where primary-target cases have a higher predicted score compared to secondary-target cases. In this context, when estimating a model's effectiveness, concordance measures how often positive or favorable outcomes (in this case, primary-target cases) receive higher predicted scores than negative or unfavorable outcomes (secondary-target cases). The higher the fraction of primary-target cases with scores exceeding those of secondary-target cases, the better the model's discrimination ability regarding the primary target.

This understanding is fundamental in assessing the predictive quality of classification models. In practice, a high concordance value indicates that the model effectively distinguishes between the two types of cases, thereby serving its purpose in predictive analytics.