- reference
- Refers to a dataset which will be considered as the source of truth in the given comparison. Many times this is some kind of observational dataset, but can be any dataset of your choosing. Errors and scores will be calculated with respect to this dataset.
- comparison
- Refers to the dataset being compared to the reference dataset. Usually, this is a model output dataset, but could be any dataset of your choosing.
- scalar
- A scalar is a single numerical value that is generated by an analysis function. In
ilamb3it can be a scalar metric such as mean absolute error, or other single numerical values such as the mean over a region and the time period. - score
- In
ilamb3a score is a scalar that represents a synthesis of model performance. It non-dimensionalizes errorsand compresses the values to be on the unit interval where 0 is the worst and 1 is the best. This allows scores to be combined to form an overall impression of model performance.