Predicting nitrogen use efficiency, nitrogen loss and dry matter intake of individual dairy cows in late lactation by including mid-infrared spectra of milk samples

Table 7 The bias, slope, and random proportions of the mean square prediction error of best prediction models in the external validation for each trait¹

Trait	Algorithm	Model	MIR	MSPE²	RE	Bias%	Slope%	Random%
NUE	PLS	2	Original	22.7(1.4)	0.19(0.01)	11.3(5.5)	6.0(4.3)	82.7(5.4)
	RR	4	Original	25.7(0.9)	0.20(0.004)	7.8(1.2)	1.4(1.0)	90.9(0.6)
	SVM	3	Original	23.4(0.5)	0.18(0.003)	2.1(1.8)	3.2(4.0)	94.7(5.3)
NL	PLS	3	Original	9.6e-03(3.2e-04)	0.20(0.003)	1.2(0.7)	0.7(0.6)	98.2(1.3)
	RR	6	No	1.4e-02(3.9e-04)	0.26(0.01)	4.3(3.7)	8.9(3.3)	86.9(2.7)
	SVM	3	Original	1.1e-02(1.8e-04)	0.21(0.004)	0.8(1.1)	2.2(0.9)	97.0(1.2)
DMI_a	PLS	6	No	47.4(4.2)	0.28(0.01)	0.8(0.3)	15.3(17.0)	83.9(16.9)
	RR	6	No	20.6(0.7)	0.18(0.01)	9.8(2.1)	10.5(3.3)	79.7(1.8)
	SVM	3	Original	15.3(0.3)	0.16(0.002)	0.4(0.3)	3.7(0.8)	96.0(0.8)

¹NUE nitrogen use efficiency, NL nitrogen loss, DMI_a 3-d moving average of dry matter intake, PLS partial least squares, RR ridge regression, SVM support vector machine, MIR mid-infrared, Original MIR without pre-treatment, No MIR is not included in the model, MSPE mean square prediction error, RE relative error, Bias% proportion of error due to mean bias, Slope% proportion of error due to deviation of the slope from 1, Random% proportion of error explained by random error. Values between brackets indicate the standard deviation
²The unit of MSPE: % × % for NUE; kg × kg for NL and DMI_a

ISSN: 2049-1891