Table 2.

AUC for four classification models used as discrimination tools to identify light pigs at weaning (i.e., live weight standardized to 21 d of age), evaluated using three different weight cut points

	Testing data set
	1	2	3	Mean ± SE
Cut point < 4.31 kg (10%)
Linear model	0.823	0.823	0.809	0.818 ± 0.005
Decision tree	0.675	0.604	0.638	0.639 ± 0.021
Random forest	0.829	0.823	0.827	0.826 ± 0.002
Generalized boosted regression	0.837	0.833	0.827	0.832 ± 0.003
Cut point < 4.90 kg (20%)
Linear model	0.812	0.807	0.802	0.807 ± 0.003
Decision tree	0.622	0.647	0.686	0.652 ± 0.019
Random forest	0.815	0.816	0.817	0.816 ± 0.001
Generalized boosted regression	0.819	0.818	0.816	0.818 ± 0.001
Cut point < 5.33 kg (30%)
Linear model	0.791	0.782	0.784	0.786 ± 0.003
Decision tree	0.685	0.526	0.612	0.608 ± 0.046
Random forest	0.800	0.801	0.792	0.798 ± 0.003
Generalized boosted regression	0.801	0.796	0.792	0.796 ± 0.003