Causality Analysis for COVID-19 among Countries Using Effective Transfer Entropy

In this study, causalities of COVID-19 across a group of seventy countries are analyzed with effective transfer entropy. To reveal the causalities, a weighted directed network is constructed. In this network, the weights of the links reveal the strength of the causality which is obtained by calculating effective transfer entropies. Transfer entropy has some advantages over other causality evaluation methods. Firstly, transfer entropy can quantify the strength of the causality and secondly it can detect nonlinear causal relationships. After the construction of the causality network, it is analyzed with well-known network analysis methods such as eigenvector centrality, PageRank, and community detection. Eigenvector centrality and PageRank metrics reveal the importance and the centrality of each node country in the network. In community detection, node countries in the network are divided into groups such that countries in each group are much more densely connected.

Automated summary

This study analyzes the causalities of COVID-19 among seventy countries using effective transfer entropy. It constructs a weighted directed network to reveal the strength of the causality, which is obtained by calculating effective transfer entropies. Transfer entropy has the advantage of quantifying the strength of causality and detecting nonlinear causal relationships. The causality network is then analyzed using network analysis methods such as eigenvector centrality, PageRank, and community detection. Eigenvector centrality and PageRank metrics reveal the importance and centrality of each country in the network, while community detection groups node countries with denser connections.