2020 年 41 巻 5 号 p. 209-218
To estimate the energy use reduction due to energy saving measures, it is necessary to understand the current state of energy consumption considering difference of energy use among households. In this paper, a multinomial logistic regression model which predicts the appliance ownership status from household attributes was developed using government statistics (Household CO2 Statistics, 2014). By integrating this model with the residential end-use energy simulation model, the residential sector energy consumption of Japan was calculated. As a result, it is clarified the household using a lot of energy owns more appliances than the others. Comparing the energy simulation result and actual energy consumption from the Household CO2 Statistics, it was verified that the variation of the energy consumption close to the actual situation can be reproduced. It was also shown that the largest cause of error between simulation results and statistical values was due to the differences in resident lifestyle and occupant’s behavior. Furthermore, it was demonstrated that households that were underestimated in the simulation model had more energy consuming behavior and lower energy saving awareness than standard households.