Figure 6

Comparison of graph measures using various artificial unweighted networks. Each panel shows the box plot of (a) global efficiency (e_glo), (b) average local efficiency (e_loc), (c) modularity (Q), (d) functional entropy (h_fun), (e) spectral entropy (h_spe), and (f) volume entropy (h_vol). The sparsity of the unweighted network was 0.09, 0.18, 0.27, 0.36, 0.45, and 0.54 from left to right. The color of the line represents RE (blue), SW (green), RA (red), SF (cyan), and magenta (HY). The order of the five types of different unweighted networks was consistent for the sparsity when the graph property was measured by the modularity Q in (c) and the volume entropy h_vol in (f). However, the order was changed more than three times depending on sparsity in the other graph measures, e_glo in (a), e_loc in (b), and h_spe in (e). h_fun in (d) was exactly the same for all graph types at the fixed sparsity.