Table 4 F1-score, precision and recall from the ensemble forecasting models for the EVA and the three validation datasets: Netherlands (NL), Austria (AT) and the French Forest Inventory (IFN).

From: EUNIS habitat maps: enhancing thematic and spatial resolution for Europe through machine learning

Strategy

F1-score

Precision

Recall

EVA

NL

AT

IFN

EVA

NL

AT

IFN

EVA

NL

AT

IFN

MA2 - Littoral biogenic habitats

0.87 ± 0.11

0.76 ± 0.34

  

0.82 ± 0.18

0.74 ± 0.33

  

0.95 ± 0.06

0.80 ± 0.36

  

N – coastal habitats

0.81 ± 0.09

0.82 ± 0.09

  

0.75 ± 0.15

0.84 ± 0.06

  

0.92 ± 0.07

0.83 ± 0.18

  

Q – Wetlands

0.73 ± 0.13

0.95 ± 0.06

0.56 ± 0.46

 

0.67 ± 0.19

0.95 ± 0.08

0.55 ± 0.45

 

0.86 ± 0.09

0.96 ± 0.07

0.57 ± 0.46

 

R – Grasslands

0.66 ± 0.17

0.40 ± 0.32

0.37 ± 0.35

 

0.78 ± 0.15

0.57 ± 0.42

0.45 ± 0.39

 

0.59 ± 0.18

0.37 ± 0.33

0.37 ± 0.36

 

S - Scrub and Tundra

0.83 ± 0.12

0.46 ± 0.39

0.31 ± 0.42

 

0.89 ± 0.06

0.49 ± 0.41

0.31 ± 0.42

 

0.79 ± 0.15

0.46 ± 0.41

0.31 ± 0.43

 

T – Forests

0.61 ± 0.20

0.38 ± 0.31

0.33 ± 0.39

0.48 ± 0.32

0.77 ± 0.15

0.47 ± 0.36

0.34 ± 0.40

0.78 ± 0.32

0.55 ± 0.23

0.37 ± 0.35

0.32 ± 0.39

0.40 ± 0.33

U - Sparsely vegetated

0.94 ± 0.03

0.81 ± 0.22

0.35 ± 0.40

 

0.95 ± 0.03

0.96 ± 0.05

0.34 ± 0.39

 

0.94 ± 0.03

0.78 ± 0.31

0.36 ± 0.23