An artificial intelligence approach of feature engineering and ensemble methods depicts the rumen microbiome contribution to feed efficiency in dairy cows

Table 2 Results from a linear mixed model for dry matter intake and gross feed efficiency traits in 454 lactating Holstein cows in the US and Canada. Individual parameters were tested using Type III sum of squares

Item	R²	Estimate	SE¹	P-value
Dry matter intake, kg/d
Parity²	0.02	0.87	0.22	< 0.001
MBW, kg	0.12	0.09	0.01	< 0.001
BEC, Mcal/d	0.05	0.17	0.02	< 0.001
NESec, Mcal/d	0.39	0.37	0.02	< 0.001
Treatment (random effect)	0.07
Total R²of the regression³	0.81
Residual feed intake (RFI)³	0.19
Milk fat efficiency, g/kg DMI
Parity²	0.01	-3.95	0.64	< 0.001
MBW	0.01	0.05	0.03	0.04
BEC	0.00	-0.03	0.07	0.63
DMI	0.18	-2.97	0.13	< 0.001
NESec	0.58	2.71	0.07	< 0.001
Treatment (random effect)	0.07
Total R²of the regression³	0.84
Residual MFE³	0.16
Milk protein efficiency, g/kg DMI
Parity²	0.02	2.84	0.49	< 0.001
MBW	0.00	-0.03	0.02	0.13
BEC	0.00	0.01	0.05	0.82
DMI	0.20	-1.80	0.10	< 0.001
NESec	0.41	1.30	0.05	< 0.001
Treatment (random effect)	0.07
Total R²of the regression³	0.69
Residual MPE³	0.31

¹Standard error of the estimate
²Primiparous were used as the reference for the parameter parity, meaning the estimate is in regard to the multiparous Holstein cow effect
³These R² were calculated based on the REML default PROC MIXED model in SAS 9.4 and used to derive the residual of feed intake (RFI), residual of MFE, and residual of MPE. These values are not the sum of each parameter R² reported in this table as those are based on the Type III sum of squares method chosen for hypothesis testing on each parameter

ISSN: 2524-4671