Hydrosphere Monitoring is a service for real-time monitoring of your production machine learning. It uses various statistical and machine learning methods to check that your production distribution matches your training one. If these distributions differ, your models will degrade, producing worse predictions. Hydrosphere Monitoring makes sure that you will catch the moment when your data begin to change so you can be always aware of your production machine learning quality.
- Statistical Drift Detection. Monitor how various statistics change in your data over time
- Multivariate Data Monitoring. Complex data drift can be detected with our multivariate data monitoring. Check outliers and anomalies in your production data streams
- Tabular, Image & Text data support out of the box.
- External Models Support. Connect models hosted outside Hydrosphere to Hydrosphere Monitoring to monitor their quality.
- Alerting. Be notified of your metrics changes. You can set up a notification about a change in the quality of your production data and respond to the problem in time.