metric_providers

Metric computation providers for forecast evaluation.

Implements various metric providers that compute performance measures for probabilistic forecasts. Each provider handles specific metric types and can be configured to work with specific quantiles or all available quantiles.

Classes

MAPEProvider(**data)	Provides Mean Absolute Percentage Error metrics.
MeanAbsoluteCalibrationErrorProvider(**data)	Provides quantile calibration metrics.
MetricProvider(**data)	Base class for forecast metric computation.
ObservedProbabilityProvider(**data)	Provides observed probability metrics.
PeakMetricProvider(**data)	Provides metrics for peak detection performance.
R2Provider(**data)	Provides R² (coefficient of determination) metrics.
RCRPSProvider(**data)	Provides the Relative Continuous Ranked Probability Score.
RIQDProvider(**data)	Provides Relative Inter Quantile Distance metrics.
RMAEPeakHoursProvider(**data)	Provides Relative Mean Absolute Error metrics for peak hours only (8:00-20:00).
RMAEProvider(**data)	Provides Relative Mean Absolute Error metrics.
RelativePinballLossProvider(**data)	Provides Relative Pinball Loss metrics for quantile predictions.