Histogram tabanlı gradyan artırma yöntemi ile Türkiye için kısa vadeli elektrik talep tahmini

Abstract

Elektrik enerjisi, sürdürülebilir ekonomik büyüme ve sosyal refah için stratejik bir üretim faktörüdür. Elektrik talebinin yüksek doğrulukla tahmini, enerji arz-talep dengesinin optimizasyonu ve şebeke güvenilirliği açısından kritik öneme sahiptir. Bu çalışmada, Türkiye elektrik piyasasının saatlik talep tahmini, 1 Aralık 2019 - 1 Mart 2025 dönemini kapsayan veri seti üzerinde HGBR ile gerçekleştirilmiştir. Zamansal gecikmeler, hareketli istatistikler, takvimsel ve mevsimsel özellikler ile çoklu enerji üretim verilerini içeren kapsamlı bir öznitelik mühendisliği uygulanmıştır. Beş katlı zaman serisi çapraz doğrulama ve bağımsız test sonucunda 0.988 R² ve %1.13 MAPE ile istatistiksel olarak anlamlı bir tahmin performansı elde edilmiştir. HGBR; XGBoost, LGBM, RF, KNN, SVM, AdaBoost, LSTM, CNN ve GRU gibi yöntemlere kıyasla üstün performans göstermiş ve histogram tabanlı optimizasyon tekniklerinin elektrik talebi tahmininde etkinliğini ortaya koymuştur. Ayrıca önerilen model, derin öğrenme modellerine kıyasla çok daha kısa eğitim süresiyle yüksek doğruluk sağlayarak gerçek zamanlı operasyonel uygulamalar için pratik bir çözüm sunmaktadır.

Keywords

• Makine öğrenmesi
• elektrik üretimi
• kısa vadeli tahmin

Demand forecasting for Türkiye using histogram-based gradient boosting method

Abstract

Electric energy is a strategic production factor for sustainable economic growth and social welfare. Accurate forecasting of electricity demand is critical for optimizing the supply-demand balance and ensuring grid reliability. In this study, hourly electricity demand forecasting for the Turkish electricity market was conducted using HGBR on a dataset spanning from December 1, 2019, to March 1, 2025. A comprehensive feature engineering process was applied, incorporating temporal lags, moving statistical features, calendar and seasonal components, and multi-source energy production data. Experimental evaluation using five-fold Time Series Split (TSS) cross-validation and an independent test set achieved a statistically significant forecasting performance with an R² of 0.988 and a Mean Absolute Percentage Error (MAPE) of 1.13%. The HGBR method demonstrated superior performance compared to XGBoost, LGBM, RF, KNN, SVM, AdaBoost, LSTM, CNN, and GRU methods, revealing the effectiveness of histogram-based optimization techniques in electricity demand forecasting. Furthermore, the proposed model provides high accuracy with significantly shorter training times compared to deep learning models, offering a practical solution for real-time operational applications.

Keywords

• Machine learning
• electricity generation
• short-term forecasting

References

1. W. Gao, A. Darvishan, M. Toghani, M. Mohammadi, O. Abedinia, and N. Ghadimi, “Different states of multi-block based forecast engine for price and load prediction,” International Journal of Electrical Power & Energy Systems, 104, pp. 423-435, 2019. https://doi.org/10.1016/j.ijepes.2018.07.014.
2. H. Nhat-Duc and T. Van-Duc, “Comparison of histogram-based gradient boosting classification machine, random Forest, and deep convolutional neural network for pavement raveling severity classification,” Automation in Construction, 148, p. 104767, 2023. https://doi.org/10.1016/j.autcon.2023. 104767.
3. K. Kaysal, E. Akarslan, and F. O. Hocaoğlu, “Türkiye kısa dönem elektrik yük talep tahmininde makine öğrenmesi yöntemlerinin karşılaştırılması,” Bilecik Şeyh Edebali Üniversitesi Fen Bilimleri Dergisi, 9 (2), pp. 693-702, 2022. https://doi.org/10.35193/bseufbd .1004827.
4. C. E. Velasquez, M. Zocatelli, F. B. Estanislau, and V. F. Castro, “Analysis of time series models for Brazilian electricity demand forecasting,” Energy, 247, p. 123483, 2022. https://doi.org/10.1016/j.energy.2022.123483.
5. E. Sezer, G. Yıldırım, and M. T. Özdemir, “Türkiye’deki Toplam Elektrik Tüketiminin Holt-Winters Üstel Düzleştirme Yöntemi İle Tahmini,” Researcher, 4 (2), pp. 1-6, 2024.
6. J. F. Torres, F. Martínez-Álvarez, and A. Troncoso, “A deep LSTM network for the Spanish electricity consumption forecasting,” Neural Computing and Applications, 34 (13), pp. 10533-10545, 2022. https://doi.org/10.1007/s00521-021-06773-2.
7. A. K. Shaikh, A. Nazir, N. Khalique, A. S. Shah, and N. Adhikari, “A new approach to seasonal energy consumption forecasting using temporal convolutional networks,” Results in Engineering, 19, p. 101296, 2023. https://doi.org/10.1016/j.rineng. 2023.101296.
8. L. Alsmadi, G. Lei, and L. Li, “Forecasting Day-Ahead Electricity Demand in Australia Using a CNN-LSTM Model with an Attention Mechanism,” Applied Sciences, 15 (7), p. 3829, 2025. https://doi.org/ 10.3390/app15073829.

9. T. G. Grandón, J. Schwenzer, T. Steens, and J. Breuing, “Electricity demand forecasting with hybrid classical statistical and machine learning algorithms: Case study of Ukraine,” Applied Energy, 355, p. 122249, 2024. https://doi.org/10.1016/j.apenergy.20 23.122249.
10. S. Zhao, Z. Xu, Z. Zhu, X. Liang, Z. Zhang, and R. Jiang, “Short and Long-Term Renewable Electricity Demand Forecasting Based on CNN-Bi-GRU Model,” ICCK Transactions on Emerging Topics in Artificial Intelligence, 2 (1), pp. 1-15, 2025. https://doi.org/ 10.62762/TETAI.2024.532253.
11. Z. Wang, Z. Chen, Y. Yang, C. Liu, X. Li, and J. Wu, “A hybrid autoformer framework for electricity demand forecasting,” Energy Reports, 9, pp. 3800-3812, 2023. https://doi.org/10.1016/j.egyr.2023.02.0 83.
12. S. Reddy, S. Akashdeep, R. Harshvardhan, and S. Kamath, “Stacking Deep learning and Machine learning models for short-term energy consumption forecasting,” Advanced Engineering Informatics, 52, p. 101542, 2022. https://doi.org/10.1016/j.aei.2022.1 01542.
13. S. Albahli, “LSTM vs. Prophet: Achieving Superior Accuracy in Dynamic Electricity Demand Forecasting,” Energies, 18 (2), p. 278, 2025. https:// doi.org/ 10.3390/en18020278.
14. H. A. Issa, H. H. Çevik, A. Yilmaz, and M. Cunkas, “A genetic programming-based ensemble method for long-term electricity demand forecasting,” PeerJ Computer Science, 11, p. e2825, 2025. https://doi.org /10.7717/peerj-cs.2825.
15. J. Adinkrah et al., “Artificial intelligence-based strategies for sustainable energy planning and electricity demand estimation: A systematic review,” Renewable and Sustainable Energy Reviews, vol. 210, p. 115161, 2025. https://doi.org/10.1016/j.rser.20 24.115161.
16. “Transparency Platform of the Energy Exchange Istanbul (EPİAŞ).” [Online]. Available: seffaflik.epias.com.tr
17. M. K. SB and J. K. AR, “Weather forecasting using application programming interface,” presented at the 2023 International Conference on Recent Advances in Science and Engineering Technology (ICRASET), IEEE, 2023, pp. 1-6. https://doi.org/10.1109/ ICRASET59632.2023.10420308.
18. M. Özalp, “Türkiye’nin yenilenebilir enerji görünümü ve geleceği,” Avrasya Etüdleri, 64, pp. 39-75, 2025.
19. Y. Nohara, K. Matsumoto, H. Soejima, and N. Nakashima, “Explanation of machine learning models using shapley additive explanation and application for real data in hospital,” Computer Methods and Programs in Biomedicine, 214, p. 106584, 2022. https://doi.org/10.1016/j.cmpb.2021.106584.
20. S. Developers, “Histogram-based Gradient Boosting Classification Tree,” Scikit-learn documentation, 2023.
21. J. Jin, C. Yang, and Y. Zhang, “An improved CMA-ES for solving large scale optimization problem,” presented at the Advances in Swarm Intelligence: 11th International Conference, ICSI 2020, Belgrade, Serbia, July 14-20, 2020, Proceedings 11, Springer, 2020, pp. 386-396. https://doi.org/10.1007/978-3-030-53956-6_34.
22. C. Bergmeir and J. M. Benítez, “On the use of cross-validation for time series predictor evaluation,” Information Sciences, 191, pp. 192-213, 2012. https:// doi.org/10.1016/j.ins.2011.12.028.
23. D. Chicco, M. J. Warrens, and G. Jurman, “The coefficient of determination R-squared is more informative than SMAPE, MAE, MAPE, MSE and RMSE in regression analysis evaluation,” Peerj computer science, 7, p. e623, 2021. https://doi.org/ 10.7717/peerj-cs.623.
24. 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,” Climate research, 30 (1), pp. 79-82, 2005. https://doi.org/10.3354/cr030079.
25. T. Chai and R. R. Draxler, “Root mean square error (RMSE) or mean absolute error (MAE)?-Arguments against avoiding RMSE in the literature,” Geoscientific model development, 7 (3), pp. 1247-1250, 2014. https://doi.org/10.5194/gmd-7-1247-20 14.
26. K. Serefoglu Cabuk et al., “Chasing the objective upper eyelid symmetry formula; R2, RMSE, POC, MAE, and MSE,” International Ophthalmology, 44 (1), p. 303, 2024. https://doi.org/10.1007/s10792-024-03157-y.
27. M. Shanmugavalli and K. M. J. Ignatia, “Comparative Study among MAPE, RMSE and R Square over the Treatment Techniques Undergone for PCOS Influenced Women,” Recent Patents on Engineering, 19 (1), p. E041223224190, 2025. https://doi.org/ 10.2174/0118722121269786231120122435.