Forecasting of Natural Gas Supplied for EU in Gazprom with ARIMA and Machine Learning Methods
Abstract
Natural gas, as an environmentally friendly energy source, is gaining importance with its capacity to respond to the increasing energy demand worldwide. Its wide range of uses and low carbon emissions place it at the heart of sustainable energy policies. This critical resource in the energy sector has a dynamic and complex structure under the influence of political, economic, socio-cultural and technological factors. In particular, the forecasting of natural gas supply is of great importance for energy planning and strategic decision making. This study examines different modelling techniques for forecasting Gazprom's natural gas supply. ARIMA, ELM and MLP models are used, and their performance is compared. ARIMA is a classical method often preferred in time series analysis and makes predictions based on past values of the data. ELM is a model based on artificial neural networks and has a fast-learning capability. MLP is a deep learning method with the ability to model complex relationships thanks to its layered structure. As a result of the comparisons, the MLP model was found to perform best. MLP has the lowest error criterion and is more successful than other models in forecasting natural gas supply. This can be attributed to the complexity of the MLP and its strong learning ability.
Keywords
References
- Bayraç, H. N. (2018). Uluslararası doğalgaz piyasasının ekonomik yapısı ve uygulanan politikalar. Eskişehir Osmangazi Üniversitesi İktisadi ve İdari Bilimler Dergisi, 13(3), 13-36.
- IEA, Energy for All, IEA, Paris, (2011). https://www.iea.org/reports/energy-for-all
- Cameron, P. D., & Brothwood, M. (2002). Competition in energy markets: law and regulation in the European Union. Oxford University Press, USA.
- Beyca, O. F., Ervural, B. C., Tatoglu, E., Ozuyar, P. G., & Zaim, S. (2019). Using machine learning tools for forecasting natural gas consumption in the province of Istanbul. Energy Economics, 80, 937-949.
- Aydin, G. (2015). Forecasting natural gas production using various regression models. Petroleum Science and Technology, 33(15-16), 1486-1492.
- Al-Fattah, S. M., & Startzman, R. A., (2001). In SPE hydrocarbon economics and evaluation symposium, OnePetro.
- Nguyen-Le, V., Kim, M., Shin, H., & Little, E. (2021). Multivariate approach to the gas production forecast using early production data for Barnett shale reservoir. Journal of Natural Gas Science and Engineering, 87, 103776.
- Zheng, C., Wu, W. Z., Xie, W., & Li, Q. (2021). A MFO-based conformable fractional nonhomogeneous grey Bernoulli model for natural gas production and consumption forecasting. Applied Soft Computing, 99, 106891.
- Sen, D., Günay, M. E., & Tunç, K. M. (2019). Forecasting annual natural gas consumption using socio-economic indicators for making future policies. Energy, 173, 1106-1118.
- Zhang, W., & Yang, J. (2015). Forecasting natural gas consumption in China by Bayesian model averaging. Energy Reports, 1, 216-220.