UniEload: Electrical load dataset for energy forecasting applications at public universities in Bangladesh

This paper presents a dataset on electrical power collected from a university campus in Bangladesh. It is meant to help research on energy forecasting in university settings. The dataset has hourly measurements of system voltage, three-phase currents (R, Y, B), and power factor (pf). These were recorded at the campus substation. Data were collected during different operational conditions, including academic periods and vacations. This provides insights into load behaviour, changes in power factor, and phase imbalance patterns in an educational setting. The dataset supports the creation and assessment of models for load forecasting, anomaly detection, and improving power efficiency. It was also combined with weather data to aid research on load forecasting that takes weather into account. The weather parameters include temperature, humidity, precipitation, wind speed, and solar radiation. All weather values match energy values and were gathered hourly and daily. This dataset is especially useful for researchers studying how artificial intelligence and machine learning can be applied in managing electrical energy. The dataset also includes notes about context, such as reduced load during national holidays. This improves its usefulness for studies that focus on events in forecasting. By making this dataset open access, it helps fill the gap in publicly available electrical load data from educational institutions in developing countries. This supports reproducible research and sustainable energy management on campus.