AI-900 Practice Question: Describe fundamental principles of machine learning on Azure

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  Accuracy

  Why it's wrong here

  Accuracy is the ratio of correct predictions to all predictions, i.e., (true positives + true negatives) divided by the total number of cases. In a rare disease scenario with, say, a 1% prevalence, a model that simply predicts 'no disease' for every patient achieves 99% accuracy while detecting zero actual cases. Because the majority class (healthy patients) dominates the metric, accuracy remains deceptively high and completely obscures the model's failure to identify any positive instances.

• ✗

  Precision

  Why it's wrong here

  Precision measures the proportion of positive predictions that are actually correct, defined as true positives divided by (true positives + false positives). In this case, the model never outputs a positive prediction, so both true positives and false positives are zero, rendering precision undefined (or assigned 0 by convention). Precision evaluates the quality of the model's positive alerts, not its coverage of real cases, so it cannot expose the fact that all disease patients are being missed. The failure to detect positives is literally a recall problem, not a precision problem.

• ✓

  Recall

  Why this is correct

  Recall, also called sensitivity, is computed as true positives divided by the total number of actual positive cases (true positives + false negatives). Here, the model predicts no positive cases, so the true positive count is zero, yielding a recall of 0%. This directly indicates that 100% of actual disease cases are missed, which is the exact failure mode in this screening scenario. For rare disease detection, high recall is essential to ensure that no affected patient is overlooked, even if it means accepting more false positives.

• ✗

  F1 score

  Why it's wrong here

  The F1 score is the harmonic mean of precision and recall, calculated as 2 × (precision × recall) / (precision + recall). Because recall is 0 in this model, the F1 score collapses to 0 no matter what precision happens to be, so it does convey that performance is poor. However, it cannot distinguish whether the problem stems from precision or recall; the metric is a single aggregate that hides the specific failure mode. A 0 F1 is a symptom, not a diagnostic — it is recall alone that cleanly identifies the model's total inability to detect any disease cases.

Go deeper

Related to this question