Compute probability an account is fake
Company: Meta
Role: Data Scientist
Category: Statistics & Math
Difficulty: easy
Interview Round: Technical Screen
A platform uses an automated classifier to flag potentially **fake accounts**.
Assume:
- Base rate: 2% of accounts are fake.
- The classifier flags a fake account with probability 90% (true positive rate).
- The classifier flags a real account with probability 5% (false positive rate).
If an account is flagged, what is the probability it is actually fake? Show your work and state any assumptions.
Quick Answer: This question evaluates understanding of conditional probability and Bayesian reasoning, specifically interpreting base rates alongside true positive and false positive rates to infer posterior probabilities for a flagged account.