Various studies utilize the data envelopment analysis (DEA) technique to measure the efficiency of decision making units (DMUs) in energy system. An inappropriate input-or output-oriented DEA model as well as heterogeneous DMUs can cause unreasonable results in efficiency analysis. K-Means clustering algorithm was applied to classify homogenous countries in the demand and supply sides of an energy system. An input-oriented DEA model was used to analyze power stations (PS) under renewable energy. While, an output-oriented model was used not only to examine PS' efficiency under non-renewables and refineries, but also in the demand analysis. Further, the output variable for the demand analysis consisted of an energy-related quality of life (QoL) indicator. The overall energy efficiency was calculated through multiplying the supply efficiency by the demand efficiency. The results of the paper showed that the highest potential energy saving (PES) sources on the supply side were the non-renewables in power stations, followed by refineries, then renewables. The demand-side analysis identified that the highest PES belonged to countries with high populations and high-income economies. In conclusion, the overall energy efficiency analysis based on QoL suggested allowances be made to use fossil fuels in poorer countries with weak economies and smaller populations. This allowance was proposed to support energy poverty reduction, improve health, and promote education.