A Novel Methodology for Identifying Associations and Correlations Between Household Appliance Behaviour in Residential Buildings

This paper reports the development of a new methodology for examining all associations and correlations between various household appliance behaviour, thereby discovering hidden patterns of occupant behaviour in residential buildings. The method is based on a basic data mining technique (association rule mining). Its strength lies in its ability to analyse both continuous and nominal data, and examine all associations and correlations automatically. To demonstrate its applicability, it was applied to the measured end-use electricity data in a selected residential building with comprehensive household appliances in Japan. The results show that both direct and indirect associations between occupant behaviour were discovered. The results obtained could provide a deep insight into the interaction between different behaviour, offer detailed recommendations for reducing building energy consumption, and enable the best input parameters of occupant behaviour prediction models to be identified.