Analysis and promotion of resident evacuation behavior with explainable machine learning (XAI) model

2023 Fiscal Year Final Research Report

• Project/Area Number: 21K04301
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 22050:Civil engineering plan and transportation engineering-related
• Research Institution: Gifu University
• Principal Investigator: takagi akiyoshi 岐阜大学, 社会システム経営学環, 教授 (30322134)
• Co-Investigator(Kenkyū-buntansha): 杉浦 聡志 北海道大学, 工学研究院, 准教授 (30648051)
• Project Period (FY): 2021-04-01 – 2024-03-31
• Keywords: 避難行動 / 豪雨災害 / 機械学習 / 説明可能なAI / XAI / アンケート / 要因分析

Outline of Final Research Achievements

One of the reasons for the continued human damage caused by heavy rain disasters is that residents have not taken appropriate evacuation actions, and various measures have been implemented. On the other hand, although a vast amount of research has been accumulated on resident evacuation behavior, appropriate resident evacuation actions have not been taken and human damage has not been reduced. In this study, we used eXplainable AI (XAI) to resident evacuation behavior questionnaire survey data consisting of standardized items collected over the past five years to clarify factors related to resident evacuation behavior, taking into account the relationship with disaster situations and local conditions.

Free Research Field

土木計画学

Academic Significance and Societal Importance of the Research Achievements

豪雨災害による人的被害が後を絶たない一原因である適切な住民避難行動が適切に取られていない課題に対して，５年間に亘って統一的な項目で構成される住民避難行動アンケート調査を実施したことは，今後の調査方法を考えるうえでも，蓄積したデータそのものについても，学術的・社会的に意義がある．
また，説明可能なAI（XAI）を用いて，住民避難行動のアンケート調査データに対して要因分析したところ，従来の統計手法で得られた結果と概ね同じ結果でありつつも，新たな結果が示されたことは，学術的にも社会的にも意義がある．