Figure 7
Correlations between marbling score expected progeny differences in the training (testing) sets and their fitted (predicted) values using BayesCpC and BRANN and SCGANN with different numbers of neurons in the hidden layer and using 700 SNPs. 1training = correlations in the training sets; testing = correlations in the testing sets; 2BayesCpC = Bayesian regression model, where BayesCπ is used for feature selection and BayesCπ with π = 0 is used for post-selection statistical inference and cross-validation; BRANN = artificial neural network with Bayesian regularization; SCGANN = artificial neural network with scaled conjugate gradient back-propagation.