Figure 3
From: Empirical evaluation of scoring functions for Bayesian network model selection
The effect of the hyperparameter α on the BDeu score. This figure plots the SHD between the networks learned by BDeu and the gold standard networks for six values of α for the Breast, Glass, Diabetes, and Hepatitis datasets. We used the behavior of each curve to evaluate both the convergence and the recovery ability of each value of α. We evaluate the recovery ability by considering both the smallest SHD for the scoring function, the size of the dataset which gives that SHD, and whether the scoring function converged to the smallest SHD, some other SHD or did not converge.