Beyond forecast accuracy: A statistical and financial evaluation of machine learning models for small hydropower forecasting

Öz

This study develops and evaluates a machine learning-based forecasting framework for improving the day-ahead production estimates of a run-of-river hydropower plant operating under Turkish electricity market conditions. The analysis is based on hourly operational data and compares operator-based heuristic forecasts with four alternative modeling approaches: Multilayer Perceptron (MLP), Random Forest (RF), standard XGBoost, and a cost-aware XGBoost (CA-XGBoost) formulation. To reflect real-world operational constraints, the framework relies on a parsimonious feature structure composed of the operator’s original forecast and a proxy variable derived from the upstream plant’s generation data with a 16-hour hydraulic lag. In this way, the study aims to provide a practical forecasting approach for data-constrained hydropower settings where detailed meteorological inputs may not be readily available. The models are evaluated through both statistical and financial criteria. Statistical performance is assessed using MAE, RMSE, and (R²), while financial performance is examined through imbalance-related market costs calculated using hourly Market Clearing Price and System Marginal Price data under the asymmetric settlement structure of the Turkish electricity market. The results show that Random Forest achieved the best overall economic performance, yielding the lowest total imbalance cost and the lowest MAE, whereas MLP produced the best RMSE and (R²) values. By contrast, the CA-XGBoost approach model did not outperform the benchmark machine learning models, although it did alter the directional structure of forecast errors. The findings reveal that improvements in statistical accuracy do not necessarily imply superior financial performance. Accordingly, the study demonstrates the importance of evaluating hydropower forecasting models not only in terms of conventional error metrics but also with respect to their economic consequences under market-based imbalance pricing. From a practical perspective, the CA- XGBoost proxy-based framework offers a lightweight and potentially scalable decision-support approach for small hydropower plants operating under limited data availability.

Anahtar Kelimeler

• Hydropower generation
• Machine learning
• Imbalance cost
• Turkish electricity market

Kaynakça

1. [1] Klein SJW, Fox ELB. A review of small hydropower performance and cost. Renewable and Sustainable Energy Reviews 2022; 169: 112898.
2. [2] Rospriandana N, Burke PJ, Suryani A, Mubarok MH, Pangestu MA. Over a century of small hydropower projects in Indonesia: a historical review. Energy, Sustainability and Society 2023; 30(1): 30.
3. [3] Zhang Y, Davis D, Brear MJ. Least-cost pathways to net-zero, coupled energy systems: A case study in Australia. Journal of Cleaner Production 2023; 392: 136266.
4. [4] Du Y, Guo Y, Wang P, Lyu X, Xu Z, Ouyang Y, et al. Pumped Hydro Storage Multi-Market Trading Strategy Considering Electricity Price Forecast Errors. 4th International Conference on Energy Engineering and Power Systems (EEPS); 2024.
5. [5] Wang J, Zhou Y, Zhang Y, Lin F, Wang J. Risk-averse optimal combining forecasts for renewable energy trading under CVaR assessment of forecast errors. IEEE Transactions on Power Systems 2023; 39(1): 2296-2309.
6. [6] Gandhi O, Zhang W, Kumar DS, Rodriguez-Gallegos CD, Yagli GM, Yang D, et al. The value of solar forecasts and the cost of their errors: A review. Renewable and Sustainable Energy Reviews 2024; 189: 113915.
7. [7] Ilseven E, Gol M. Hydro-optimization-based medium-term price forecasting considering demand and supply uncertainty. IEEE Transactions on Power Systems 2017; 33(4): 4074-4083.
8. [8] Aydogdu A, Tor OB, Guven AN. CVaR-based stochastic wind-thermal generation coordination for Turkish electricity market. Journal of Modern Power Systems and Clean Energy 2019; 7(5): 1307-1318.

9. [9] Buke B, Sayin M, Tanrisever F. A dynamic multi-level iterative algorithm for clearing European electricity day-ahead markets: An application to the Turkish market. Journal of the Operational Research Society 2024; 75(4): 784-798.
10. [10] Matsumoto T, Bunn D, Yamada Y. Mitigation of the inefficiency in imbalance settlement designs using day-ahead prices. IEEE Transactions on Power Systems 2021; 37(5): 3333-3345.
11. [11] Gunduz, S, Ugurlu U, Oksuz I. Data Imputation for Electricity Price Forecasting in the Intraday Market. Avrupa Bilim ve Teknoloji Dergisi 2021; (25): 334-340.
12. [12] Artantas C. Green Electricity Promotion in Turkey. In: Promotion of Green Electricity in Germany and Turkey: A Comparison with Reference to the WTO and EU Law. Cham: Springer Nature Switzerland; 2023. pp. 169-187.
13. [13] Emre T. Renewable Energy Transition in Turkiye: The Impact of Solar and Wind Based Capacity Growth on Market Prices and Cost Dynamics. Balkan Journal of Electrical and Computer Engineering 2025; 13(3): 338-345.
14. [14] Selcuk O, Acar B, Dastan SA. System integration costs of wind and hydropower generations in Turkey. Renewable and Sustainable Energy Reviews 2022; 156: 111982.
15. [15] Ozozen A, Kayalica MO, Kayakutlu G. Day-ahead power market behavior for a small supplier: case of Turkish market. Environmental Economics 2018; 9(2): 70-79.
16. [16] Singh U, Vadhera S. Random forest and xgboost technique for short-term load forecasting. 1st International Conference on Sustainable Technology for Power and Energy Systems (STPES); 2022.
17. [17] Maciejewski D, Mudryk K, Sporysz M. Forecasting Electricity Production in a Small Hydropower Plant (SHP) Using Artificial Intelligence (AI). Energies 2024; 17(24): 6401.
18. [18] Arifoglu A, Kandemir T. Electricity Price Forecasting in Turkish Day-Ahead Market via Deep Learning Techniques. Mehmet Akif Ersoy University Journal of Economics and Administrative Sciences Faculty 2022; 9(2): 1433-1458.
19. [19] Yilan Y, Beykent A. Day-Ahead Electricity Price Forecasting Using the XGBoost Algorithm: An Application to the Turkish Electricity Market. Computers, Materials & Continua 2026; 86(1): 1.
20. [20] Atalay BA, Zor K. An Innovative Approach for Forecasting Hydroelectricity Generation by Benchmarking Tree-Based Machine Learning Models. Applied Sciences 2025; 15(19): 10514.
21. [21] Bulut M. Improving deep learning forecasting model based on LSTM for Türkiye's hydro-electricity generation. Sakarya University Journal of Computer and Information Sciences 2024; 7(3): 325-337.
22. [22] Atalan YA, Sahin H, Keskin A, Atalan A. Strategic forecasting of renewable energy production for sustainable electricity supply: A machine learning approach considering environmental, economic, and oil factors in Turkiye. PLOS ONE 2025; 20(8): e0328290.
23. [23] Siridhipakul C, Vateekul P. Enhance Attentional LSTM Models for Power Consumption Forecasting Using Asymmetric Loss and Renewable Energy Factors. 8th International Conference on Computer and Communications Management; 2020.
24. [24] Ghanim J, Issa M, Awad M. An asymmetric loss with anomaly detection LSTM framework for power consumption prediction. 21st Mediterranean Electrotechnical Conference (MELECON); 2022.
25. [25] Lin JL, Zhang Y, Zhu K, Chen B, Zhang F. Asymmetric Loss Functions for Contract Capacity Optimization. Energies 2020; 13(12): 3123.
26. [26] Wang Z, Mae M, Yamane T, Ajisaka M, Nakata T, Matsuhashi R. Novel Custom Loss Functions and Metrics for Reinforced Forecasting of High and Low Day-Ahead Electricity Prices Using Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) and Ensemble Learning. Energies 2024; 17(19): 4885.
27. [27] Polat E, Guler MG, Ulukus MY. Renewable GenCo bidding strategy using newsvendor-based neural networks: An example from Turkish electricity market. Electric Power Systems Research 2024; 231: 110301.
28. [28] Song Y, Yoon Y, Jin Y. Impact Analysis of Price Cap on Bidding Strategies of VPP Considering Imbalance Penalty Structures. Energies 2025; 18(15): 3927.
29. [29] Shang Y, Xu Y, Shang L, Fan Q, Wang Y, Liu Z. A method of direct, real-time forecasting of downstream water levels via hydropower station reregulation: A case study from Gezhouba Hydropower Plant, China. Journal of Hydrology 2019; 573: 895-907.
30. [30] Lin R, Zhou Z, You S, Rao R, Kuo CCJ. Geometrical interpretation and design of multilayer perceptrons. IEEE Transactions on Neural Networks and Learning Systems 2022; 35(2): 2545-2559.
31. [31] Wang S, Hasan M, Lu M. Global sensitivity analysis methodology for construction simulation models: multiple linear regressions versus multilayer perceptions. Journal of Construction Engineering and Management 2024; 150(5): 04024035.
32. [32] Dralus G, Kusznier MDJ, Dralus J. Application of artificial intelligence algorithms in multilayer perceptron and Elman networks to predict photovoltaic power plant generation. Energies 2023; 16(18): 6697.
33. [33] Mei J, Wang C, Luo S, Xu W, Deng Z. Short-term wind power prediction based on encoder-decoder network and multi-point focused linear attention mechanism. Sensors 2024; 24(17): 5501.
34. [34] Gonzalez C, Mira-McWilliams J, Juarez I. Important variable assessment and electricity price forecasting based on regression tree models: Classification and regression trees, Bagging and Random Forests. IET Generation, Transmission & Distribution 2015; 9(11): 1120-1128.
35. [35] Wang P, Xu K, Ding Z, Du Y, Liu W, Sun B, et al. An online electricity market price forecasting method via random forest. IEEE Transactions on Industry Applications 2022; 58(6): 7013-7021.
36. [36] Dudek G. A comprehensive study of random forest for short-term load forecasting. Energies 2022; 15(20): 7547.
37. [37] Magalhaes B, Bento P, Pombo J, Calado MDR, Mariano S. Short-term load forecasting based on optimized random forest and optimal feature selection. Energies 2024; 17(8): 1926.
38. [38] Beloev HI, Saitov SR, Filimonova AA, Chichirova ND, Babikov OE, Iliev IK. Short-Term Electrical Load Forecasting Based on XGBoost Model. Energies 2025; 18(19): 5144.
39. [39] Algarvio, H, Couto A, Estanqueiro A. A double pricing and penalties separated imbalance settlement mechanism to incentive self balancing of market parties. 20th International Conference on the European Energy Market (EEM); 2024.
40. [40] EPIAS. MCP, SMP and Imbalance Price Listing. [Online] Available at: https://seffaflik.epias.com.tr/electricity/electricity-market-reports/mcp-smp-and-imbalance-price-listing (Last Access: 24/02/2026).
41. [41] EPIAS. 2024 Faaliyet Raporu. 2024. [Online] Available at: https://www.epias.com.tr/epias-kurumsal/faaliyet-raporlarimiz/ (Last Access: 24/02/2026).