SavedAdd your journal

Journals / Muş Alparslan Üniversitesi Mühendislik Mimarlık Fakültesi Dergisi / 2026 / Cilt: 7 Sayı: 1

Comparative Evaluation of AI and Statistical Models for Forecasting Fossil Fuel Electricity Generation

Fosil Yakıtla Elektrik Üretiminin Tahmini: Hibrit Yapay Zeka-İstatistiksel Çok Ülkeli Analiz

Zeydin Pala(MUŞ ALPARSLAN ÜNİVERSİTESİ)0000-0002-2642-7788
Pages
1–21
Publication Date
2026-06-28
DOI
—
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

The results indicate that forecasting performance varies considerably across countries and depends on the underlying characteristics of each time series.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:Forecasting fossil fuel-based electricity generation remains important for energy planning, particularly in countries undergoing different stages of energy transition.
📊
Key Findings & Data:This study investigates the forecasting performance of several statistical and machine-learning approaches using annual fossil fuel electricity generation data from six countries: Türkiye, Germany, the United Kingdom, France, Iran, and Ukraine.
💡
Conclusion & Impact:The results indicate that forecasting performance varies considerably across countries and depends on the underlying characteristics of each time series.
🔬Design:Modeling & Simulation
📍Scope:2022 Verileri
⚙️Variables:Enerji Geçişi · Hibrit Tahmin · Zaman Serisi Analizi

Abstract

Forecasting fossil fuel-based electricity generation remains important for energy planning, particularly in countries undergoing different stages of energy transition. Reliable forecasts can support decision-makers in balancing energy security concerns with long-term sustainability objectives. This study investigates the forecasting performance of several statistical and machine-learning approaches using annual fossil fuel electricity generation data from six countries: Türkiye, Germany, the United Kingdom, France, Iran, and Ukraine. The dataset covers the period from 1985 to 2022 and includes countries with distinct energy structures and policy trajectories. Nine forecasting models were evaluated, including traditional statistical techniques (NAÏVE, AUTO.ARIMA, HOLT-WINTERS, ETS, THETAF, and TBATS) and neural network-based methods (NNETAR, MLP, and ELM). Model performance was assessed using rolling validation strategies and three commonly used error measures: RMSE, MAE, and MAPE. The results indicate that forecasting performance varies considerably across countries and depends on the underlying characteristics of each time series. Neural network-based models generally performed better in countries exhibiting more complex or irregular generation patterns, whereas conventional statistical methods remained competitive for relatively stable series. Among the evaluated approaches, ELM achieved the lowest forecasting errors for France and Ukraine, while AUTO.ARIMA and ETS provided highly accurate results for Iran. Rather than identifying a universally superior forecasting technique, the findings highlight the importance of selecting models according to the structural properties of national energy systems. The study provides a comparative perspective on fossil fuel electricity forecasting and offers insights that may support future energy planning and transition strategies.

Özet

Fosil yakıta dayalı elektrik üretiminin doğru tahmin edilmesi, küresel enerji geçiş stratejileri için kritik bir zorluk olmaya devam ediyor; ancak mevcut yaklaşımlar genellikle hem karmaşık doğrusal olmayan dinamikleri hem de bölgeye özgü politika etkilerini yakalayacak metodolojik çok yönlülükten yoksundur. Bu çalışma, jeopolitik olarak farklı altı ülkeye (Türkiye, Almanya, Birleşik Krallık, Fransa, İran ve Ukrayna) gelişmiş uzun vadeli tahminler sunmak için gelişmiş derin öğrenme modellerini (MLP, ELM, NNETAR) yerleşik istatistiksel yöntemlerle (ARIMA, ETS, TBATS) sistematik olarak entegre eden yeni bir hibrit modelleme çerçevesi sunmaktadır. 1985-2022 yıllarını kapsayan yıllık verileri kullanarak, sürekli başlangıç ​​doğrulama stratejisi kullanıyoruz ve performansı RMSE, MAE ve MAPE ölçümleri aracılığıyla değerlendiriyoruz. Bulgularımız, derin öğrenme modellerinin değişken, doğrusal olmayan trendleri yakalamada istatistiksel muadillerinden sürekli olarak daha iyi performans gösterdiğini ortaya koyuyor (ELM'nin Fransa için %8,75'lik bir MAPE elde etmesi buna örnektir), AUTO.ARIMA gibi geleneksel modeller ise İran'da gösterildiği gibi istikrarlı rejimlerde etkili olmaya devam ediyor (MAPE: %2,40). Daha da önemlisi, bu araştırma yalnızca tahmin doğruluğunu geliştirmekle kalmıyor, aynı zamanda model performansını ulusal enerji geçiş yollarına bağlayarak eyleme dönüştürülebilir, politikayla ilgili bilgiler sağlıyor. Hibrit çerçevelerin özel karbondan arındırma stratejilerine nasıl rehberlik edebileceğini, şebeke istikrarını nasıl optimize edebileceğini ve ulusal enerji politikalarının küresel sürdürülebilirlik hedefleriyle uyumlaştırılmasını nasıl destekleyebileceğini, böylece dünya çapındaki politika yapıcılara ölçeklenebilir ve veri odaklı bir karar destek aracı sunabileceğini gösteriyoruz.

🔖

Cite This Article

Pala, Z. (2026). Comparative Evaluation of AI and Statistical Models for Forecasting Fossil Fuel Electricity Generation. Muş Alparslan Üniversitesi Mühendislik Mimarlık Fakültesi Dergisi, Cilt: 7 Sayı: 1, 1–21. https://ajindex.com/articles/1619979

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (44)

  1. S. Pfenninger and J. Keirstead, “Renewables, nuclear, or fossil fuels? Scenarios for Great Britain’s power system considering costs, emissions and energy security,” Appl. Energy, vol. 152, pp. 83–93, 2015, doi: 10.1016/j.apenergy.2015.04.102.
  2. M. Tunç, Ü. Çamdali, and C. Parmaksizoglu, “Comparison of Turkey’s electrical energy consumption and production with some European countries and optimization of future electrical power supply investments in Turkey,” Energy Policy, vol. 34, no. 1, pp. 50–59, 2006, doi: 10.1016/j.enpol.2004.04.027.
  3. S. Erat, A. Telli, O. M. Ozkendir, and B. Demir, “Turkey’s energy transition from fossil-based to renewable up to 2030: milestones, challenges and opportunities,” Clean Technol. Environ. Policy, vol. 23, no. 2, pp. 401–412, 2021, doi: 10.1007/s10098-020-01949-1.
  4. D. Ohlhorst, “Germany’s energy transition policy between national targets and decentralized responsibilities,” Journal of Integrative Environmental Sciences, vol. 12, no. 4, pp. 303–322, 2015, doi: 10.1080/1943815X.2015.1125373.
  5. M. Hossein Jahangir, E. Bazdar, and A. Kargarzadeh, “Techno-economic and environmental assessment of low carbon hybrid renewable electric systems for urban energy planning: Tehran-Iran,” City and Environment Interactions, vol. 16, no. June, p. 100085, 2022, doi: 10.1016/j.cacint.2022.100085.
  6. V. Stanytsina et al., “Fossil Fuel and Biofuel Boilers in Ukraine: Trends of Changes in Levelized Cost of Heat,” Energies (Basel)., vol. 15, no. 19, 2022, doi: 10.3390/en15197215.
  7. Z. Pala and F. Şevgin, “Statistical modeling for long-term meteorological forecasting: a case study in Van Lake Basin,” Natural Hazards, vol. 120, no. 2024, pp. 14101–14116, 2024, doi: 10.1007/s11069-024-06747-2.
  8. I. Tugal and F. Sevgin, “Analysis and forecastıng of temperature using time series forecasting methods A Case Study of Mus,” Thermal Science, vol. 27, no. 4, pp. 3081–3088, 2023, doi: 10.2298/TSCI2304081T.
  9. Z. Pala and İ. Tuğal, “Effect of frequency-dependent test length on prediction performance in monthly/quarterly time series analysis,” Knowl. Based. Syst., vol. 315, Apr. 2025, doi: 10.1016/j.knosys.2025.113294.
  10. M. Hossain, B. Rekabdar, S. J. Louis, and S. Dascalu, “Forecasting the weather of Nevada: A deep learning approach,” Proceedings of the International Joint Conference on Neural Networks, vol. 2015-Septe, pp. 1–6, 2015, doi: 10.1109/IJCNN.2015.7280812.
  11. E. Yaldız and Z. Pala, “Time Series Analysis of Radiological Data of Outpatients and Inpatients in Emergency Department of Mus State Hospital,” International Conference on Data Science, Machine Learning and Statistics - 2019 (DMS-2019), pp. 234–236, 2019.
  12. N. Srivastava, G. Hinton, A. Krizhevsky, I. Sutskever, and R. Salakhutdinov, “Dropout: A simple way to prevent neural networks from overfitting,” Journal of Machine Learning Research, vol. 15, pp. 1929–1958, 2014.
  13. J. H. Hwang and S. H. Yoo, “Energy consumption, CO2 emissions, and economic growth: Evidence from Indonesia,” Qual. Quant., vol. 48, no. 1, pp. 63–73, 2014, doi: 10.1007/s11135-012-9749-5.
  14. M. Mohsin, S. Naseem, M. Sarfraz, and T. Azam, “Assessing the effects of fuel energy consumption, foreign direct investment and GDP on CO2 emission: New data science evidence from Europe & Central Asia,” Fuel, vol. 314, no. December 2021, p. 123098, 2022, doi: 10.1016/j.fuel.2021.123098.
  15. T. Etem, Z. Pala, and I. Bozkurt, “Electromagnetic pollution measurement in the system rooms of a university,” in 2017 13th International Conference Perspective Technologies and Methods in MEMS Design, MEMSTECH 2017 - Proceedings, 2017. doi: 10.1109/MEMSTECH.2017.7937548.
  16. Z. Pala, İ. H. Ünlük, and E. Yaldız, “Forecasting of electromagnetic radiation time series: An empirical comparative approach,” Appl. Comput. Electromagn. Soc. J., vol. 34, no. 8, pp. 1238–1241, 2019.
  17. C. Chen, Z. Wu, S. Sun, P. Ban, Z. Ding, and Z. Xu, “Forecasting the ionospheric f 0 F 2 parameter one hour ahead using a support vector machine technique,” J. Atmos. Sol. Terr. Phys., vol. 72, no. 18, pp. 1341–1347, 2010, doi: 10.1016/j.jastp.2010.09.022.
  18. Z. Pala and R. Atici, “Forecasting Sunspot Time Series Using Deep Learning Methods,” Sol. Phys., vol. 294, no. 5, 2019, doi: 10.1007/s11207-019-1434-6.
  19. Z. Pala, “Comparative study on monthly natural gas vehicle fuel consumption and industrial consumption using multi-hybrid forecast models,” Energy, vol. 263, no. PC, pp. 1–21, 2023, doi: 10.1016/j.energy.2022.125826.
  20. N. Li, J. Wang, L. Wu, and Y. Bentley, “Predicting monthly natural gas production in China using a novel grey seasonal model with particle swarm optimization,” Energy, vol. 215, 2021, doi: 10.1016/j.energy.2020.119118.
  21. S. Dhillon, C. Madhu, D. Kaur, and S. Singh, “A Solar Energy Forecast Model Using Neural Networks: Application for Prediction of Power for Wireless Sensor Networks in Precision Agriculture,” Wirel. Pers. Commun., vol. 112, no. 4, pp. 2741–2760, 2020, doi: 10.1007/s11277-020-07173-w.
  22. R. J. Hyndman and G. Athanasopoulos, Forecasting: Principles and Practice, 2nd editio. Australia: Monash University, 2018.
  23. Z. Pala and A. F. Pala, “Comparison of ongoing COVID-19 pandemic confirmed cases / deaths weekly forecasts on continental basis using R statistical models,” Dicle University Journal of Engineering, vol. 4, pp. 635–644, 2021, doi: 10.24012/dumf.1002160.
  24. S. M. Basha, Y. Zhenning, D. S. Rajput, R. D. Caytiles, and N. C. S. N. Iyengar, “Comparative study on performance analysis of time series predictive models,” International Journal of Grid and Distributed Computing, vol. 10, no. 8, pp. 37–48, 2017, doi: 10.14257/ijgdc.2017.10.8.04.
  25. F. Petropoulos and I. Svetunkov, “A simple combination of univariate models,” Int. J. Forecast., vol. 36, no. 1, pp. 110–115, 2020, doi: 10.1016/j.ijforecast.2019.01.006.
  26. Y. Yang, W. Gao, and C. Guo, “Aero-engine lubricating oil metal content prediction using non-stationary time series ARIMA model,” Proceedings - 2017 10th International Symposium on Computational Intelligence and Design, ISCID 2017, vol. 2, no. 4, pp. 51–54, 2018, doi: 10.1109/ISCID.2017.173.
  27. S. Gelper, R. Fried, and C. Croux, “Robust forecasting with exponential and holt-winters smoothing,” J. Forecast., vol. 29, no. 3, pp. 285–300, 2010, doi: 10.1002/for.1125.
  28. M. Dong et al., “Deformation prediction of unstable slopes based on real-time monitoring and deepar model,” Sensors (Switzerland), vol. 21, no. 1, pp. 1–18, 2021, doi: 10.3390/s21010014.
  29. Z. Pala, “Examining EMF Time Series Using Prediction Algorithms With R,” vol. 44, no. 2, pp. 223–227, 2021.
  30. Z. H. Munim, “State-space TBATS model for container freight rate forecasting with improved accuracy,” Maritime Transport Research, vol. 3, no. November 2021, p. 100057, 2022, doi: 10.1016/j.martra.2022.100057.
  31. I. Bozkurt and Z. Pala, “Prediction of impact behavior of notched composite pipes repaired with different composite patches by machine learning algorithms,” J. Adhes. Sci. Technol., vol. 39, no. 10, pp. 1521–1558, 2025, doi: 10.1080/01694243.2025.2459718.
  32. Z. Pala, I. Bozkurt, and T. Etem, “Estimation of low frequency electromagnetic values using machine learning,” in 2017 13th International Conference Perspective Technologies and Methods in MEMS Design, MEMSTECH 2017 - Proceedings, 2017. doi: 10.1109/MEMSTECH.2017.7937550.
  33. I. Bozkurt, Z. Pala, and T. Etem, “Measurement and evaluation of low frequency electromagnetic field in camera observation rooms,” in 2017 13th International Conference Perspective Technologies and Methods in MEMS Design, MEMSTECH 2017 - Proceedings, 2017. doi: 10.1109/MEMSTECH.2017.7937554.
  34. Z. Pala and A. F. Pala, “Perform Time-series Predictions in the R Development Environment by Combining Statistical-based Models with a Decomposition-based Approach,” Muş Alparslan University Journal of Engineering and Architecture Faculty, vol. 1, no. 1, pp. 55–63, 2020.
  35. Y. Liao and C. Liang, “A Temperature Time Series Forecasting Model Based on DeepAR,” 2021 7th International Conference on Computer and Communications, ICCC 2021, pp. 1588–1593, 2021, doi: 10.1109/ICCC54389.2021.9674623.
  36. A. Luis, S. Maia, F. D. A. T. De Carvalho, and T. B. Ludermir, “Forecasting models for interval-valued time series,” 2008, doi: 10.1016/j.neucom.2008.02.022.
  37. Y. Park and H. S. Yang, “Convolutional neural network based on an extreme learning machine for image classification,” Neurocomputing, vol. 339, pp. 66–76, 2019, doi: 10.1016/j.neucom.2018.12.080.
  38. X. Chen, Z. Y. Dong, K. Meng, Y. Xu, K. P. Wong, and H. W. Ngan, “Electricity price forecasting with extreme learning machine and bootstrapping,” IEEE Transactions on Power Systems, vol. 27, no. 4, pp. 2055–2062, 2012, doi: 10.1109/TPWRS.2012.2190627.
  39. Z. Pala and İ. H. Ünlük, “Comparison of hybrid and non-hybrid models in short-term predictions on time series in the R development environment,” DÜMF Mühendislik Dergisi, vol. 2, pp. 199–204, 2022, doi: 10.24012/dumf.1079230.
  40. S. Kim and H. Kim, “A new metric of absolute percentage error for intermittent demand forecasts,” Int. J. Forecast., vol. 32, no. 3, pp. 669–679, 2016, doi: 10.1016/j.ijforecast.2015.12.003.
  41. Z. Pala, R. Atıcı, and E. Yaldız, “Forecasting Future Monthly Patient Volume using Deep Learning and Statistical Models,” Wirel. Pers. Commun., vol. 130, no. 2, pp. 1479–1502, 2023, doi: 10.1007/s11277-023-10341-3.
  42. C. J. Willmott and K. Matsuura, “Advantages of the mean absolute error (MAE) over the root mean square error (RMSE) in assessing average model performance,” Clim. Res., vol. 30, no. 1, pp. 79–82, Dec. 2005, doi: 10.3354/cr030079.
  43. H. Li, Y. Yang, Y. Cheng, Y. Jin, H. Luo, and L. Zhang, “Application of Time Series Model in Relative Humidity Prediction,” J. Phys. Conf. Ser., vol. 1584, no. 1, 2020, doi: 10.1088/1742-6596/1584/1/012017.
  44. R. J. Hyndman and G. Athanasopoulos, Forecasting: Principles and Practice, 2nd editio. Australia: Monash University, 2018.