Figure 4

Performance of the Louvain method [[46]], fast modularity method [[47]], the spin-glass based method [[48]], the infoMAP method [[49]], label propagation [[50]] and the random-walk based method [[51]] in networks with mean degree 10. Fill circles, open circles and triangles represent networks with Poisson, geometric and power-law degree distributions, respectively. Each data point represents the average over ten modular random graphs. Error bars represent standard deviations. The solid line is the reference line where estimated modularity is equal to the input modularity.