Table 3
Comparing the skill scores with previous models. The dashed entries mean the data is unavailable (see Whitman et al., 2022, for more details).
| Model | POD | FAR | TSS | HSS | POFD | CSI | Accuracy | Precision |
|---|---|---|---|---|---|---|---|---|
| Our BiLSTM model | ||||||||
| 1-day ahead | 0.618 | 0.087 | 0.531 | 0.732 | 0.001 | 0.583 | 0.99 | 0.913 |
| 2-day ahead | 0.412 | 0.125 | 0.287 | 0.553 | 0.001 | 0.389 | 0.985 | 0.875 |
| 3-day ahead | 0.147 | 0 | 0.147 | 0.252 | 0 | 0.147 | 0.980 | 1 |
| UMASEP-10 (Núñez, 2011) | 0.822 | 0.219 | – | – | – | – | – | – |
| PCA (Papaioannou et al., 2018) | 0.587 | 0.245 | – | 0.65 | – | – | – | – |
| SPARX (Dalla et al., 2017) | 0.5 | 0.57 | – | – | – | – | – | – |
| SPRINTS (Engell et al., 2017) | 0.56 | 0.34 | – | 0.58 | 0.32 | 0.3 | – | – |
| REleASE (Malandraki & Crosby, 2018) | 0.63 | 0.3 | – | – | – | – | – | – |
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.