Fig. 4
From: Module-based regularization improves Gaussian graphical models when observing noisy data
Application of Standard and Modular GLASSO to Covid-19 incidence data. The Standard GLASSO (log-likelihood) and Modular GLASSO (codelength savings) suggest vastly different regularization strengths for the Covid-19 data (a). The Standard GLASSO reveals no modular structure in the resulting network, while the Modular GLASSO uncovers the 14 modules represented by different colors on the world map (b). The modules exhibit a geographical signal as adjacent countries tend to belong to the same module, with some interesting exceptions