Decoding Crude Oil Trade Dynamics: A Comparative Analysis of Machine Learning Models

Globally, supply and demand dynamics, energy security considerations, and volatile international relations influence the crude oil energy landscape. This research aims to predict the formation and dissolution of crude oil trade links with specific reference to major crude oil exporting countries by employing classification-based machine learning algorithms involving Logistic Regression, Logistic Ridge Regression, Random Forest, and Support Vector Machine (SVM). It harnesses a dataset covering two decades comprising node-specific and edge-specific features to capture the economic, historical, and geopolitical nuances affecting crude oil trade linkages. The analysis reveals that out of all models employed, the Random Forest Classifier can aptly predict both active and non-active crude oil trade links, outperforming traditional econometric models and, further emphasizing the immense potential of employing ML algorithms in Energy Economics. These outcomes offer indispensable insights for policymakers and help them ensure energy security by anticipating alterations in trade links and enabling adaptive strategies in response to evolving global predicaments.