Dashboard
📊 The STAMM Dashboard
One interactive interface for process engineers and ML specialists working with soft sensors in industrial production — real-time process data, model deployments, drift monitoring, and post-run assessment, all in one place.
The four sections
Pick a section below to jump into its guide — each one is illustrated with screenshots and a short video tutorial.
Data Source
Pick an experiment and explore every variable that goes with it — including a live link to the IBISBA Hub for project-level metadata.
- Experiment selector
- Statistical summary of variables
- Last measurement & valid points over time
Soft Sensors
Deploy any soft sensor from the Model Registry alongside a running experiment and visualize its predictions in real time.
- Model selector with full configuration view
- Live predictions next to sensor signals
- Adjustable time windows (6h, 1d, 1m, all)
Monitoring
Continuous drift monitoring with univariate and multivariate detectors, plus a model-divergence panel to compare candidates.
- Drift detectors with interpretable scores
- Train–test density comparison
- Model divergence vs the deployed baseline
Simulation Assessment
Post-run evaluation: label anomalies, filter the data, and generate multi-sheet reports from finished experiments.
- Interactive chart over the chosen window
- Filter & flag anomalies inline
- Downloadable multi-sheet reports