<![CDATA[Purpose – This exploratory pilot study aims to develop an interpretable machine learning framework for predicting social vulnerability in the coastal areas of Cilacap Regency, Indonesia. Design/methods/approach – The study utilized an optimised eXtreme Gradient Boosting (XGBoost) algorithm integrated with Explainable Artificial Intelligence (XAI) using SHapley Additive exPlanations (SHAP), based on socio-economic data collected from seven coastal sub-districts during 2024–2025. The dataset includes indicators such as poverty rate, population density, education level, and infrastructure accessibility. Due to the limited sample size (n = 7), model evaluation was conducted using Leave-One-Out Cross-Validation (LOOCV). Findings – The proposed model achieved an accuracy of 94.2% and an AUC-ROC value of 0.96 within the experimental dataset. SHAP analysis indicates that socio-economic vulnerability and geographic isolation are among the most influential contributing factors. Research implications/limitations – Nevertheless, the findings should be cautiously interpreted because the limited regional dataset constrains broader generalisability. Despite these limitations, this study demonstrates the potential of integrating machine learning and XAI to provide interpretable and data-driven insights that may support policymakers in prioritising mitigation strategies for high-risk coastal communities. Originality/value – Social vulnerability is a critical component of disaster risk management, particularly in coastal regions where complex socio-economic factors interact dynamically. However, conventional approaches often fail to capture non-linear relationships and provide limited predictive interpretability. ]]>
Copyrights © 2026
