Device Activity Report with Complete Knowledge Dataset

This technical report presents the Device Activity Report with Complete Knowledge (DARCK) dataset, which is designed for Non-Intrusive Load Monitoring (NILM) research. The dataset was collected over six months in a two-person household in Germany and includes power readings from 51 different appliances, as well as aggregate readings from the mains. The report outlines the measurement setup, data collection methods, and preprocessing steps undertaken to ensure data quality and usability. It highlights the dataset's completeness, as it monitors every appliance, including lighting, which is often omitted in other datasets. The report also discusses the challenges faced during data collection, such as device disconnections and the need for calibration, and details the postprocessing techniques applied to enhance the dataset's utility for researchers. The DARCK dataset aims to fill a significant gap in publicly available NILM datasets by providing high-quality, fully disaggregated data suitable for evaluating machine learning algorithms and energy consumption models.