Figure 2: Significant second-order Markov constraints on flow.
From: Memory in network flows and its effects on spreading dynamics and community detection
(a–f) First- and second-order conditional entropy for all nodes of the six analysed networks. Blue nodes show a significant memory effect, because the null hypothesis that the data are generated from a first-order Markov model can be rejected. Red nodes do not show a significant effect. The memory effect is the difference in entropy between a first- and second-order Markov model. Las Vegas, among all cities, shows the strongest memory effect. Traffic is dominated by visitors who return to the city from which they came. In the other extreme, nodes that we could not significantly distinguish from a first-order Markov model typically have low connectivity and relatively small entropies.
