Volume 59

Clustering Japanese Municipalities from A Perspective of Energy Demand Sectors Using Compositional Data Analysis with A Total Takahiro Yoshida, Daisuke Murakami, Ryuta Tsurumi, Hiroaki Onodera, Yujiro Hirano

Abstract

Many municipalities worldwide are setting net-zero emission targets. In Japan, over 1,000 of 1,791 municipalities, representing 99.7% of the population, have committed to achieving net zero by 2050. Despite these desirable commitments, most municipalities still lack a clear vision or concrete solutions for reaching their goals. To help local decision-makers develop effective decarbonization plans, it is crucial to identify municipalities with similar energy situations. This study aims to categorize municipalities using regional energy demand data. To effectively cluster municipalities based on energy demand, both the total energy volume and the composition of energy sectors were considered. We employed a Compositional Data Analysis (CoDA) approach to adequately incorporate both volume and sector composition information. While still rare in energy-related studies, the CoDA approach combined with both relative and absolute information provides a robust framework for classifying municipalities. From the results, we obtained intuitive and interpretable clusters for tacking the issue.

Keywords Japan energy database, sector compositions, compositional data analysis, compositional clustering with a total