A Systematic Review of Machine Learning Approaches for Electricity Theft Detection under Class Imbalance: Implications for Utilities in Low-Resource Settings
Abstract
Electricity theft accounts for much of the non-technical loss that utilities in low- and middle-income countries absorb, and machine learning on consumption data is the most studied way to detect it. Theft cases are rare in these data, so models must handle class imbalance, and how they do so affects how far reported results can be trusted. Following PRISMA 2020, we searched the Directory of Open Access Journals for peer-reviewed studies published from 2016 to 2026 that trained a classifier for electricity theft or non-technical loss and handled class imbalance explicitly. Of 64 records, 57 studies (58 reports) met the criteria. SMOTE and related oversampling methods were the most common remedy (22 studies), followed by hybrid resampling (10) and generative models (9). Thirty-seven studies (65%) used the public State Grid Corporation of China dataset, and one used data from an African utility. Of the 55 studies that resampled or generated data, 11 kept resampling to the training data, by explicit statement or by design, and four stated that balancing came before the train and test split; the other 40 did not make the order clear. Accuracy was reported by 77% of studies, MCC by 32% and PR-AUC by 21%. No study used a time-ordered split, three tuned the decision threshold, and two checked predictions against real cases. Two studies reported values that cannot all hold for the same test set. High reported accuracies are therefore weak evidence of field performance. Utilities such as the Kenya Power and Lighting Company should judge models on inspection outcomes, time-ordered validation and precision-recall measures before deployment.
References
Abro, S. A., Laghari, J. A., Memon, S. A., Khan, T. A., Memon, I., Nasir, H., & Fatima, K. (2025). Non-technical loss detection in power distribution networks using machine learning. Scientific Reports, 15(1), 36189. https://doi.org/10.1038/s41598-025-20048-z
Adil, M., Javaid, N., Qasim, U., Ullah, I., Shafiq, M., & Choi, J.-G. (2020). LSTM and bat-based RUSBoost approach for electricity theft detection. Applied Sciences, 10(12), 4378. https://doi.org/10.3390/app10124378
Alfaverh, F., Gan, H., Miroshnyk, V., Bin Saeed, Z., Blinov, I., Shymaniuk, P., Tarassodi, P., & Mporas, I. (2026). Electricity theft detection from electricity and gas measurements using machine learning. Energies, 19(9), 2045. https://doi.org/10.3390/en19092045
Ali, A., Khan, L., Javaid, N., Aslam, M., Aldegheishem, A., & Alrajeh, N. (2024). Exploiting machine learning to tackle peculiar consumption of electricity in power grids: A step towards building green smart cities. IET Generation, Transmission & Distribution, 18(3), 413–445. https://doi.org/10.1049/gtd2.13056
Ali, A., Khan, L., Javaid, N., Bouk, S. H., Aldegheishem, A., & Alrajeh, N. (2025). Mitigating anomalous electricity consumption in smart cities using an AI-based stacked-generalization technique. IET Renewable Power Generation, 19(1), e12785. https://doi.org/10.1049/rpg2.12785
Ammar, M., Javaid, N., & Arishi, A. (2025). An AI-explained systematic modular approach for enhanced electricity theft detection for urbanized smart grid environment. Alexandria Engineering Journal, 129, 998–1023. https://doi.org/10.1016/j.aej.2025.07.040
Asif, M., Nazeer, O., Javaid, N., Alkhammash, E. H., & Hadjouni, M. (2022). Data augmentation using BiWGAN, feature extraction and classification by hybrid 2DCNN and BiLSTM to detect non-technical losses in smart grids. IEEE Access, 10, 27467–27483. https://doi.org/10.1109/ACCESS.2022.3150047
Chawla, N. V., Bowyer, K. W., Hall, L. O., & Kegelmeyer, W. P. (2002). SMOTE: Synthetic minority over-sampling technique. Journal of Artificial Intelligence Research, 16, 321–357. https://doi.org/10.1613/jair.953
Chicco, D., & Jurman, G. (2020). The advantages of the Matthews correlation coefficient (MCC) over F1 score and accuracy in binary classification evaluation. BMC Genomics, 21(1), 6. https://doi.org/10.1186/s12864-019-6413-7
Coelho, F. J. S., Feitosa, A. R. S., Alcântara, A. L. M., Li, K., Lima, R. F., Silva, V. R., & da Silva-Filho, A. G. (2023). HyMOTree: Automatic hyperparameter tuning for non-technical loss detection based on multi-objective and tree-based algorithms. Energies, 16(13), 4971. https://doi.org/10.3390/en16134971
Ghori, K. M., Awais, M., Khattak, A. S., Imran, M., Fazal-E-Amin, & Szathmary, L. (2021). Treating class imbalance in non-technical loss detection: An exploratory analysis of a real dataset. IEEE Access, 9, 98928–98938. https://doi.org/10.1109/ACCESS.2021.3095145
Glauner, P., Meira, J. A., Valtchev, P., State, R., & Bettinger, F. (2017). The challenge of non-technical loss detection using artificial intelligence: A survey. International Journal of Computational Intelligence Systems, 10(1), 760–775. https://doi.org/10.2991/ijcis.2017.10.1.51
Gong, X., Tang, B., Zhu, R., Liao, W., & Song, L. (2020). Data augmentation for electricity theft detection using conditional variational auto-encoder. Energies, 13(17), 4291. https://doi.org/10.3390/en13174291
He, H., & Garcia, E. A. (2009). Learning from imbalanced data. IEEE Transactions on Knowledge and Data Engineering, 21(9), 1263–1284. https://doi.org/10.1109/TKDE.2008.239
Janthong, S., Chalermyanont, K., & Duangsoithong, R. (2023). Unbalanced data handling techniques for classifying energy theft and defective meters in the Provincial Electricity Authority of Thailand. IEEE Access, 11, 46522–46540. https://doi.org/10.1109/ACCESS.2023.3274543
Kapoor, S., & Narayanan, A. (2023). Leakage and the reproducibility crisis in machine-learning-based science. Patterns, 4(9), 100804. https://doi.org/10.1016/j.patter.2023.100804
Kaufman, S., Rosset, S., Perlich, C., & Stitelman, O. (2012). Leakage in data mining: Formulation, detection, and avoidance. ACM Transactions on Knowledge Discovery from Data, 6(4), 1–21. https://doi.org/10.1145/2382577.2382579
Kenya Power and Lighting Company. (2023). Annual report and financial statements 2022/2023.
Khan, N., Amir Raza, M., Ara, D., Mirsaeidi, S., Ali, A., Abbas, G., Shahid, M., Touti, E., Yousef, A., & Bouzguenda, M. (2023). A deep learning technique, AlexNet, to detect electricity theft in smart grids. Frontiers in Energy Research, 11, 1287413. https://doi.org/10.3389/fenrg.2023.1287413
Khan, N., Shahid, Z., Alam, M. M., Sajak, A. A. B., Nazar, M., & Mazliham, M. S. (2024). A novel deep learning technique to detect electricity theft in smart grids using AlexNet. IET Renewable Power Generation, 18(6), 941–958. https://doi.org/10.1049/rpg2.12846
Kwarteng, M. Y., Effah, F. B., Kwegyir, D., & Frimpong, E. A. (2023). ANN-based electricity theft classification technique for limited data distribution systems. Jurnal Nasional Teknik Elektro, 12(1), 8–16. https://jnte.ft.unand.ac.id/index.php/jnte/article/view/1072
Lin, G., Feng, H., Feng, X., Wen, H., Li, Y., Hong, S., & Ni, Z. (2021). Electricity theft detection in power consumption data based on adaptive tuning recurrent neural network. Frontiers in Energy Research, 9, 773805. https://doi.org/10.3389/fenrg.2021.773805
Liu, H., Li, Z., & Li, Y. (2020). Noise reduction power stealing detection model based on self-balanced data set. Energies, 13(7), 1763. https://doi.org/10.3390/en13071763
Messinis, G. M., & Hatziargyriou, N. D. (2018). Review of non-technical loss detection methods. Electric Power Systems Research, 158, 250–266. https://doi.org/10.1016/j.epsr.2018.01.005
Monteiro, T., Castor, G., Castillo Correa, C., Arias, H., Ñaupari Huatuco, D., & Molina Rodríguez, Y. (2025). A hybrid machine learning framework for electricity fraud detection: Integrating isolation forest and XGBoost for real-world utility data. Energies, 18(23), 6249. https://doi.org/10.3390/en18236249
Mujeeb, S., Javaid, N., Ahmed, A., Gulfam, S. M., Qasim, U., Shafiq, M., & Choi, J.-G. (2021). Electricity theft detection with automatic labeling and enhanced RUSBoost classification using differential evolution and Jaya algorithm. IEEE Access, 9, 128521–128539. https://doi.org/10.1109/ACCESS.2021.3102643
Narayanan, A., & Hardy, T. (2023). Synthetic data generation for machine learning model training for energy theft scenarios using co-simulation. IET Generation, Transmission & Distribution, 17(5), 1035–1046. https://doi.org/10.1049/gtd2.12619
Norouzi, A., Ahmadi, I., Nazari, M., Pourrostami, H., & Hoseini-Kordkheili, H. (2025). A novel framework to detect electricity theft in regions with traditional metering: applying a hybrid machine learning-based approach to low-sampling-rate data. Results in Engineering, 25, 104478. https://doi.org/10.1016/j.rineng.2025.104478
Orozco-Cerón, O. W., Joaqui-Barandica, O., & Manotas-Duque, D. F. (2026). Comparing machine and deep learning for electricity theft detection from monthly billing data in an emerging energy market. Technologies, 14(9), 568. https://doi.org/10.3390/technologies14090568
Page, M. J., McKenzie, J. E., Bossuyt, P. M., Boutron, I., Hoffmann, T. C., Mulrow, C. D., Shamseer, L., Tetzlaff, J. M., Akl, E. A., Brennan, S. E., Chou, R., Glanville, J., Grimshaw, J. M., Hróbjartsson, A., Lalu, M. M., Li, T., Loder, E. W., Mayo-Wilson, E., McDonald, S., . . . Moher, D. (2021). The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. BMJ, 372, n71. https://doi.org/10.1136/bmj.n71
Pamir, Javaid, N., Javed, M. U., Houran, M. A., Almasoud, A. M., & Imran, M. (2023). Electricity theft detection for energy optimization using deep learning models. Energy Science & Engineering, 11(10), 3575–3596. https://doi.org/10.1002/ese3.1541
Qu, Z., Li, H., Wang, Y., Zhang, J., Abu-Siada, A., & Yao, Y. (2020). Detection of electricity theft behavior based on improved synthetic minority oversampling technique and random forest classifier. Energies, 13(8), 2039. https://doi.org/10.3390/en13082039
Rahaman, M. A., & Mohamad Idris, R. (2026). A stacking ensemble with Pareto optimization for scalable electricity theft detection via hybrid data repair and lightweight deployment. Scientific Reports, 16(1), 14548. https://doi.org/10.1038/s41598-026-39693-z
Rajitroj, N., Seepromting, K., & Chatthaworn, R. (2025). 1D-CNN and a weight-based balancing technique for improved detection of non-technical losses in power distribution systems. E3S Web of Conferences, 629, 06005. https://doi.org/10.1051/e3sconf/202562906005
Saito, T., & Rehmsmeier, M. (2015). The precision-recall plot is more informative than the ROC plot when evaluating binary classifiers on imbalanced datasets. PLOS ONE, 10(3), e0118432. https://doi.org/10.1371/journal.pone.0118432
Sánchez-Mojica, K. Y., & Suárez, O. J. (2026). Artificial intelligence model and balancing techniques for controlling non-technical losses in the electrical energy sector. AiBi Revista de Investigación, Administración e Ingeniería, 14(1), 1–7. https://doi.org/10.15649/2346030X.5635
Santos, M. S., Soares, J. P., Abreu, P. H., Araújo, H., & Santos, J. (2018). Cross-validation for imbalanced datasets: Avoiding overoptimistic and overfitting approaches. IEEE Computational Intelligence Magazine, 13(4), 59–76. https://doi.org/10.1109/MCI.2018.2866730
Savian, F. de S., Siluk, J. C. M., Garlet, T. B., do Nascimento, F. M., Pinheiro, J. R., & Vale, Z. (2021). Non-technical losses: A systematic contemporary article review. Renewable and Sustainable Energy Reviews, 147, 111205. https://doi.org/10.1016/j.rser.2021.111205
Sun, Y., Lee, J., Kim, S., Seon, J., Lee, S., Kyeong, C., & Kim, J. (2023). Energy theft detection model based on VAE-GAN for imbalanced dataset. Energies, 16(3), 1109. https://doi.org/10.3390/en16031109
Sun, Y., Sun, X., Hu, T., & Zhu, L. (2023). Smart grid theft detection based on a hybrid multi-time scale neural network. Applied Sciences, 13(9), 5710. https://doi.org/10.3390/app13095710
Wolff, R. F., Moons, K. G. M., Riley, R. D., Whiting, P. F., Westwood, M., Collins, G. S., Reitsma, J. B., Kleijnen, J., & Mallett, S. (2019). PROBAST: A tool to assess the risk of bias and applicability of prediction model studies. Annals of Internal Medicine, 170(1), 51–58. https://doi.org/10.7326/M18-1376
Zheng, Z., Yang, Y., Niu, X., Dai, H.-N., & Zhou, Y. (2018). Wide and deep convolutional neural networks for electricity-theft detection to secure smart grids. IEEE Transactions on Industrial Informatics, 14(4), 1606–1615. https://doi.org/10.1109/TII.2017.2785963
