Fig. 3: Time complexity analysis for temporal topological features versus sequentially static features.
From: Sequential stacking link prediction algorithms for temporal networks
a Time in seconds to compute temporal-extended network features, time-series modeled temporal features, and sequentially static features for 4 different types of features on 24 random edge-correlated T-SBM realizations with 50 nodes and 10 layers each. (20 parallel processes with 8 GB memory, see Acknowledgments) b Comparing AUC scores for link prediction from these temporal features (y-axis, blue) or time-series modeled features (y-axis, orange) versus the corresponding sequentially stacked features (x-axis). Both panels display data obtained in the partially-observed setting.
