Predictive Analysis of Monthly Electricity Consumption Using Rule-Based and Machine Learning Models

Abstract

Accurate monthly electricity consumption (EC) forecasting is essential for power providers to allocate resources efficiently, develop reasonable sales plans, and support the creation of reliable smart grids and precise demand-side management policies. Given that factors such as climate, population, and economic conditions can significantly impact EC, it is crucial to consider a wide range of variables in medium-term EC forecasts. This paper addresses a gap in the existing literature by evaluating the performance of the M5 rule model—a relatively underutilized technique—in comparison with popular machine learning (ML) models like Random Forest (RF) and Support Vector Machine (SVM). The motivation for selecting the M5 rule regression technique stems from its effective feature selection process, which is simpler and more straightforward than the complex feature selection methods employed by other models. Using an aggregated dataset from the Czech Transmission System Operator, the study applies these three regression techniques independently to forecast monthly EC. The results demonstrate that the M5 rule regression model outperforms both SVM and RF models for monthly forecasts, achieving an impressive correlation coefficient (R²) value of 0.9063, compared to 0,8915 for SVM and 0,8598 for RF. Additionally, the M5 rule achieves the lowest Mean Absolute Error (MAE) of 16772.29, compared to 17477.57 for SVM and 21390.68 for RF, as well as the lowest Root Mean Squared Error (RMSE) of 22287.94, compared to 23114.17 for SVM and 26658.89 for RF. Furthermore, M5 rule shows superior performance in terms of relative errors, with a Relative Absolute Error (RAE) of 43.12% and a Relative Root Mean Squared Error (RRSE) of 45.74%, while RF and SVM show higher values. The M5 rule model also identifies air temperature, relative humidity, and clear sky surface irradiance as the most influential features in predicting EC. These findings offer valuable implications for power management companies, aiding in the strategic planning of power generation and supply. By accurately forecasting EC and understanding key influencing factors, companies can better avoid issues of overproduction or shortages, leading to more efficient and reliable power management.

Keywords

• M5 rule
• SVM
• RF
• tree-based regression
• monthly EC forecasting

Kural Tabanlı ve Makine Öğrenimi Modelleri ile Aylık Elektrik Tüketiminin Tahmine Dayalı Analizi

Abstract

Aylık elektrik tüketiminin (ET) doğru tahmin edilmesi, enerji sağlayıcıların kaynakları verimli tahsis etmesi, gerçekçi satış planları geliştirmesi ve güvenilir akıllı şebekeler ile doğru talep tarafı yönetim politikaları oluşturması açısından kritik öneme sahiptir. İklim, nüfus ve ekonomik koşullar gibi etkenlerin ET’yi önemli ölçüde etkileyebilmesi nedeniyle, orta vadeli tahminlerde çok çeşitli değişkenlerin dikkate alınması gereklidir. Bu çalışma, literatürde nispeten az kullanılan M5 kural modelinin performansını, yaygın makine öğrenimi (MÖ) tekniklerinden Rastgele Orman (RO) ve Destek Vektör Makinesi (DVM) ile karşılaştırarak bu alandaki boşluğu ele almaktadır. M5 kural regresyonunun tercih edilmesinin nedeni, diğer modellerdeki karmaşık özellik seçimi yöntemlerine kıyasla daha basit ve doğrudan bir özellik seçimi süreci sunmasıdır. Çekya İletim Sistemi Operatöründen elde edilen toplulaştırılmış veri kullanılarak üç model ayrı ayrı uygulanmıştır. Bulgular, M5 kural modelinin aylık tahminlerde RO ve DVM’den daha iyi sonuçlar verdiğini göstermektedir (R² = 0,9063). Ayrıca, Ortalama Mutlak Hata (OMH), Kök Ortalama Kare Hata (KOKH) ve göreli hata ölçütlerinde de en düşük değerlere ulaşılmıştır. M5 modeli, hava sıcaklığı, bağıl nem ve açık gökyüzü yüzey ışınımını en etkili değişkenler olarak belirlemiştir. Bu sonuçlar, enerji yönetiminde stratejik planlama için önemli çıkarımlar sunmaktadır.

Keywords

• M5 Kuralı
• DVM
• RO
• Ağaç-Tabanlı Regression

References

1. [1] A. Cabrera, L. G. B. Ruiz, D. Criado-Ramón, C. D. Barranco, and M. C. Pegalajar, Application of fuzzy and conventional forecasting techniques to predict energy consumption in buildings, Int. J. Intell. Syst., vol. 2023, 2023, Art. no. 4391555, pp. 1–12.
2. [2] M. Jawad et al., Genetic algorithm-based non-linear auto-regressive with exogenous inputs neural network short-term and medium-term uncertainty modelling and prediction for electrical load and wind speed, J. Eng., 2018, pp. 721–729.
3. [3] S.-M. Jung, S. Park, S.-W. Jung, and E. Hwang, Monthly electric load forecasting using transfer learning for smart cities, Sustainability, vol. 12, no. 15, p. 6364, 2020.
4. [4] S. R. Khuntia, J. L. Rueda, and M. A. van der Meijden, Forecasting the load of electrical power systems in mid- and long-term horizons: A review, IET Gener. Transm. Distrib., vol. 10, no. 15, pp. 3971–3977, 2016.
5. [5] T. Gao, D. Niu, Z. Ji, and L. Sun, Mid-term electricity demand forecasting using improved variational mode decomposition and extreme learning machine optimized by sparrow search algorithm, Energy, vol. 261, p. 125328, 2022.
6. [6] N. B. Behmiri, C. Fezzi, and F. Ravazzolo, Incorporating air temperature into mid-term electricity load forecasting models using time-series regressions and neural networks, Energy, vol. 278, p. 127831, 2023.
7. [7] L. Di Persio and N. Fraccarolo, Energy consumption forecasts by gradient boosting regression trees, Mathematics, vol. 11, no. 4, p. 1068, 2023.
8. [8] A. Mystakidis et al., Energy forecasting: A comprehensive review of techniques and technologies, Energies, vol. 17, no. 6, p. 1662, 2024.

9. [9] R. Srivastava, A. N. Tiwari, and V. K. Giri, Solar radiation forecasting using MARS, CART, M5, and random forest model: A case study for India, Heliyon, vol. 5, no. 11, Art. no. e02692, 2019.
10. [10] P. Geurts, D. Ernst, and L. Wehenkel, Extremely randomized trees, Mach. Learn., vol. 63, no. 1, pp. 3–42, 2006.
11. [11] H. Wei-Chiang, Electric load forecasting by support vector model, Appl. Math. Model., vol. 33, no. 5, pp. 2444–2454, 2009.
12. [12] P. Bunnoon, K. Chalermyanont, and C. Limsakul, Mid term load forecasting of the country using statistical methodology: Case study in Thailand, in Proc. Int. Conf. Signal Process. Syst., vol. 1, 2009, pp. 924–928.
13. [13] R. Hyndman, A. B. Koehler, J. K. Ord, and R. D. Snyder, Forecasting with Exponential Smoothing: The State Space Approach. Berlin: Springer, 2008.
14. [14] D. Oliveira, E. Meira, C. Oliveira, and F. Luiz, Forecasting mid-long term electric energy consumption through bagging ARIMA and exponential smoothing methods, Energy, vol. 144, pp. 776–788, 2018.
15. [15] C. Hamzaçebi, H. A. Es, and R. Çakmak, Forecasting of Turkey’s monthly electricity demand by seasonal artificial neural network, Neural Comput. Appl., vol. 31, pp. 2217–2231, 2019.
16. [16] B. Jin and X. Xu, Wholesale price forecasts of green grams using the neural network, Asian J. Econ. Bank., 2024.
17. [17] Q. Ji et al., Short- and medium-term power demand forecasting with multiple factors based on multi-model fusion, Mathematics, vol. 10, p. 2148, 2022.
18. [18] B. Jin and X. Xu, Price forecasting through neural networks for crude oil, heating oil, and natural gas, Meas.: Energy, vol. 1, no. 1, p. 100001, 2024.
19. [19] T. Ahmad and H. Chen, Potential of three variant machine-learning models for forecasting district level medium-term and long-term energy demand in smart grid environment, Energy, vol. 160, pp. 1008–1020, 2018.
20. [20] B. Jin and X. Xu, Carbon emission allowance price forecasting for China Guangdong carbon emission exchange via the neural network, Global Finance Rev., vol. 6, no. 1, p. 3491, 2024.
21. [21] M. H. L. Lee et al., A comparative study of forecasting electricity consumption using machine learning models, Mathematics, vol. 10, p. 1329, 2022.
22. [22] E. Barakat, Modeling of nonstationary time-series data. Part II. Dynamic periodic trends, Int. J. Electr. Power Energy Syst., vol. 23, pp. 63–68, 2001.
23. [23] F. Apadula, A. Bassini, A. Elli, and S. Scapin, Relationships between meteorological variables and monthly electricity demand, Appl. Energy, vol. 98, pp. 346–356, 2012.
24. [24] B. Jin and X. Xu, Forecasting wholesale prices of yellow corn through the Gaussian process regression, Neural Comput. Appl., vol. 36, no. 15, pp. 8693–8710, 2024.
25. [25] B. Jin and X. Xu, Predictions of steel price indices through machine learning for the regional northeast Chinese market, Neural Comput. Appl., vol. 36, no. 33, pp. 20863–20882, 2024.
26. [26] B. Jin and X. Xu, Pre-owned housing price index forecasts using Gaussian process regressions, J. Model. Manag., 2024.
27. [27] M. De Felice, A. Alessandri, and F. Catalano, Seasonal climate forecasts for medium-term electricity demand forecasting, Appl. Energy, vol. 137, pp. 435–444, 2015.
28. [28] Z. Hu, Y. Bao, R. Chiong, and T. Xiong, Mid-term interval load forecasting using multioutput support vector regression with a memetic algorithm for feature selection, Energy, vol. 84, pp. 419–431, 2015.
29. [29] O. Yuksel, M. Bayraktar, and M. Sokukcu, Comparative study of machine learning techniques to predict fuel consumption of a marine diesel engine, Ocean Eng., vol. 286, p. 115505, 2023.
30. [30] B. Singh and A. K. Jana, Forecast of agri-residues generation from rice, wheat and oilseed crops in India using machine learning techniques, Environ. Res., vol. 245, p. 117993, 2024.
31. [31] A. Srivastav et al., Predictive analysis of recycled concrete properties at elevated temperatures using M5 pruned rule classifiers, Asian J. Civ. Eng., vol. 25, pp. 2623–2640, 2024.
32. [32] N. T. Ngo et al., An ensemble machine learning model for enhancing the prediction accuracy of energy consumption in buildings, Arab. J. Sci. Eng., vol. 47, pp. 4105–4117, 2022.
33. [33] A. K. Srivastava et al., A day-ahead short-term load forecasting using M5P machine learning algorithm along with elitist genetic algorithm and random forest-based hybrid feature selection, Energies, vol. 16, p. 867, 2023.
34. [34] M. Lydia and G. Edwin Prem Kumar, Soft computing models for forecasting day-ahead energy consumption, Mater. Today Proc., vol. 58, pp. 473–477, 2022.
35. [35] S. A. Lesik, Applied Statistical Inference with MINITAB, 2nd ed. U.S.: Chapman and Hall, 2018.
36. [36] S. Raschka and V. Mirjalili, Python Machine Learning: Machine Learning and Deep Learning with Python, scikit-learn, and TensorFlow, 2nd ed. UK: Packt Publishing, 2017.
37. [37] A. S. Ahmad et al., A review on applications of ANN and SVM for building electrical energy consumption forecasting, Renew. Sustain. Energy Rev., vol. 33, pp. 102–109, 2014.
38. [38] L. Lang, L. Tiancai, A. Shan, and T. Xiangyan, An improved random forest algorithm and its application to wind pressure prediction, Int. J. Intell. Syst., vol. 36, no. 8, pp. 4016–4032, 2021.
39. [39] G. Dudek, A comprehensive study of random forest for short-term load forecasting, Energies, vol. 15, p. 7547, 2022.
40. [40] E. Elbasi, A. E. Topcu, and S. Mathew, Prediction of COVID-19 risk in public areas using IoT and machine learning, Electronics, vol. 10, p. 1677, 2021.