Table 5 Median robustness of normalised degree variance, \(\bar {v}\), and heterogeneity, ρ, to subsampling across 184 real-world networks
%n removed | 5% | 10% | 15% | 20% | 25% |
med\((|\bar {v}-\bar {v}_{sub}|)\) | 0.0004 | 0.0009 | 0.0013 | 0.0017 | 0.0022 |
med (|ρ−ρsub|) | 0.0013 | 0.0024 | 0.0034 | 0.0034 | 0.0039 |
% m lost (mean) | 9.7% | 19.6% | 28.1% | 36.3% | 44.3% |
%m removed inversely to degree | 5% | 10% | 15% | 20% | 25% |
med\((|\bar {v}-\bar {v}_{sub}|)\) | 0.0012 | 0.0023 | 0.0034 | 0.0043 | 0.0050 |
med (|ρ−ρsub|) | 0.0027 | 0.0052 | 0.0079 | 0.0107 | 0.0131 |
%m removed randomly | 5% | 10% | 15% | 20% | 25% |
med\((|\bar {v}-\bar {v}_{sub}|)\) | 0.0018 | 0.0040 | 0.0067 | 0.0078 | 0.0099 |
med (|ρ−ρsub|) | 0.0022 | 0.0045 | 0.0069 | 0.0101 | 0.0125 |