Question

A logistics company has trained a demand-forecasting model. Every night, they need to generate delivery volume predictions for all 5,000 warehouses at once, using data collected during the day. There is no need for instant responses; the results simply must be ready before the morning shift. Which type of inference best fits this requirement?

Accepted Answer

Batch inference that processes the full dataset on a schedule

Answer

Real-time inference through a persistent, always-on endpoint

Answer

Continuous online training that updates the model with each new record

Answer

Reinforcement learning to let the model self-correct predictions overnight