Skip to main content

Table 1 Machine learned (ML) feature weights of the predictive sets selected in Constud

From: Shrubby cinquefoil (Dasiphora fruticosa (L.) Rydb.) mapping in Northwestern Estonia based upon site similarities

Feature

Num. of classes

Radius (m)

ML feature weights in predicting

Presence/absence in unclear partition

Coverage in

Predicted presence sites

Probable presence partition

Present land cover

15

0

 

0.8

1

Historical land cover

6

0

 

0.3

0.1

Historical land cover

6

200

  

0.4

Historical land cover

2

1000

  

0.1

Soil

25

0

  

1

Soil

25

200

  

0.3

Proportion of presences

 

100

2.9

 

0.1

Proportion of presences

 

1000

0.5

0.6

5.5

Mean coverage

 

100

0.2

2.0

0.2

Mean coverage

 

1000

0.3

 

0.6

Distance weighted mean coverage

 

10,000

1.1

 

1.8

  1. The three ML sets of feature weights match the machine learning operations in Fig. 4