You are working on a platform that runs large fleets of virtual machines and containers in the cloud. Capacity is often overprovisioned to avoid incidents, but that drives up cost and lowers hardware utilization. You want to use machine learning to predict demand and recommend actions such as right-sizing, autoscaling, or workload placement.
How would you optimize cloud infrastructure efficiency using machine learning?