Machine Learning

TrackMe ships its own native machine-learning engine — per-tenant statistical models that learn each entity’s normal numeric behaviour and flag deviations, feeding the impact score rather than raising alerts on their own.

• Outlier detection — how the models work, seasonality, confidence, scoring, and which components benefit, with a full in-depth reference.

• Incremental baseline training — why a model may retain bounded evidence across training cycles, how that changes the statistical meaning of its baseline, and when to prefer a full rolling-window rebuild.

• Outlier detection
• Incremental baseline training
• Outlier detection — in depth