A glass-studio “bubble / no-bubble” set is 190 no-bubble tiles and 10 bubble tiles. A model calls every tile no-bubble: 0 true bubbles found, 10 bubbles missed, 0 false bubble calls, 190 true no-bubble calls. Why does 95% accuracy mislead, and which named metric should the tester also compute?
Select an answer to reveal the explanation.
Short Explanation
A model that calls every tile no-bubble scores 95 percent accuracy while recall on the rare bubble class is 0 percent. Accuracy hides that rare-class failure. Compute recall (sensitivity) on the bubble class. Do not invent AUC, ROC, or specificity for this LO.
Full Explanation
Accuracy is (0 + 190) / 200 = 95% while recall on the rare bubble class is 0 / 10 = 0%. Testers must interpret named metrics on an imbalanced matrix. AUC, ROC, and specificity are not required names in this LO.