Detection of Anomalous Energy Consumption Patterns For Electricity Theft Detection In Honduras Using Machine Learning Techniques

Electricity theft poses a significant challenge to the reliability and efficiency of power distribution in Honduras. In this paper, we present a machine learning-based framework to detect anomalous energy consumption patterns that may indicate potential cases of electricity theft. Our approach employs a two-stage methodology: first, exploratory data analysis and dimensionality reduction are performed using Principal Component Analysis (PCA) to reveal the underlying structure of household energy usage; second, unsupervised anomaly detection techniques—namely, Isolation Forest and Local Outlier Factor (LOF)—are applied to identify irregular consumption behaviors. The framework is validated on a dataset collected from 48 households in the department of Cortés, capturing variables related to household demographics, appliance usage, energy consumption, and billing. The experimental results demonstrate that our approach can successfully flag potential anomalies, providing a scalable and replicable solution to mitigate non-technical losses in the energy sector.