Moving from Offline to Online Machine Learning with River
summary
Claims River is a practical, pure‑Python library that lets models predict one / learn one on streaming data, lowering CPU and memory by updating on each observation and adapting to concept drift. It merges Creme and scikit‑multiflow work, ships classifiers, regressors, anomaly detectors (half‑space trees), scalers and metrics, and treats features and models as everything is a dictionary so snapshots serialize to JSON. Examples use NYC taxi data and integrate with brokers like Kafka / Quick Streams, and progressive validation evaluates performance online.
AI-written summary · about
chapters (8)
- 0:00Streaming and online ML
- 3:14Stream processing explained
- 8:10Offline versus online ML
- 12:00Why online ML matters
- 18:22Online ML use cases
- 21:50Batch and stream processing
- 22:33Introducing River library
- 29:56Summary and getting started
related talks
23mMoving from Offline to Online Machine Learning with River
30mMoving from Offline to Online Machine Learning with River
30mReconceptualizing Machine Learning for the Real-time World
22mOnline Machine Learning with Creme
41mStreams to the River: Powering an Online XGBoost Classifier with Kafka
37mAutomated Machine Learning: Building a Pipeline
26mFrom Data to Decision: Monitoring and Improving Water Networks using AI/ML
40mReal-time Learning with Spark