Three years of hourly data from 3021 smart heat meters installed in Danish residential buildings

Abstract

The now widespread use of smart heat meters for buildings connected to district heating networks generates data at an unknown extent and temporal resolution. This data encompasses information that enables new data-driven approaches in the building sector. Real-life data of sufficient size and quality are necessary to facilitate the development of such methods, as subsequent analyses typically require a complete equidistant dataset without missing or erroneous values. Thus, this work presents three years (2018-01-03 till 2020-12-31) of screened, interpolated, and imputed data from 3,021 commercial smart heat meters installed in Danish residential buildings. The screening aimed to detect data from not used meters, resolve issues caused by the data storage process and identify erroneous values. Linear interpolation was used to obtain equidistant data. After the screening, 0.3% of the data were missing, which were imputed using a weighted moving average based on a systematic comparison of nine different imputation methods. The original and processed data are published together with the code for data processing (https://doi.org/10.5281/zenodo.6563114).