Evaluation summary statistics

What are summary statistics?

Summary statistics describe or summarize a collection of evaluation statistics, such as a collection of statistics across geographic features. For example, you may want to summarize the skill across a larger domain by plotting a histogram of the skill scores from the individual locations within it or by calculating an average skill or a standard deviation. In short, summary statistics provide descriptive information about a broader collection of evaluation statistics.

What dimensions can be summarized?

As of WRES v6.19, summary statistics can be produced for each of two dimensions within an evaluation:

1. All geographic features (e.g., declared as features); and
2. For each geographic feature group (e.g., declared as feature_groups).

What summary statistics are supported?

As of WRES v6.19, the following summary statistics can be calculated for each supported dimension (see 2. What dimensions can be summarized?):

• Mean
• Median
• Standard deviation
• Minimum
• Maximum
• Mean absolute
• Histogram (produced with 10 bins, by default)
• Box plot
• Quantiles (0.1, 0.5 and 0.9, by default)

How do I declare summary statistics?

See How do I declare summary statistics? on the Declaration Language](https://github.com/NOAA-OWP/wres/wiki/Declaration-Language) page.

Can I calculate sampling uncertainties for summary statistics?

No. For a complete answer see: Can I calculate sampling-uncertainties for summary statistics? on the Sampling uncertainty assessment page.