A statistical quantitative analysis of the correlations between socio-demographic characteristics and household occupancy patterns in residential buildings in China

Abstract

Occupancy patterns that describe occupant presence or absence can result in a significant impact in energy use of residential buildings; thus, they are critical to the energy modeling of residential buildings. Occupancy patterns currently embedded in building energy simulation tools or recommended in the standards are uniform, ignoring that different occupancy patterns exist for households with different socio-demographic characteristics. Consequently, this study objective is to extract the typical occupancy patterns of households and quantify the correlations between household occupancy patterns and socio-demographic characteristics based on statistical approaches. Clustering analysis was used to identify representative types of occupancy patterns based on the occupancy profiles of daily occupant presence data. Then, correlations between occupancy patterns and household characteristics were identified by investigating three statistical methods: nonparametric analysis, decision trees, and association rule mining. To illustrate the quantitative analysis, datasets collected by a large-scale questionnaire survey in two Chinese cities were used for the case study. The results revealed that socio-demographic characteristics, such as occupant age, employment status and household composition, have statistically significantly associated with occupancy patterns. The statistical quantitative analysis can be used to estimate household occupancy patterns to further reduce the performance gap between the simulation and actual energy consumption of residential buildings.