Performance Metrics for Classifiers

A video motivating the need for additional classifier performance metrics beyond just accuracy and an introduction to those metrics. The R language (and particularly {tidymodels}) is utilized. Accuracy, precision, recall, and ROC-AUC are introduced and calculated. The video ends with a short discussion of multi-class log-loss as another alternative metric which is focused on the predicted probabilities (the model's certainty in its predictions) rather than focusing solely on the predicted classes. Here's the link to the Quarto Document from the video: github.com/agm...