Table 3 Predictive mean- squared error (PMSE) between observed and predicted values for grain yield (GY) and days to heading (DTH) in 12 environments for seven models
TraitEnvironmentBLBRRBayes ABayes BRKHSRBFNNBRNN
113.0213.1812.7213.2311.0210.8511.52
211.8912.3710.6511.2810.1910.7210.44
38.188.447.317.596.296.256.63
421.5922.2721.7921.6721.1422.6421.49
DTH58.869.238.488.377.958.028.21
814.7215.2214.5414.5813.1213.1914.81
921.3821.4423.7123.9320.5019.8420.62
107.728.517.277.576.666.517.36
116.837.126.596.746.035.966.51
1213.6014.4213.5613.4613.2514.8615.75
Average6.096.475.996.285.319.129.25
10.070.090.070.070.070.070.07
20.060.080.060.060.060.070.07
30.060.070.060.060.050.050.05
GY40.220.240.230.230.200.190.21
50.390.440.260.270.350.330.36
60.130.150.120.130.120.130.13
70.400.410.430.440.380.370.39
Average0.060.070.050.050.050.070.06
  • Fitted models were Bayesian LASSO (BL), RR-BLUP (BRR), Bayes A, Bayes B, reproducing kernel Hilbert space regression (RKHS), radial basis function neural networks (RBFNN) and Bayesian regularized neural networks (BRNN) across 50 random partitions of the data with 90% in the training set and 10% in the validation set. The models with lowest PMSE are underlined.