A Python-Based SVR-FLAHP Computational Framework for Landslide Vulnerability Modeling

The relocation of settlements due to strategic development projects, such as the Bagong Dam construction in Trenggalek, necessitates accurate landslide vulnerability analysis in landslide-prone areas. Conventional assessment methods often face limitations in handling subjectivity, implementation complexity, and data dependency, highlighting the need for a more objective and flexible approach. This study addresses these challenges by developing an interactive Python-based computational framework, SVR-FLAHP-Sys, that integrates Support Vector Regression (SVR) for objective prediction and Fuzzy Logic-AHP (FL-AHP) for robust weighting. A key innovation is the framework's ability to overcome data scarcity through automated training data generation using the Gemini API. The Python ecosystem was chosen for its mature scientific libraries, such as scikit-learn, which facilitate the seamless integration of these complex methods into a cohesive, reproducible workflow. The implemented framework demonstrates high computational efficiency, capable of completing the entire analysis workflow in under a minute. The case study analysis successfully identified the bedrock layer (36.95%) and slope gradient (30.61%) as the most dominant factors, producing a final vulnerability map where approximately 45% of the study area is classified as moderate to high risk. This research contributes a functional framework that leverages machine learning for objective, transparent, and reproducible vulnerability analysis, providing an accessible tool to directly support evidence-based policymaking for safer settlement planning in disaster-prone regions.