Skip to main content
Figure 8 | EPJ Data Science

Figure 8

From: A path-based approach to analyzing the global liner shipping network

Figure 8

Percentage of edges and shipping length (real shipping distance between ports, measured in kilometers) by edge type in the original work (left) and with our updated minimum-route paths filtered with distance threshold \(\alpha =1.15\) (right). The role of the core was simultaneously underestimated for all shortest paths between peripheral ports and overestimated for only paths between peripheral ports that pass through at least 1 core edge. The first bar in each plot shows the percentage of edges in the graph that fall into each category (sum of in and out feeders in gray to the left of each bar in the right plot). The second bar shows the percentage of total shipping length for edges in each category in the graph. The numbers in parentheses are the length percentage divided by the edge percentage. The majority of edges in both representations are local (64.1% and 64.8%), but local edges make up relatively less of the total length among all edges in the graphs. The third bar in each plot shows the percentage of length by category for all paths between peripheral ports (75% of shortest paths in the previous analysis vs. 26% here), while the fourth bar shows the same quantity but only for paths that pass through at least 1 core edge (25% previously vs. 74% here). Using shortest paths through the undirected co-route graph we find that the role of the core was under-estimated for all paths (16.6% vs. 24.2%). However, the length-to-edge percentage ratios are similar for all categories (core: 5.2 vs. 4.6; feeder: 1.7 vs. 1.3; local: 0.4 vs. 0.5). Limiting to only paths that pass through the structural core, the role of the core appears to have been overestimated (62.0% vs. 27.7%) and the role of local edges was underestimated in the previous work (4.6% vs. 33.3%). In this case, the length-to-edge percentage ratio for the core is much smaller (19.4 vs. 5.3), while the feeder (1.0 vs. 1.3) and local (0.1 vs. 0.5) ratios are again similar. Out-feeder edges, which are edges that leave the structural core, make up a higher percentage of both edges and lengths than in-feeder edges, which go from outside the core to inside

Back to article page