Scholarlink

<p>Predicting residents' participation in flood response is key to improving community resilience and emergency management. This study combines machine learning (ML) and agent-based modeling (ABM) to predict flood response behaviors in Zhengzhou, China, considering information asymmetry in emergency resource accessibility. First, the XGBoost models identified fire station and hospital proximity as key factors influencing residents' participation in flood response. However, while these resources were physically close, residents’ perceptions of accessibility were much lower, creating a gap between actual and perceived accessibility. This information asymmetry then formed the basis for the predictive ABM, which simulated how improving perception through outreach efforts would affect participation. Second, simulation results indicated that enhancing residents’ perception of emergency resources significantly increased participation, with high participation rising from 7.18% to 14.06%, medium participation increasing from 25.98% to 46.34%, and low participation decreasing from 66.84% to 39.6%. Importantly, the improvement in participation was uniform across the study area, highlighting the consistent effectiveness of this intervention across diverse urban contexts. This study’s integration of ML and ABM presents a significant methodological advancement, offering a robust framework for predictive modeling of disaster response participation. This novel approach is crucial in demonstrating that reducing information asymmetry concerning emergency resource accessibility effectively enhances community engagement in disaster management.</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

Resident Participation in Flood Response: A Machine Learning and Agent-Based Simulation Study of Zhengzhou, China

  • 期刊论文

  • 发表日期：

    2025年6月16日

查看全文
  • Yuxiao Wang,
  • Xinyue Han,
  • Wei Ma,
  • Zanmei Wei,
  • Zhouying Song,
  • Mengmeng Zhang,
  • Huaxiong Jiang

摘要

Predicting residents' participation in flood response is key to improving community resilience and emergency management. This study combines machine learning (ML) and agent-based modeling (ABM) to predict flood response behaviors in Zhengzhou, China, considering information asymmetry in emergency resource accessibility. First, the XGBoost models identified fire station and hospital proximity as key factors influencing residents' participation in flood response. However, while these resources were physically close, residents’ perceptions of accessibility were much lower, creating a gap between actual and perceived accessibility. This information asymmetry then formed the basis for the predictive ABM, which simulated how improving perception through outreach efforts would affect participation. Second, simulation results indicated that enhancing residents’ perception of emergency resources significantly increased participation, with high participation rising from 7.18% to 14.06%, medium participation increasing from 25.98% to 46.34%, and low participation decreasing from 66.84% to 39.6%. Importantly, the improvement in participation was uniform across the study area, highlighting the consistent effectiveness of this intervention across diverse urban contexts. This study’s integration of ML and ABM presents a significant methodological advancement, offering a robust framework for predictive modeling of disaster response participation. This novel approach is crucial in demonstrating that reducing information asymmetry concerning emergency resource accessibility effectively enhances community engagement in disaster management.