Module monitoring

Expand description

Production monitoring with drift detection Production monitoring with drift detection and concept drift algorithms

This module provides:

Core monitoring (requires monitoring feature): TransformationMonitor with KS, PSI, MMD, Wasserstein drift detection, performance metrics, and alerting.
Drift detection: DriftDetector trait with KS, PSI, Wasserstein, and MMD implementations for comparing reference vs. test distributions.
ADWIN: ADaptive WINdowing algorithm for online concept drift detection in streaming data.

Re-exports§

ARIMAConfig: Configuration for ARIMA (AutoRegressive Integrated Moving Average) model
AdvancedAnomalyDetector: Advanced anomaly detection system
AlertConfig: Alert configuration
AnomalyFeedback: Feedback for anomaly detection tuning
AnomalyInsights: Anomaly insights summary
AnomalyRecord: Anomaly record for historical analysis
AnomalyThresholds: Anomaly detection thresholds
ChangePointConfig: Configuration for change point detection
DriftDetectionResult: Data drift detection result
EnsembleAnomalyDetector: Ensemble anomaly detector combining multiple methods
ForecastModel: Time series forecasting model configuration
IsolationForestConfig: Configuration structures for various anomaly detection methods
LOFConfig: Configuration for Local Outlier Factor (LOF) detection
MLAnomalyDetector: Machine learning anomaly detector
OneClassSVMConfig: Configuration for One-Class SVM anomaly detection
PerformanceMetrics: Performance degradation metrics
SeasonalConfig: Configuration for seasonal time series decomposition
StatisticalDetector: Statistical anomaly detector using multiple statistical methods
TimeSeriesAnomalyDetector: Time series anomaly detector
TimeSeriesPoint: Time series data point
TransformationMonitor: Production monitoring system