Table 2.
Win rate evaluation of link prediction
CHA | SPM | SBM | stacking basic | stacking basic + CH + SPM + SBM | ||
---|---|---|---|---|---|---|
precision | Biological | 0.607 | 0.353 | 0.223 | 0.179 | 0.158 |
Economic | 0.653 | 0.527 | 0.354 | 0.249 | 0.244 | |
Informational | 0.744 | 0.272 | 0.167 | 0.083 | 0.050 | |
Social | 0.932 | 0.111 | 0.022 | 0.027 | 0.019 | |
Technological | 0.521 | 0.366 | 0.121 | 0.246 | 0.253 | |
Transportation | 0.349 | 0.491 | 0.360 | 0.434 | 0.423 | |
Mean | 0.634 | 0.353 | 0.208 | 0.203 | 0.191 | |
AUC precision | Biological | 0.542 | 0.271 | 0.161 | 0.152 | 0.133 |
Economic | 0.669 | 0.556 | 0.332 | 0.278 | 0.275 | |
Informational | 0.656 | 0.311 | 0.117 | 0.061 | 0.056 | |
Social | 0.938 | 0.244 | 0.016 | 0.031 | 0.027 | |
Technological | 0.480 | 0.343 | 0.136 | 0.233 | 0.234 | |
Transportation | 0.414 | 0.434 | 0.377 | 0.394 | 0.411 | |
Mean | 0.616 | 0.360 | 0.190 | 0.192 | 0.189 | |
AUC-PR | Biological | 0.611 | 0.234 | 0.193 | 0.118 | 0.082 |
Economic | 0.877 | 0.468 | 0.410 | 0.364 | 0.362 | |
Informational | 0.811 | 0.250 | 0.133 | 0.033 | 0.039 | |
Social | 0.932 | 0.134 | 0.015 | 0.023 | 0.019 | |
Technological | 0.561 | 0.379 | 0.200 | 0.401 | 0.350 | |
Transportation | 0.606 | 0.471 | 0.414 | 0.529 | 0.517 | |
Mean | 0.733 | 0.323 | 0.228 | 0.245 | 0.228 | |
AUC-ROC | Biological | 0.407 | 0.158 | 0.322 | 0.182 | 0.322 |
Economic | 0.078 | 0.028 | 0.900 | 0.040 | 0.040 | |
Informational | 0.367 | 0.044 | 0.483 | 0.189 | 0.339 | |
Social | 0.927 | 0.411 | 0.026 | 0.904 | 0.905 | |
Technological | 0.129 | 0.007 | 0.360 | 0.377 | 0.440 | |
Transportation | 0.417 | 0.006 | 0.214 | 0.460 | 0.274 | |
Mean | 0.387 | 0.109 | 0.384 | 0.359 | 0.387 | |
AUC-mROC | Biological | 0.550 | 0.223 | 0.165 | 0.116 | 0.079 |
Economic | 0.557 | 0.192 | 0.289 | 0.006 | 0.003 | |
Informational | 0.744 | 0.206 | 0.139 | 0.022 | 0.028 | |
Social | 0.944 | 0.259 | 0.015 | 0.026 | 0.025 | |
Technological | 0.421 | 0.236 | 0.066 | 0.213 | 0.210 | |
Transportation | 0.237 | 0.214 | 0.251 | 0.226 | 0.251 | |
Mean | 0.576 | 0.222 | 0.154 | 0.101 | 0.099 | |
NDCG | Biological | 0.551 | 0.215 | 0.164 | 0.099 | 0.069 |
Economic | 0.577 | 0.117 | 0.368 | 0.005 | 0.002 | |
Informational | 0.800 | 0.200 | 0.117 | 0.017 | 0.022 | |
Social | 0.932 | 0.460 | 0.015 | 0.046 | 0.037 | |
Technological | 0.443 | 0.170 | 0.057 | 0.260 | 0.231 | |
Transportation | 0.297 | 0.154 | 0.243 | 0.294 | 0.271 | |
Mean | 0.600 | 0.219 | 0.161 | 0.120 | 0.105 | |
MCC | Biological | 0.607 | 0.353 | 0.223 | 0.179 | 0.158 |
Economic | 0.653 | 0.526 | 0.354 | 0.250 | 0.244 | |
Informational | 0.744 | 0.272 | 0.167 | 0.083 | 0.050 | |
Social | 0.931 | 0.112 | 0.022 | 0.027 | 0.019 | |
Technological | 0.521 | 0.369 | 0.124 | 0.249 | 0.257 | |
Transportation | 0.349 | 0.491 | 0.360 | 0.434 | 0.423 | |
Mean | 0.634 | 0.354 | 0.208 | 0.204 | 0.192 |
550 real-world networks of Ghasemian et al.1 are considered. For each network and for each link prediction method, the link prediction evaluation framework is applied (10 repetitions). The table reports, for each network domain and for each evaluation measure, the win rate of each method over the networks of that domain and over the 10 repetitions. For each evaluation measure and for each link prediction method, the mean over the network domains is also reported. For each evaluation measure and for each domain, the highest mean result is highlighted in bold, as well as the highest overall mean result over the domains.