Enhancing the Electricity Theft Detection using Extreme Gradient Boosting with Optuna Optimization in Smart Grid

Electricity theft poses significant challenges to power utilities, leading to considerable financial losses and operational inefficiencies. This study introduces an innovative approach combining Extreme Gradient Boosting (XG-Boost) with the Optuna optimization method (OOM) to improve electricity theft detection. XG-Boost is known for its speed and accuracy in handling large datasets, offering robust feature selection and powerful boosting capabilities. Optuna is an efficient hyperparameter tuning technique to optimize model performance. Experiments conducted on real-world datasets demonstrate the effectiveness of our approach, with the proposed model surpassing conventional models in terms of Accuracy, F1, AUC, and Kappa Score. Furthermore, computational efficiency has also been considered a key parameter that further improves the practicality of the model. This study highlights the potential of integrating gradient boosting machines with optimization techniques to enhance electricity theft detection, thereby contributing to the reduction of non-technical losses and promoting the efficiency of smart power distribution systems.