Classification evaluations Quiz (Hard)

The F1 Score is unsuitable for which of the following scenarios?

What is the difference between micro-averaging and macro-averaging when computing precision, recall, and F1 score in a multi-class classification setting?

How does the choice of threshold impact the performance of a probabilistic classifier? 

In a scenario where a classification model is tested on a highly imbalanced dataset, what phenomenon might occur if accuracy is used as the sole evaluation metric?

In a clinical trial for a new drug, which metric would be more important to maximize to ensure all potential patients are identified, even at the risk of including some healthy individuals?

In which scenario is log-loss a more appropriate metric than AUC-ROC?

Why is Balanced Accuracy particularly useful in evaluating classifiers on imbalanced datasets?

What does “log loss” measure in the context of a probabilistic classifier?

Which of the following statements is true about Log Loss?

What does the Matthews Correlation Coefficient (MCC) measure in the context of a binary classification problem?

In which scenario is the Matthews Correlation Coefficient (MCC) particularly useful?