Batch Prediction Pattern
← Back to ML System Design Patterns
Pre-compute predictions offline, serve from cache/DB. Low latency at serving time but stale predictions. Good for recommendation emails, daily reports, and non-time-sensitive predictions.
Related
- Real-Time Prediction Pattern (fresh but slower)
- Batch Inference (implementation)