Fig. 6
From: Machine learning meets complex networks via coalescent embedding in the hyperbolic space
Greedy routing (GR) on real networks. The eight real networks whose statistics are reported in Table 1 have been mapped using the hyperbolic embedding techniques and the greedy routing in the geometrical space has been evaluated. The barplot report for each method the mean GR-score and standard error over the networks. The GR-score is a metric to evaluate the efficiency of the greedy routing, which assumes values between 0, when all the routings are unsuccessful, and 1, when all the packets reach the destination through the shortest path (see “Methods” section for details). Both the EA (a) and non-EA (b) variants are reported, in order to check whether the equidistant adjustment might affect the navigability. A black arrow points the coalescent embedding algorithm RA-ncMCE that offers the best performance regardless the use of node angular adjustment. The mean GR-score of RA-ncMCE is not statistically different from the one of the HyperMap-based algorithms (permutation test p value >0.2 in all the pairwise comparisons)
