Open Access
Research Article

Table 3

Evaluation of the NN-based and the baseline models in the context of the classification problem.

Time horizon (h) Precision Recall Fscore FAR Threat score Heidke Skill Score
3 0.85 (0.64) 0.84 (0.07) 0.84 (0.12) 0.010 (0.002) 0.73 (0.06) 0.83 (0.11)
6 0.85 (0.66) 0.83 (0.07) 0.84 (0.12) 0.010 (0.002) 0.73 (0.07) 0.83 (0.11)
9 0.86 (0.67) 0.82 (0.07) 0.84 (0.12) 0.009 (0.002) 0.72 (0.07) 0.83 (0.11)
12 0.87 (0.68) 0.80 (0.07) 0.83 (0.13) 0.009 (0.002) 0.71 (0.07) 0.82 (0.12)
15 0.87 (0.69) 0.79 (0.07) 0.83 (0.13) 0.009 (0.002) 0.70 (0.07) 0.81 (0.12)
18 0.87 (0.69) 0.77 (0.07) 0.82 (0.13) 0.009 (0.002) 0.69 (0.07) 0.80 (0.12)
21 0.87 (0.70) 0.75 (0.07) 0.81 (0.13) 0.009 (0.002) 0.68 (0.07) 0.79 (0.12)
24 0.87 (0.69) 0.73 (0.07) 0.80 (0.13) 0.009 (0.002) 0.66 (0.07) 0.78 (0.12)

Notes. The results obtained with the NN-based model are given in bold. The results obtained with the baseline are given in brackets.

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.