Skip to main content
. Author manuscript; available in PMC: 2021 Nov 7.
Published in final edited form as: Br J Math Stat Psychol. 2019 Jun 17;73(2):187–212. doi: 10.1111/bmsp.12173

Figure 3.

Figure 3.

Risk of the estimated precision matrices. Lower scores are closer to the true network. The probability of a connection varies from 20% to 80%. γ denotes the EBIC parameter (equation 5). EBIC = extended Bayesian information criterion; HD = Hamming distance; KL = Kullback–Leibler divergence. The ribbons are ±1 SD.