The final classification score is only useful when you can explain its denominator. This practice set alternates fitted classifiers, a feature-free baseline and different metrics. Each fictional team has fresh records, so compute the requested result from the supplied arrays rather than carrying a number from an earlier task.
Keep the positive class fixed at 1. Precision describes the alerts a team would review, while recall describes the positive cases it would find. F1 balances those two rates, and accuracy counts all correct labels. A model that predicts only the majority class can make accuracy look better than its positive-class coverage. The tasks use the earlier methods without introducing another estimator. Preserve the test partition throughout, including when you fit the simplest baseline.