[熱身賽第二輪] 評估分類系統(2) [Warm-up Round 2] Evaluating Classification Systems (2)

Big Concept: Evaluating Classification Systems Topic: Why 99% Accuracy Can Still Be a Bad Model Bloom's Taxonomy: understand, apply, analyze, evaluate, remember Learning Goal: Use a confusion matrix to compare accuracy, precision, and recall, and judge whether a 99% accurate classifier is acceptable for spam filtering, fraud detection, or disease screening. Key Learning Points: confusion matrix; false positives vs. false negatives; precision; recall; task-dependent error cost