This paper demonstrates the use of benchmarking methods for residential buildings in an energy rich economy. Three bottom-up methods were used, ordinary least square (OLS), support vector machines (SVM), and engineering modelling (EM). EM highlighted the energy inefficiency in these buildings, where the EM simulations showed the potential of halving the average measured consumption of 20.5 MWh/year. SVM could demarcate buildings and households with consumption of >16.0 MWh/year which are considered to be inefficiency. OLS is simple to use and the non-linear least square, showing R-2>0.8, performs better than linear least square fit. This work allows two EUI benchmark measures to be determined and they have average values of 2035 kWh/person/year and 56 kWh/m(2)/year. (C) 2014 The Authors. Published by Elsevier B.V.