Figure 2

Partial least squares regression model optimization for Australian Selection Index using cross-validation. Shown is the mean prediction error (MSEP) in the training (MSEP_training) data set, the average MSEP in the 5-fold cross-validation samples (MSEP_CV), the proportion of EBV (Var_EBV) and SNP variance (Var_SNP) explained in the training data for models with an increasing number of latent components; the optimal prediction model includes the first 5 latent components, identified by the smallest MSEP_CV.