Choosing the Right Metric

Match the evaluation metric to the problem's cost of errors.

For a cancer-screening model where missing a real case is far worse than a false alarm, which metric should be prioritized?