Anomaly Detection in Conventional Meters and Electricity Consumption Using Optimized Extreme Gradient Boosting

With the progress of innovative information technologies, smart energy meters have designed a substantial quantity of real-time electricity consumption. Energy consumption is based on the various factors such as humidity, building area, temperature, habitation and so on and these factors are cooperatively describing the energy meter background. The smart meter supports to capable to identify data abnormalities for efficient monitoring as well as investigation. The anomaly detection supports for better decision making for minimizing the energy consumption. In this research, Machine Learning (ML) approach of Optimized Extreme Gradient Boosting (OXGBoosting) for the detection and classification of anomaly for energy consumption in conventional meters. This research collects the large-scale anomaly dataset for estimating the effectiveness. The proposed method is estimated by various performance metrices and it achieves an accuracy of 0.95, precision of 0.92, recall of 0.92, F1-score of 0.91 and AUC of 0.95 respectively when compared to the previous studies such as Extreme Gradient Boosting (XGB), Two-Class Boosted Decision Tree and NGBoost.