Residential Electricity Consumption: Dataset Combining Multi-round Longitudinal Surveys and Energy Provider Data

A significant portion of the end users of electricity consists of residential consumers, often exceeding that of other consumer categories, particularly in developing countries. Effective demand-side management strategies increasingly rely on Artificial Intelligence (AI) and Machine Learning (ML), yet their success depends on access to high-quality, comprehensive datasets. However, finding datasets that collect insights into consumers' drivers, behaviors, and attitudes regarding domestic electricity usage in addition to information on electricity consumption while with adequate coverage, and clear documentation remains challenging.

This study presents a novel dataset collected from multiple Sri Lankan cities, combining 20 months of smart-meter and non-smart-meter electricity consumption records with three rounds of longitudinal surveys conducted with 4000 households. The dataset captures household living conditions, appliance usage, and behavioral changes over time, offering an intersection of granular energy consumption and consumer-driven factors.

With applications in non-intrusive load monitoring, inefficient appliance detection, and demand flexibility analysis, this dataset enables deeper insights into residential energy consumption dynamics while supporting data-driven approaches for optimizing energy management, informing policy, and advancing AI-driven solutions for sustainable electricity use.