| Keywords |
Energy consumption ; Households energy use ; K-mean clustering ; Decision tree algorithm |
| Abstract |
In order to establish policy for effective energy saving of residential sector, it is necessary to make a reliable prediction about the energy usage of households. To do this, analyzing factors affecting the energy consumption is essential. In this study, we conducted K-mean cluster analysis and decision tree analysis using micro data of 「Household Energy Standing Survey」in 2015 to derive the factors affecting the energy consumption of households which has high uncertainty characteristics. And also, we analyzed the change of energy consumption according to detailed condition change of each variable. As a result of k-means clustering analysis, total energy consumption was classified into three groups: high, middle, and low. In addition, statistical analysis of the differences of variables for each group was conducted. Decision tree analysis showed structural relationships between variables that cannot be easily determined in a linear relationship. The main heating fuel, number of household members and household characteristics were found to have the greatest effect on annual energy consumption in decision tree analysis. |