CSCI 3151 - M12 - Evaluation metrics for regression & classification скачать в хорошем качестве

CSCI 3151 - M12 - Evaluation metrics for regression & classification

This lecture focuses on how we actually judge whether a machine learning model is “good.” We cover the most common evaluation metrics for regression and classification, including MSE, MAE, accuracy, precision, recall, F1, and ROC–AUC, and explain what each one really measures. A central theme is that metrics are not neutral: different choices encode different assumptions, costs, and stakeholder priorities. You’ll see how seemingly strong performance can be misleading under class imbalance, poor threshold choices, or mismatched objectives, and how to reason about these failure modes before deploying a model. Course module page: https://web.cs.dal.ca/~rudzicz/Teaching/CS...