Figure 4
From: Bayesian inference of transition matrices from incomplete graph data with a topological prior
Effects of the use of constraints on the inference of a transition matrix in empirical data. From left to right, the panels show the results for wikispeedia, neural, ecosystem, flights, and powergrid networks. We measure performance with the Frobenius distance between the transition matrix of the ground truth and that of the inferred model. Error bars represent the 95% confidence interval (mostly too small to be visible). In accordance with the synthetic results, the inclusion of the constraints allows BaCon to recover the transition matrix more precisely than both the frequentist and the noninformative Bayesian approaches