Table 2 Predictive performance for the 13 most-commonly dominating canopy tree species for predictions corrected for taxonomic reporting bias and for uncorrected (raw) predictions

From: Multispecies deep learning using citizen science data produces more informative plant community models

Species

n

Correction weight

Bias corrected

Raw

F1

Correct (#)

F1

Correct (#)

Picea abies

732

1.86

0.67

496

0.45

238

Fagus sylvatica

407

2.11

0.56

354

0.57

260

Abies alba

207

0.79

0.03

4

0.32

107

Larix decidua

114

0.39

0.25

17

0.46

49

Acer pseudoplatanus

80

1.07

0

0

0.02

1

Fraxinus excelsior

68

1.28

0.07

3

0.09

4

Castanea sativa

40

1.29

0.5

17

0.58

21

Pinus sylvestris

32

2.23

0.24

14

0.17

5

Betula pendula aggr.

25

0.2

0

0

0.06

1

Alnus incana

21

1.8

0.24

4

0.16

2

Pinus mugo

20

0.77

0.15

2

0.3

5

Quercus petraea

19

1.92

0

0

0

0

Pinus cembra

17

0.43

0.44

7

0.45

13

  1. n refers to number of test observations; F1 score is the harmonic mean of precision and recall; and correct (#) refers to number of correctly classified observations per species. Source data are provided as a Source Data file.
Back to article page