Treffer: Code Repository for 'Uncovering Spatiotemporal Urban Flood Dynamics: An Explainable GeoAI Approach to Land Cover Change Over Two Decades'

This repository contains the Python scripts and related files used in the study titled "Explainable GeoAI for Assessing the Impact of Land Use Dynamics on Flood Susceptibility Evolution." The research combines geospatial artificial intelligence (GeoAI) with explainable machine learning to investigate how temporal changes in land use contribute to the evolution of flood susceptibility. The study employs Random Forest and Logistic Regression models to predict flood-prone areas. It incorporates SHapley Additive exPlanations (SHAP) to interpret the contribution of individual predictors, with an emphasis on land use transformation over time. Included in this repository: Data preprocessing and feature engineering scripts Model training, cross-validation, and evaluation workflows SHAP-based interpretability analysis Visualisation scripts for SHAP summary plots and flood susceptibility maps The dataset used in this study is not publicly available due to confidentiality agreements. While the dataset used in the study is not publicly available due to confidentiality constraints, the code is structured for adaptability to other spatio-temporal flood-related datasets