A sub-sequence clustering method for identifying daily indoor environmental patterns from massive time-series data

Indoor Environmental Quality (IEQ) affects human comfort, productivity and health. The rich values behind massive IEQ data are urgently required for the improvement of building performance. Clustering is one common approach for data mining. However, current time-series clustering methods are not applicable to IEQ data, owing to its high dimension and complex pattern. This study proposes a sub-sequence clustering framework for the extraction of daily IEQ patterns. Two case studies were conducted: 1) three main daily patterns of air temperature were extracted from 734 curves in an office building, and 2) six typical daily patterns of CO₂ concentration were identified from 1884 curves in a university building. Post-clustering analyses, including a classification tree, were also performed for knowledge discovery. Finally, the clustering performance of the proposed method was compared with that of previous methods. The results indicate that the sub-sequence clustering method is appropriate for identifying daily IEQ patterns.