Fig 5. Precision recall characteristic plots for p = 40 under Models 1(a)-(d).

BVS-SL + G0_κ represents the Bayes variable selection with structure learning with belief parameter κ for all edges. Pencred, SSVS, Lasso, ENET, represent the penalized credible regions approach, stochastic search variable selection, L₁ penalized regression, and elastic net, respectively. The curves for SSL and SCAD are not presented to ensure greater clarity of the plot.