A machine learning engineer is evaluating a model designed to identify rare fraudulent transactions. The model correctly identifies 90 out of 100 actual fraudulent transactions but also flags 50 legitimate transactions as fraudulent. There are 10,000 total legitimate transactions. What is the False Positive Rate (FPR) for this model?
- A0.1
- B0.005
- C0.05
- D0.01
Show answer & explanationAnswer & explanation
Correct answer: B. 0.005
The False Positive Rate (FPR) is calculated as False Positives / (False Positives + True Negatives). In this scenario, False Positives (FP) are the 50 legitimate transactions incorrectly flagged as fraudulent. True Negatives (TN) are the legitimate transactions that were correctly identified as legitimate. Since there are 10,000 total legitimate transactions and 50 were falsely flagged, the number of True Negatives is 10,000 - 50 = 9,950. Therefore, FPR = 50 / (50 + 9,950) = 50 / 10,000 = 0.005.
Why the other options are wrong
- A. This is incorrect; it might be the ratio of actual fraudulent transactions missed (10/100) or an arbitrary value.
- C. This might be confused with a percentage or a ratio with an incorrect denominator.
- D. This would be the False Negative Rate if 100 were the total actual fraudulent transactions and 10 were missed (10/100).
False Positive Rate (FPR)
The False Positive Rate (FPR), also known as the fallout, is a metric that quantifies the proportion of negative instances that were incorrectly classified as positive by a model.
- Calculated as False Positives / (False Positives + True Negatives).
- Indicates the rate of Type I errors.
- Relevant in scenarios where misclassifying a negative as positive is costly (e.g., false alarms).
Memory trick: Confusion Matrix: The four corners tell the story.