Table 1 The relationship between probability of occurrence of false presences (A) and false absences (B) and a set of uncorrelated environmental covariates.

From: Integrating drone-borne thermal imaging with artificial intelligence to locate bird nests on agricultural land

 

β

SE

z

p

(A) False presence

Intercept

− 1.767

0.133

13.26

< 0.001

Cloud cover

− 0.005

0.001

3.63

< 0.001

Temperature

0.028

0.011

2.50

0.013

Wind speed

0.020

0.018

1.09

0.276

Substrate (un-ploughed)

− 0.199

0.089

2.24

0.025

Drone height (25 m)

− 0.053

0.087

0.61

0.542

(B) False absence

Intercept

− 0.867

0.737

1.18

0.240

Cloud cover

− 0.002

0.007

0.31

0.759

Temperature

− 0.098

0.062

1.58

0.113

Substrate (un-ploughed)

− 0.453

0.392

1.15

0.250

Drone height (25 m)

0.974

0.460

2.12

0.035

  1. Effect size (β), standard error (SE), test statistics (z) as well as p values are derived from averaging across the set of best supported models (reported in Table S1) run separately for false presence and false absence (see “Methods” section for more details). For the two categorical variables with two classes each, results consider un-ploughed fields as a reference for the substrate variable (compared to ploughed substrate), and height of 25 m for the drone height variable (compared to 15 m height).
Back to article page