TECHNOLOGY PROFILE · TECH-0228
River
Online machine-learning library for streaming observations.
Reference link only — not reviewed this edition. This profile preserves the prepared catalog’s original descriptions. Advantages depend on the workload; comparisons are editorial, not benchmark results.
Where it has an edge
Updates one sample at a time instead of repeatedly retraining batch models.
Best-fit problems
Drifting streams, incremental classifiers and streaming anomaly detection.
Tradeoffs & cautions
Online learners may sacrifice batch accuracy; drift handling is not automatic correctness.
What to test before choosing
Use leakage-safe cross-validation and compare quality, inference latency and training cost to a simple baseline.
Runtime & hardware
Usually CPU; selected GPU support
Access & licensing
Open-source core; verify extensions
Execution
Local / self-hosted
Lifecycle
Not independently audited
Comparison basis
Editorial problem-fit judgement; no head-to-head benchmark performed
Connected concepts
Alternatives to investigate
scikit-learn · Vowpal Wabbit