<![CDATA[Urban flooding poses a growing challenge in rapidly urbanizing regions due to the combined effects of climate variability, land-use change, and infrastructure limitations. This study proposes a hybrid framework integrating the Fuzzy Analytical Hierarchy Process (Fuzzy-AHP), ensemble machine learning, and sensitivity analysis to support urban flood risk assessment. Fuzzy-AHP is employed to incorporate expert judgment and address uncertainty through triangular fuzzy numbers, while Random Forest and XGBoost are used to capture non-linear relationships and temporal patterns in heterogeneous flood-related data. The framework is applied to 1,008 observations from 12 districts in Bekasi City, Indonesia, covering the period 2018–2024. Model performance indicates strong discriminatory capability in distinguishing flood and non-flood conditions. Sensitivity analysis is explicitly positioned as a policy-oriented diagnostic and prioritization tool, enabling the identification of influential variables relevant for seasonal planning and early warning strategies. The results highlight the dominant role of climate-related factors, particularly rainfall and temporal variables, in shaping urban flood risk. Overall, the proposed framework demonstrates the complementary integration of expert knowledge and data-driven learning, offering a transferable methodological reference for flood risk assessment in complex urban environments.]]>
Copyrights © 2025
