Search

• This study develops a disaster risk index (DRI) for assessing disaster risks in coastal communities. • Logit Boost ML model was found the optimal model for this research. • Gabura had the highest DRI value of 0.09, ahead of Southkhali (0.06) and Banishanta (0.04) unions. • The study underscores the importance of evolving from traditional approaches to advanced data-driven tools. Communities in disaster-prone coastal areas are more vulnerable to socioeconomic, environmental, and ecological problems. Knowing how disaster-prone coastal communities are, and, ideally, their resilience against current disaster risks, is an essential first step in reducing consequences. Therefore, this study aims to develop a disaster risk index (DRI) for assessing disaster risks in three coastal communities in southwest Bangladesh. Five machine learning (ML) models and holistic approaches were adopted to conduct the analysis. This study utilized ML-based feature selection methods and considered hazard, exposure, and vulnerability components to identify the most influential indicators for evaluating disaster risk. The researchers systematically developed six distinct feature sets using filter-based feature selection methods. Five ML models then evaluate these sets to determine the optimal set of features. Evaluation metrics were used to validate the performance of ML models, such as accuracy, precision, F-measure, recall, and the receiver operating characteristics (ROC) curve. Based on the Logit Boost ML model, the optimal feature set (set 3) had the best accuracy (0.997), precision (0.997), recall (0.997), F-measure (0.997), and ROC (0.996) compared to other feature selection sets. According to the findings, Gabura had the highest DRI value of 0.09, ahead of the Southkhali (0.06) and Banishanta (0.04) unions. The study underscores the importance of evolving from traditional approaches to advanced data-driven tools to understand disaster risks. The paper's findings can provide policymakers in coastal areas with new insight into the disaster risk differences among the inhabitants, leading to more effective disaster management practices and risk reduction strategies. [ABSTRACT FROM AUTHOR]

• This paper provides a comprehensive conceptual analysis of disaster resilience in climate-vulnerable communities. • The research employs the Walker and Avant Method to elucidate the notion of resilience within the framework of climate vulnerability. • A practical framework and a set of ecological indicators are proposed for environmental impact assessments in disaster resilience. • The research provides tangible illustrations of the proposed approaches and indicators in practical contexts. Resilience is a prior term of debate in vulnerability research irrespective of discipline. Nowadays the resilience concept has been used abnormally in many cases without realizing its intended meaning. Therefore, this study clarifies the resilience concept in the context of climate vulnerability. An integrated review of the literature has been done by the PRISMA approach to identify the most relevant documents. Besides Walker and Avant Method has been used to analyze the concept of resilience. The use of concept in different disciplines, historical perspectives, dimensions, attributes, consequences, and relationship with vulnerability and adaptability have been reviewed to clarify it. This study also provides a theoretical and workable definition of resilience from the perspective of climate change and natural hazards. A conceptual framework has been developed with concept mapping, contributing to the existing debate on resilience study. A case analysis has been done by focusing on a climate-vulnerable community living in the riverine islands in Bangladesh. This article shows that socio-ecological systems approach to defining small island disaster resilience has shortcomings. Additionally, the research-policy divide in resilience studies is exacerbated by the paucity of studies that use the participatory approach to identify disaster resilience indicators. The study makes a compelling case for further research on how to turn academic discoveries into a resource that might benefit local people, particularly those on small islands. The study suggests that social community-based resilience should be developed to protect geographically isolated communities. [ABSTRACT FROM AUTHOR]