A multisensor dataset of energy use and activity in Dutch social housing

This record contains the HOMESENSE dataset, a harmonised multisensor release from 11 pseudonymised social-rental dwellings in Limburg, the Netherlands. It covers 366 consecutive local calendar days from 26 November 2024 through 26 November 2025. Available measurements include indoor environmental quality (temperature, relative humidity, CO2, dimensionless TVOC, barometric pressure and light), PIR- and door-sensor activity signals, thermostatic radiator-valve telemetry, appliance-level electricity, whole-dwelling electricity grid import and export, gas measurements where available, and Maastricht weather data. Sensor availability differs by dwelling. The release uses a common 10-minute publication grid. Each measurement is paired with a quality-control field identifying observed, interpolated, reset-imputed or missing values. The package contains 14 files: H01.csv through H11.csv, preprocess_release.py, getting_started.ipynb and demo_analysis.ipynb. Each household CSV contains 52,704 timestamped rows; column counts vary with the installed sensors. The notebooks demonstrate reproducible loading and analysis, while the Python script documents release processing from a non-public intermediate sensor export. The dataset supports research on activity patterns, heating control, indoor thermal conditions, residential energy use, low-frequency load attribution and coarse-resolution non-intrusive load monitoring. PIR signals indicate detected activity rather than ground-truth occupancy. Household identifiers H01-H11 are pseudonyms. Ethical approval for the monitoring campaign was granted by the Ethical Committee CETO of Open Universiteit (Reference Ceto_RP782), and all participants provided informed consent. Raw unprocessed household records, address information and direct personal identifiers are not distributed.