Using machine learning to improve the accuracy of genomic prediction of reproduction traits in pigs

Table 5 Accuracy and mean squared error (MSE) of genomic prediction of TNB and NBA in younger individuals from seven methods

Hyperparameters	Method	TNB¹			NBA²
Hyperparameters	Method	Accuracy³	MSE	Optimal hyperparameters⁴	Accuracy³	MSE	Optimal hyperparameters⁴
	GBLUP	0.355^ab	11.598	–	0.264^ab	10.203	–
	ssGBLUP	0.408^b	11.221	–	0.288^ab	9.974	–
	BayesHE	0.357^ab	11.566	–	0.262^ab	10.143	–
Tuning	SVR	0.307^a	11.488	kernel = ‘rbf’; gamma = 0.00005; C = 14	0.229^a	10.235	kernel = ‘rbf’; gamma = 0.00005; C = 13
	KRR	0.362^ab	11.367	kernel = ‘rbf’; gamma = 0.000001; λ = 0.07	0.266^ab	10.121	kernel = ‘rbf’; gamma = 0.000001; λ = 0.12
	RF	0.385^ab	11.337	n_estimators = 430; max_depth = None	0.285^ab	10.116	n_estimators = 400; max_depth = None
	Adaboost.R2_KRR	0.395^b	11.254	n_estimators = 70; kernel = ‘rbf’, gamma = 0.00001, λ = 1	0.328^b	9.794	n_estimators = 60; kernel = ‘rbf’, gamma = 0.00001, λ = 0.9
Default	SVR	0.271	11.858	–	0.17	10.37	–
	KRR	0.346	11.538	–	0.259	10.158	–
	RF	0.26	11.867	–	0.179	10.335	–
	Adaboost.R2_KRR	0.36	11.392	–	0.322	9.797	–

¹ TNB: total number of piglets born
² NBA: number of piglets born alive
³ Accuracy: the correlation between corrected phenotypes and predicted values of the validation population;
⁴Optimal hyperparameters: The optimal hyper-parameters of each machine learning method obtained by using grid search
The different superscript of accuracy indicates the significant difference by the Hotelling-Williams test

ISSN: 2049-1891