Skip to main content

Table 3 Mean accuracy increase of NetSparse relative to each other method and its standard error on the mean calculated over ten replicates each, times 100

From: Bayesian neural networks with variable selection for prediction of genotypic values

ScenarioGBLUPBayesBGBLUP-ADBayesB-ADGBLUP-ADAA
Base\(1.2\pm 0.6\)\(1.0\pm 0.5\)\({\textit{2.2}\pm \textit{0.7}}\)\({\textit{3.3}\pm \textit{0.6}}\)\({\textit{2.5}\pm \textit{0.6}}\)
\(S_{10}\)\({\textit{27.7}\pm \textit{1.6}}\)\({\textit{6.7}\pm \textit{0.8}}\)\({\textit{28.1}\pm \textit{1.5}}\)\({\textit{9.3}\pm \textit{1.1}}\)\({\textit{28.5}\pm \textit{1.4}}\)
\(S_{100}\)\({\textit{5.2}\pm \textit{1.1}}\)\({\textit{2.4}\pm \textit{0.7}}\)\({\textit{5.8}\pm \textit{1.1}}\)\({\textit{5.4}\pm \textit{0.7}}\)\({\textit{6.1}\pm \textit{1.1}}\)
\(S_{1000}\)\(-0.0\pm 0.2\)\({\textit{1.1}\pm \textit{0.3}}\)\(0.3\pm 0.2\)\({\textit{3.0}\pm \textit{0.7}}\)\(0.4\pm 0.3\)
\(D_{\text {medium}}\)\({\textit{0.8}\pm \textit{0.3}}\)\(0.6\pm 0.5\)\(0.0\pm 0.5\)\(0.8\pm 0.6\)\(0.4\pm 0.5\)
\(D_{\text {extreme}}\)\({\textit{0.6}\pm \textit{0.8}}\)\(0.6\pm 0.6\)\(-6.3\pm 0.6\)\(-6.1\pm 0.7\)\(-5.8\pm 0.7\)
\(E_A\)\({\textit{0.7}\pm \textit{0.2}}\)\(0.6\pm 0.7\)\({\textit{1.4}\pm \textit{0.4}}\)\({\textit{3.2}\pm \textit{0.6}}\)\({\textit{1.3}\pm \textit{0.5}}\)
\(E_C\)\(0.6\pm 0.4\)\({\textit{0.9}\pm \textit{0.3}}\)\(0.9\pm 1.0\)\(1.5\pm 1.1\)\(1.0\pm 1.0\)
\(E_I\)\({\textit{1.5}\pm \textit{0.5}}\)\({\textit{1.5}\pm \textit{0.4}}\)\({\textit{2.8}\pm \textit{0.6}}\)\({\textit{3.9}\pm \textit{0.6}}\)\({\textit{2.7}\pm \textit{0.6}}\)
  1. Significant entries, determined with the Benjamini-Hochberg procedure for \(\alpha =0.05\) for the one-sided paired t-test corresponding to the hypotheses \(\mathbb {E}\left( \rho _{\text {NetSparse}}-\rho _{\text {Method}}\right) =0\), are marked in italic