You are comparing model options for a supervised learning task and want to explain one common ensemble method clearly.
How would you explain random forests?