Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Deep Causal Learning to Explain and Quantify The Geo-Tension's Impact on Natural Gas Market

Philipp Kai Peter, Yulin Li, Ziyue Li, Wolfgang Ketter

TL;DR

The paper addresses how geopolitical shocks like the Russia-Ukraine war affect German natural gas demand ($NGD$) and how to quantify these effects. It combines mutual-information-driven feature selection, $LSTM$-based nonlinear Granger causality, and counterfactual forecasting with $LSTM$ and $Prophet$ to estimate war-free scenarios. Key contributions include accurate deep-learning prediction of $NGD$, sector-specific quantification of war effects across $LDZ$, $IND$, and $GTP$, and a transferable framework for rapid adaptation to shocks. The approach supports energy planning and policy-making under geopolitical uncertainty, with code and data available at the referenced GitHub repository.

Abstract

Natural gas demand is a crucial factor for predicting natural gas prices and thus has a direct influence on the power system. However, existing methods face challenges in assessing the impact of shocks, such as the outbreak of the Russian-Ukrainian war. In this context, we apply deep neural network-based Granger causality to identify important drivers of natural gas demand. Furthermore, the resulting dependencies are used to construct a counterfactual case without the outbreak of the war, providing a quantifiable estimate of the overall effect of the shock on various German energy sectors. The code and dataset are available at https://github.com/bonaldli/CausalEnergy.

Deep Causal Learning to Explain and Quantify The Geo-Tension's Impact on Natural Gas Market

TL;DR

The paper addresses how geopolitical shocks like the Russia-Ukraine war affect German natural gas demand () and how to quantify these effects. It combines mutual-information-driven feature selection, -based nonlinear Granger causality, and counterfactual forecasting with and to estimate war-free scenarios. Key contributions include accurate deep-learning prediction of , sector-specific quantification of war effects across , , and , and a transferable framework for rapid adaptation to shocks. The approach supports energy planning and policy-making under geopolitical uncertainty, with code and data available at the referenced GitHub repository.

Abstract

Natural gas demand is a crucial factor for predicting natural gas prices and thus has a direct influence on the power system. However, existing methods face challenges in assessing the impact of shocks, such as the outbreak of the Russian-Ukrainian war. In this context, we apply deep neural network-based Granger causality to identify important drivers of natural gas demand. Furthermore, the resulting dependencies are used to construct a counterfactual case without the outbreak of the war, providing a quantifiable estimate of the overall effect of the shock on various German energy sectors. The code and dataset are available at https://github.com/bonaldli/CausalEnergy.
Paper Structure (12 sections, 2 equations, 5 figures)

This paper contains 12 sections, 2 equations, 5 figures.

Table of Contents

  1. Introduction
  2. Related Work
  3. Prediction Methods for Natural Gas Demand
  4. Impact of the Russia-Ukraine war on Natural Gas Demand
  5. Methodology
  6. Research Approach
  7. Data Sources and Definition
  8. Implementation and Results
  9. Data Exploration and Mutual Information Analysis
  10. Nonlinear Granger Causality
  11. Counterfactual Intervention Studies
  12. Discussion and Conclusion

Figures (5)

  • Figure 1: (1) Selection of influencing factors for each sector of LDZ, IND, and GTP; (2) Deep Non-linear Granger Causality based on LSTM and Prophet; (3) Intervention studies based on counterfactual prediction using only pre-war training data.
  • Figure 2: (a) Scatter plots of relationships between NGD and multiple factors; (b) Mutual information analysis; (c) Selected factors (denoted with "$\times$") per sector and the aggregated NGD
  • Figure 3: LSTM-based Granger Causality Test Result
  • Figure 4: Monthly difference ($\Delta$) (in mcm) of forecasts with (red) and without (green) the war comparing with actual demands: top (LDZ), mid (IND), and bottom (GTP).
  • Figure 5: Monthly difference ($\Delta$) (in percentage) of forecasts with (blue) and without (orange) the war comparing with actual demands: top (LDZ), mid (IND), and bottom (GTP).