A company has a binary classification model in production. An ML engineer needs to develop a new version of the model. The new model version must maximize correct predictions of positive labels and ne

Question

Options

• AAccuracy
• BPrecision
• CRecall
• DSpecificity

• A
  87% (26)
• B
  3% (1)
• C
  3% (1)
• D
  7% (2)

Explanation

Accuracy is the correct choice because it is the only metric that accounts for correct predictions across both classes: it is calculated as (TP + TN) / (TP + TN + FP + FN), meaning it rewards the model for getting positive and negative labels right. When the goal is to maximize overall correctness, accuracy directly represents that objective.

Precision (B) is wrong because it only measures how many predicted positives are actually positive - it says nothing about how well the model handles negative labels. Recall (C) is wrong for the opposite reason: it only measures how many actual positives are caught, completely ignoring negative label performance. Specificity (D) is the mirror of recall - it only measures correct identification of actual negatives, so it ignores positive label performance.

Memory tip: Think of accuracy as the "both sides" metric - its numerator adds True Positives and True Negatives together. If you see a question asking about maximizing correctness on both positive and negative labels simultaneously, that's the signature of accuracy.

No community discussion yet for this question.