posted on 2021-10-29, 17:32authored byEdward J. Gregr, Dana R. Haggarty, Sarah C. Davies, Cole Fields, Joanne Lessard
The difference in (A) the True Skill Statistic (TSS) and (B) the True Negative Rate (TNR) between the 20 and 100 m models across depth ranges, shown by region. Scores are based on model fit to the build testing partition. Values below 0 indicate a higher score by the 20 m model. This shows how regional model performance is generally better across all depths and regions, except for SOG, and identifies possible sampling biases in the 0–5 and 200+ depth ranges.