An ML engineer is using Amazon QuickSight anomaly detection to detect very high or very low machine operating temperatures compared to normal. The ML engineer sets the Severity parameter to Low and…

Question

Options

• AIncreased anomaly identification frequency and increased recall
• BDecreased anomaly identification frequency and decreased recall
• CIncreased anomaly identification frequency and decreased recall
• DDecreased anomaly identification frequency and increased recall

• A
  10% (6)
• B
  81% (47)
• C
  3% (2)
• D
  5% (3)

Explanation

With Direction = All, QuickSight anomaly detection flags temperature readings that are either significantly higher OR lower than expected. Changing Direction to 'Lower than expected' restricts detection to only anomalies where the temperature is below the expected range. This has two consequences: (1) Anomaly identification frequency decreases because only one direction of deviation (low) is now flagged instead of both directions. (2) Recall decreases because the model now misses all true high-temperature anomalies - they exist in the data but are no longer surfaced. Recall measures the fraction of actual anomalies that are detected; ignoring one direction means more true anomalies go undetected, lowering recall.

No community discussion yet for this question.