This public tabular dataset along with associated Jupyter notebooks provides aggregated measurements of sensors (light/noise/motion, smart camera-based counter, bluetooth device presence) in university rooms, and the ability to inquire about the likelihood of occupancy at a desired reservation time. It is meant to contribute to smarter room reservation systems based on room atmosphere models, and therefore to the digitalisation in facility management.

All input data are contained in the folder 'parquets' in Parquet format. They have been aggregated and discretised from corresponding CSV raw data through the notebook 'RoomatchModelCreation.ipynb'. Each room/sensor pair represents a matrix of 7 week days x 96 15-minute slots. A description of the raw data format, for potential reimplementation, is given in the file 'RoomatchRawData'.

The second notebook 'RoomatchUnifiedModel.ipynb' merges the model information per room and creates room atmosphere models to estimate room occupancy and utilisation. The folder 'models' contains the resulting matrix overlay models.

The exemplary room reservation system 'RoomatchOccupancyClient.py' exploits that model information to make a guess on when a room will likely be available.

Five exemplary rooms are part of this dataset, measured over several weeks in 2023 in an actual university setting:

room1: Two person office over two weeks; all sensors
room2: Larger shared research office over a couple of days; no camera
room3: Meeting room over one week; no camera
room4: Lecture room for a few hours; no camera
room5: Larger shared research office, active corner, for some days; no camera
