| Title |
Assessing the Accuracy of Energy Consumption Prediction Models for Elementary Schools Through Classroom-based Group Segmentation |
| Authors |
정창헌(Cheong, Chang Heon) ; 황석호(Hwang, Seok-Ho) ; 김지영(Kim, Jiyeong) |
| DOI |
https://doi.org/10.5659/JAIK.2024.40.5.157 |
| Keywords |
Elementary school; Energy consumption estimation model; Number of classes; Number of students |
| Abstract |
This study explored whether classifying elementary schools by size could improve the accuracy of predicting energy consumption. Analyzing
energy use across 18,539 data points from various elementary schools, it became apparent that annual energy consumption patterns differed
between schools with fewer than 10 classes and those with more than 10 classes. Consequently, prediction accuracy improved when schools
were categorized into two groups based on size. Among the models proposed, multi-linear regression, which used the number of classes and
students as explanatory variables and annual energy consumption as the response variable, provided the most accurate predictions. A simpler
model that employed linear regression with only the number of classes also showed comparable accuracy. In contrast, a model using energy
consumption per student had the lowest predictive power. A limitation of this study was the limited access to various explanatory variables,
which could have contributed to the lower accuracy in some cases. Future research aims to address this by incorporating a broader range of
explanatory variables to refine the prediction model. |