Geopolitical Risk and Country-Level CO2 Emissions: A Deep Learning Approach Comparing LSTM, CNN and ConvLSTM
DOI:
https://doi.org/10.32479/ijeep.17764Keywords:
Geopolitical Risks, CO2 Emissions, Deep Learning, LSTM Model, Climate Change, Environmental SustainabilityAbstract
The investigation in this research employs advanced deep learning methods to analyze the bidirectional relationship between geopolitical risks and CO2 emissions in China, India, and the USA across the timeframe of 1990 to 2019. Data sourced from the Recent GPR Index and CO2 emissions was utilized. The exploration of how geopolitical risks influence climate change provides valuable insights for decision-makers in policymaking roles. The study's outcomes indicate a positive correlation between geopolitical risks and CO2 emissions across all three nations, with both variables exhibiting an ascending trend throughout the studied period. Noteworthy is the consistent superior performance of the Long Short-Term Memory (LSTM) model when compared to the Convolutional Neural Network (CNN) and Convolutional Long Short-Term Memory (ConvLSTM) models in short term prediction. This consistent effectiveness in CO2 emission prediction showcases the strength of the LSTM model. The divergence in model efficacy among the different nations highlights the significance of tailoring CO2 emission prediction models to each country's unique attributes. These insights should significantly influence policymakers as they strategize approaches to mitigate geopolitical risks, curtail CO2 emissions, and to adopt well-informed strategies in combating climate change.Downloads
Downloads
Published
2025-06-25
How to Cite
Sanusi, O. I., Aliyu, S., & Safi, S. (2025). Geopolitical Risk and Country-Level CO2 Emissions: A Deep Learning Approach Comparing LSTM, CNN and ConvLSTM. International Journal of Energy Economics and Policy, 15(4), 109–117. https://doi.org/10.32479/ijeep.17764
Issue
Section
Articles