PRODUCTION SCENARIO
A payments company has 400,000 past transactions that its investigations team already marked as fraud or legitimate. It wants a model that flags new transactions as they arrive.
Which machine learning approach matches this data and goal?
Answering here is anonymous. Nothing is saved unless you sign in.
Show answer and explanation
Answer: Supervised learning, because the historical outcomes are already labeled
Supervised learning trains on labeled data where the correct output is already known, which is what the marked transactions provide. Unsupervised learning works on unlabeled data, and reinforcement learning learns from rewards rather than from a labeled history.
Official Source: https://cloud.google.com/learn/what-is-machine-learning ↗