Similarity Analysis in Multilayer Temporal Food Trade Network
We analyze export/import food trade network that contains several layers. Each layer accounts for a particular commodity that countries trade with. The network has directed weighted edges. We look at statistical and topological similarity of layers in order to detect dependencies between different products trade. The measures include the estimation of out-degree correlation as well as the analysis of communities. We apply a normalization technique to the initial graphs that takes into account individual attributes of nodes and the possibility of groups formation. The most important elements of the networks are considered in order to compare different layers. Additionally, we analyze the network in time and detect the most similar periods of trade. The analysis of trade in dynamics gives the opportunity to track changes in export/import patterns. The results may have a significant contribution to the further analysis of food security of countries and the development of trade processes.