Table 5 Performance of nonlinear modelling methods comparisons and correlation between actual and predicted values on test and training sets.

From: Spectro-temporal acoustic elements of music interact in an integrated way to modulate emotional responses in pigs

Emotional Index

Model

RMSE train

RMSE test

Correlation

Positive high arousal index

GAM

2.10

2.33

0.63

RF

2.26

2.78

0.12

DT

2.87

3.80

0.08

XGB

2.59

2.66

0.24

Negative high index

GAM

1.25

1.44

0.50

RF

1.52

1.31

0.60

DT

1.68

1.57

0.11

XGB

1.66

1.65

0.07

Positive low emotions index

GAM

0.95

1.08

0.56

RF

0.97

1.23

0.32

DT

1.07

1.60

0.31

XGB

1.04

1.22

-0.10

  1. RMSE root mean square error, GAM generalized additive model, RF random forests model, DT decision trees model, XGB eXtreme gradient boosting model.
Back to article page