This dataset contains the metadata (or point name) and timeseries data (one week's worth) of streams from three office buildings: one from the University of Virginia and two from UC Berkeley.

We manually created the type ground truth (e.g., temperature, humidity, etc) for all the streams.

If you have further questions, feel free to contact me at hong AT virginia DOT edu.

The dataset is available here.


If you use the dataset, please consider citing the following papers:

Dezhi Hong, Hongning Wang, Jorge Ortiz, Kamin Whitehouse.
The Building Adaptor: Towards Quickly Applying Building Analytics at Scale.
In BuildSys'15 (bibtex)

Dezhi Hong, Hongning Wang, Kamin Whitehouse.
Clustering-based Active Learning on Sensor Type Classification in Buildings.
In CIKM'15 (bibtex)

Arka Bhattacharya, Dezhi Hong, David Culler, Jorge Ortiz, Kamin Whitehouse, Eugene Wu.
Automated Metadata Construction to Support Portable Building Applications.
In BuildSys'15 (bibtex)