Open Access

Table 3

Performance metrics tested in the optimization process. Each metric produces a different set of combination weights (i.e., a different ensemble). In each case a label is shown in parentheses that is used throughout the rest of the manuscript. Categorical metrics are calculated using 2 × 2 contingency table after probabilistic forecasts are turned into deterministic forecasts by choosing a threshold value, P th. See Appendix A for their definitions.

Probabilistic Categorical
Brier score (BRIER) Brier score (BRIER_C)
Mean absolute error (MAE) True skill statistic (TSS)
Linear correlation coefficient (LCC)a Heidke skill score (HSS)
Rank (nonlinear) correlation coefficient Accuracy (ACC)
(NLCC_ρ, NLCC_τ) Critical success index (CSI)
Reliability (REL) Gilbert skill score (GSS)
Resolution (RES)
Relative operating characteristic (ROC) curve area

NLCC_ρ and NLCC_τ are Spearman’s rank correlation and Kendall’s τ correlation, correspondingly.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.