Metrics
ClimateLearn provides the following metrics for deterministic predictions.
Mean squared error (MSE)
Mean squared error skill score (MSESS)
Mean absolute error (MAE)
Root mean squared error (RMSE)
Anomaly correlation coefficient (ACC)
Pearson’s correlation coefficient (Pearson)
Mean bias
Normalized root mean squared error (NRMSEs)
Normalized root mean squared error in global mean (NRMSEg)
For probabilistic forecasts, the library provides the following metrics.
Gaussian continuous ranked probability score (Gaussian CRPS)
Gaussian spread
Gaussian spread-skill ratio (Gaussian SSR)
We refer to the following sources for the definitions and motivations for these metrics: