Scholarlink

<p>Higher education institutions are pivotal in advancing low-carbon transitions through education, research, and innovation. Students’ daily behaviors and carbon awareness critically impact zero-carbon campus development, but systematic assessment of individual carbon footprints remains limited. To address this gap, this study introduces an innovative framework that integrates Life Cycle Assessment (LCA) with machine learning (ML). The approach specifically combines unsupervised clustering with classification algorithms to identify and validate distinct carbon emission patterns, thereby significantly enhancing the reliability and interpretability of findings over traditional approaches. Through analysis of 414 validated questionnaires from Guangdong University of Technology, results show that the average annual individual carbon emissions amount to 1438.07&#xa0;kg of CO₂, dominated by food consumption (36.7%) and energy usage (38.8%). ML clustering identifies distinct emission patterns. The high-emission groups were characterized by a 72% increase in meat consumption and 18% higher electricity usage compared to low-emission clusters. Furthermore, ML results quantitatively demonstrate that students with high carbon awareness are about 60% more likely to belong to low-emission groups, whereas those with poor awareness show approximately 65% prevalence in high-emission clusters, underscoring that enhancing carbon awareness is critical for emission reduction. These findings provide valuable insights into student behaviors and offer a data-driven foundation for targeted strategies to reduce campus carbon footprints.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine learning-assisted investigation on individual emissions of college students: a case study in Southern China

  • 期刊论文

  • 发表日期：

    2025年12月22日

查看全文
  • Zhiwen Cen,
  • Chuhao Lin,
  • Junyao Wang,
  • Yepeng Li,
  • Huijuan Qiu,
  • Weichi Li,
  • Xuelan Zeng,
  • Junfeng He,
  • Weixun Zeng

摘要

Higher education institutions are pivotal in advancing low-carbon transitions through education, research, and innovation. Students’ daily behaviors and carbon awareness critically impact zero-carbon campus development, but systematic assessment of individual carbon footprints remains limited. To address this gap, this study introduces an innovative framework that integrates Life Cycle Assessment (LCA) with machine learning (ML). The approach specifically combines unsupervised clustering with classification algorithms to identify and validate distinct carbon emission patterns, thereby significantly enhancing the reliability and interpretability of findings over traditional approaches. Through analysis of 414 validated questionnaires from Guangdong University of Technology, results show that the average annual individual carbon emissions amount to 1438.07 kg of CO₂, dominated by food consumption (36.7%) and energy usage (38.8%). ML clustering identifies distinct emission patterns. The high-emission groups were characterized by a 72% increase in meat consumption and 18% higher electricity usage compared to low-emission clusters. Furthermore, ML results quantitatively demonstrate that students with high carbon awareness are about 60% more likely to belong to low-emission groups, whereas those with poor awareness show approximately 65% prevalence in high-emission clusters, underscoring that enhancing carbon awareness is critical for emission reduction. These findings provide valuable insights into student behaviors and offer a data-driven foundation for targeted strategies to reduce campus carbon footprints.