Systematic data mining-based framework to discover potential energy waste patterns in residential buildings

Energy feedback systems are recently proposed to help occupants understand and improve their energy use behavior. Despite many potential benefits, the question remains, whether useful and straightforward knowledge are transferred to the occupants about their energy use patterns. In this context, the key is to develop methodologies that can effectively analyze occupants' energy use behavior and distinguish their energy-inefficient behavior (if any). Previous studies seldom considered the dynamics of occupancy, which may result in misleading information to the occupants and inefficacy in recognizing the actual wasteful behavior. To fill this gap, this study proposes a data mining framework with a combination of change point analysis (CPA), cluster analysis, and association rule mining (ARM) to explore the relationship between occupancy and building energy consumption, aiming at identifying potential energy waste patterns and to provide useful feedback to the occupants. To demonstrate the capability of the developed framework, it was applied to datasets collected from two different apartments located in Lyon, France. Results indicate that different energy waste patterns can be effectively discovered in both apartments through the proposed framework and a substantial amount of energy savings can be achieved by modifying occupants' energy use behavior. The proposed framework is flexible and can be adaptive to households with different occupancy patterns and habitual energy-use behavior. Nevertheless, the discovered energy saving potentials and benchmark values are limited to the apartments considered in this study and similar analysis based on the proposed framework are needed in wider building stocks to explore its generalizability. (C) 2019 Elsevier B.V. All rights reserved.