Predicting international trade communities from world development indicators
This project aimed to determine the predictive power of world development indicators for international trade communities, while inspecting changes and trends in these communities over time. It was hypothesized that world development indicators may contain the right balance of both sociologically and economically-driven information to predict international trade communities.
In order to test this hypothesis, a weighted, directed network was created using export FOB data from the International Monetary Fund’s (IMF) Direction of Trade (DOT) statistics. Communities were found in this network using the modularity maximization algorithm explained in [1]. These community classifications were then predicted from the World Bank’s World Development Indicators using a random forest classifier.
- Leicht EA, Newman ME. “Community structure in directed networks”. Phys Rev Lett. 2008 Mar 21; 100(11):118703. Epub 2008 Mar 21.