Anomaly Detection and Forecasting with River ML + InfluxDB 3
Session Date: Jun 18, 2026
Time: 8:00am (PT) | 3:00pm (GMT) | 4:00pm (BST)
Operational systems generate time series data continuously, and whether you’re detecting anomalies or forecasting what comes next, every second of lag between an event and a response has a cost. Batch ML models trained overnight can’t catch anomalies that emerge this afternoon, and they can’t forecast accurately against data that’s already hours old. InfluxDB’s River plugins bring online machine learning directly into the Processing Engine, so models update incrementally as new data arrives, keeping predictions current without a separate pipeline or retraining cycle. In this training, we will give an overview and tutorial of how to leverage the new River plugins for the InfluxDB 3 Processing Engine. You will be walked through the full process of setting up these plugins live, from getting the InfluxDB processing engine up and running to installing and configuring the River plugins, as well as a live demo of what they will look like in action.
- InfluxDB setup walkthrough and an explanation of the Processing Engine, including how write triggered, scheduled, and HTTP plugins work
- The case for online ML: why incrementally updating models outperform offline batch training for live time series workloads
- Installing and configuring the RiverML Anomaly Detector, Auto Profiler, and Forecaster plugins
- Live demo: anomaly detection and forecasting responding in real time, connected to downstream systems and alerting
