Identifying factors influencing energy saving cognition-behavior gaps in shared residential spaces using interpretable machine learning : A dormitory case study
Research output: Journal Publications and Reviews › RGC 21 - Publication in refereed journal › peer-review
Author(s)
Related Research Unit(s)
Detail(s)
Original language | English |
---|---|
Article number | 111809 |
Journal / Publication | Journal of Building Engineering |
Volume | 100 |
Online published | 7 Jan 2025 |
Publication status | Published - 15 Apr 2025 |
Link(s)
Abstract
Residential buildings are important contributors to building energy consumption. Occupants' energy-saving behaviors and cognition have significant impacts on building energy saving, which are often influenced by factors such as group and environment, especially in shared residential spaces. Several studies have examined factors influencing energy-saving behaviors, highlighting a significant gap between these behaviors and individuals' cognition regarding energy conservation. However, why such a gap exists have not been well explored. This research aims to identify and assess factors influencing occupants' energy-saving cognition-behavior gaps in shared residential spaces, using university dormitories as a case study. By comparing five interpretable machine learning algorithms, Random Forest was found to be the optimal algorithm. The contribution of different factors to the model was explained through SHapley Additive exPlanations (SHAP). Results show that compared with individual dimension factors, interaction dimension factors such as willingness to responding to cultural measures had a more important influence on energy-saving cognition-behavior gaps. Factors exhibited the importance variability across four groups of occupants. Management strategies designed to bridge energy-saving cognition-behavior gaps for each occupant group were proposed accordingly. This research innovatively proposed a new perspective of separating energy-saving cognition and behavior to explore influencing factors of occupants' energy conservation. Furthermore, this research provided valuable insights for building managers to develop tailored energy-saving strategies aligned with occupants’ cognitive and behavioural differences. By addressing these variations, managers can implement targeted interventions to enhance energy efficiency, optimize facility operations, and foster greater occupant engagement - key to advancing sustainability in the built environment. © 2025 Elsevier Ltd
Research Area(s)
- Building energy saving, Cognition-behavior gaps, Influencing factors, Interpretable machine learning, Shared residential spaces
Citation Format(s)
Identifying factors influencing energy saving cognition-behavior gaps in shared residential spaces using interpretable machine learning: A dormitory case study. / Wang, Hanqi; Li, Jie; Mao, Peng.
In: Journal of Building Engineering, Vol. 100, 111809, 15.04.2025.
In: Journal of Building Engineering, Vol. 100, 111809, 15.04.2025.
Research output: Journal Publications and Reviews › RGC 21 - Publication in refereed journal › peer-review