Which is the best-fit response variable for modelling the energy consumption of households? An analysis based on survey data

Highlights: • Appropriateness of response variables for estimating energy use in households. • Analysis of the significance of household-related variables in energy use. • INLA modelling of real energy consumption data retrieved from household surveys. • Great influence of socio-economic variables on households' energy use. • Need to include the number of occupants in the response variable kWh/m²·occp·year. Multiple variables can affect the energy performance of a building and, generally, they can be divided into four groups: household typology, construction features, technical building systems and socio-economic profile. The aim of this work is to determine the best-fit response variable for modelling the energy consumption of households. The study seeks to answer two main research questions: RQ1 What is the most appropriate response variable to model the energy use for the household sector? And RQ2 What influence do a set of household-related variables have on households' energy use and what kinds affect it the most? To address both questions, firstly, different response variables were explored and the one with the best fit for modelling the energy use, according to statistical comparison parameters, was selected. Secondly, a global sensitivity analysis was performed to analyse the significance of each covariate included in the models. The results revealed that the socio-economic profile of the occupants is the most influential cluster and including occupation as a rate in the energy indicator contributes notably to improving the model. The study highlights the occupants' profile as a key factor and offers a valuable basis for practitioners to make decisions regarding energy policies.