Skip to main content

Table 2 Mean squared prediction error (MSPE) for the LASSO, Bayesian LASSO (BLASSO), genomic BLUP (GBLUP), reproducing kernel Hilbert space (RKHS) regression, random forests (RF) and Bayesian additive regression trees (BART) methods evaluated on the simulated QTLMAS2010 data when dominance and epistatic effects were added

From: Genome-wide prediction using Bayesian additive regression trees

Method Mean squared prediction error (MSPE)
LASSO
 minMSE 83.377
 minMSE + 1SE 84.832
BLASSO 71.857
GBLUP 92.296
RKHS  
 \(h = 0.05\) 92.361
 \(h = 0.1\) 91.852
 \(h = 0.25\) 91.906
RF M = 10 M = 25 M = 50 M = 100 M = 200 M = 400 M = 600
  107.908 105.123 100.784 101.992 100.327 100.900 99.836
BART M = 10 M = 25 M = 50 M = 100 M = 200 M = 400 M = 600
 \(q = 0.9\)
 \(\kappa\) = 2 80.717 76.892 70.845 65.294 65.196 66.283 66.906
 \(\kappa\) = 3 79.277 72.720 67.061 65.120 64.943 65.542 66.593
 \(\kappa\) = 4 87.030 71.401 65.635 64.353 65.149 66.483 68.050
 \(\kappa\) = 5 79.249 71.243 67.748 64.741 65.611 68.290 70.510
 \(q = 0.95\)
 \(\kappa\) = 2 86.328 70.452 67.744 65.465 65.308 65.801 66.998
 \(\kappa\) = 3 76.438 69.833 67.123 65.522 65.045 65.513 66.601
 \(\kappa\) = 4 86.653 74.651 67.164 67.220 65.074 66.544 68.163
 \(\kappa\) = 5 90.456 69.571 65.085 66.086 65.790 68.298 70.566
  1. The lowest MSPE obtained with each method is highlighted in italics. h is the bandwidth of the radial basis function kernel. M is the number of trees for RF and BART, and \(q\) and \(\kappa\) are hyperparameters of the BART priors. The stopping criteria for the regularization coefficient λ in LASSO were obtained based on tenfold cross-validation both at minimum MSE and minimum MSE plus 1 standard error [42]