Ali-Elgendy/coastal-vulnerability-ml-egypt

Code and datasets for comparing data-driven machine learning architectures (XGBoost) against traditional knowledge-driven indices (InVEST) for coastal vulnerability assessment along the Egyptian Mediterranean coast. # Coastal Vulnerability Assessment: Machine Learning vs. Traditional Models (Egypt) This repository contains the datasets and Python code utilized in the comparative coastal vulnerability assessment of the Egyptian Mediterranean Coast. ## Overview The study directly compares the traditional knowledge-driven InVEST Coastal Vulnerability Index against an advanced data-driven machine learning architecture (XGBoost). The model processes 40 years of spatiotemporal shoreline morphodynamics (1985-2025) alongside high-resolution hydrodynamic and anthropogenic drivers. ## Repository Contents * `data/`: Contains the processed 100-meter transect datasets, including the 8 physical/hydrodynamic variables and the historical Linear Regression Rate (LRR). * `notebooks/`: Jupyter notebooks containing the data preprocessing, model training (XGBoost, RF, SVM, ANN), hyperparameter tuning, and performance evaluation. * `results/`: Output feature importance metrics and vulnerability classifications. ## Usage To run the notebooks, ensure you have the following dependencies installed: `pip install pandas numpy scikit-learn xgboost matplotlib seaborn` ## Citation If you utilize this code or dataset, please cite the corresponding paper: > Elgendy, A., Morgan, H., El-Askary, H., & Mucsi, L. (2026). Comparing Knowledge-Driven and Data-Driven Machine Learning Frameworks for Coastal Vulnerability Assessment on the Egyptian North Coast.