Knowledge discovery of indoor environment patterns in mild climate countries based on data mining applied to in-situ measurements

• Published in: Sustainable cities and society Vol. 30; pp. 37 - 48
• Main Authors: Ramos, Nuno M.M., Almeida, Ricardo M.S.F., Simões, M. Lurdes, Pereira, Pedro F.
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.04.2017
• Subjects: Data mining; Indoor environment; Social housing; User behaviour; Social housing; Indoor environment; User behaviour; Data mining
• ISSN: 2210-6707; 2210-6715
• DOI: 10.1016/j.scs.2017.01.007

•Proposal of a data mining methodology for indoor environment measurements.•Hygrothermal characterization of a mild climate non-rehabilitated social housing sample.•Tailored application of principal components and cluster analysis to hygrothermal data.•User behaviour as an important factor for clusterization. Temperature and relative humidity values, from a sample of 24 flats with homogeneous architectural features and social strata, were continuously measured during the heating season and a typical summer period. The results proved the existence of discomfort during the heating season, revealing energy poverty patterns, but at the same time sensible differences that could only be explained by user actions. This led to a deeper analysis of the data in search of relevant patterns and causes for the inhomogeneity. A methodology for exploring the resulting large data set is proposed in the present paper, based on the application of data mining techniques. The calculation of meaningful percentiles of hygrothermal variables, followed by the application of a principal components analysis, allowed for the cluster analysis to the flats. As a result, clusters with a specific hygrothermal pattern were found, which meant that the factors leading to such performance could be explored, thus revealing the importance of users and their behaviour.