Impact-Based Alerting & ML Outliers

Replace hardcoded rules with a flexible scoring system. Anomalies contribute weighted scores that determine entity status — Green, Orange, or Red.

DensityFunction algorithm with custom algorithm support. Learns time-of-day and day-of-week seasonality patterns automatically.

Low and normal confidence modes based on historical data sufficiency, preventing false positives during initial learning phases.

Min/max value boundaries with auto-correction for insignificant deviations, keeping alerts meaningful.

Test scoring changes and ML model adjustments before applying them to production, reducing risk of alert storms.

Detailed score breakdowns showing exactly which anomalies contributed to each entity's status, with full audit visibility.