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 2 Prediction models for nitrogen use efficiency, nitrogen loss and dry matter intake

Models	Predictors¹	Number of input variables	Spectra pre-treatment	Algorithms	Count²
Model 1	MIR	215	None, MSC, SNV	PLS, RR, SVM	9
Model 2	+ MY, parity	217			9
Model 3	+ MY, parity, BCS	218			9
Model 4	+ MY, parity, BCS, DIM_g	219			9
Model 5	+ MY, parity, BCS, DIM_g, DIP	220			9
Model 6	MY, parity, BCS, DIM_g, DIP, Protein, fat, lactose, MUN (excluding MIR)	9	No MIR		3

MIR mid-infrared, MY milk yield, BCS body condition score, DIM_g days in milk grouped by 5 days, DIP days in pregnancy, MUN milk urea nitrogen, MSC multiplicative scatter correction, SNV standard normal variate, PLS partial least squares, RR ridge regression, SVM support vector machine
¹For Model 2 to 5, the additional predictor of next model is based on Model 1, and Model 6 includes all additional predictors, except for MIR spectra
²Number of prediction models developed using this set of predictors: 3 algorithms times 3 types of MIR spectra for models 1–5 = 9 models

ISSN: 2049-1891