Which term describes the proportion of secondary outcome cases that rank highly in predicted cases?

The term that best describes the proportion of secondary outcome cases that rank highly in predicted cases is "Cumulative lift." This concept is utilized in predictive modeling to evaluate the effectiveness of a model in identifying relevant cases.

Cumulative lift quantifies the improvement in prediction compared to random selection. It measures how much better the model performs with respect to identifying the target outcome. For instance, if a model can accurately rank a higher proportion of positive cases within the top ranks of predicted observations, this speaks to the model’s ability to effectively pull out cases of interest from the dataset.

Unlike sensitivity, which assesses the model's ability to correctly identify positive cases among actual positives, cumulative lift focuses on the ranking and proportion of predicted cases within a certain threshold, making it a more suitable metric for the context of proportionate ranking. The response typically refers to the variable or outcome that the model attempts to predict, while the false positive fraction measures the proportion of incorrectly predicted positive cases among actual negatives; neither addresses the ranking aspect directly. Thus, cumulative lift is the term that best captures the relationship described in the question.