Skip to main content
Figure 5 | EPJ Data Science

Figure 5

From: Temporal social network reconstruction using wireless proximity sensors: model selection and consequences

Figure 5

Characterisation of the reconstructed network. (a) Rates of reconstructed events by different methods (for colours see key). (b) Inter-event time distribution between original and reconstructed event on single links. (c) Distribution of the duration of interactions. Dashed lines on panels (b) and (c) depict approximating power-law functions with exponents 1.8 and 2.1 respectively. (d) Weighted static representation of a reconstructed network using the BiLSTM-RSSI method. Here link widths are proportional to the time spent together when both nodes were present and node sizes are proportional to node degrees. Nodes are coloured according to the original class partitions with darker nodes indicating adults (teachers, assistants or interns). For better visualisation we removed links which correspond to the weakest 3% of weights

Back to article page