hannagetachew070421/Ethiopian-Soil-Erosion-

Ethiopian Soil Erosion Prediction Model 🇪🇹🌱 Overview This project focuses on predicting and analyzing soil erosion regimes in Ethiopia using advanced machine learning and deep learning techniques.By integrating physical geographical models with data science, the project aims to provide actionable insights for environmental conservation and sustainable land management. Key Features Hybrid Modeling: Combines the Revised Universal Soil Loss Equation (RUSLE) with deep learning frameworks. Feature Extraction: Utilizes Self-Supervised Autoencoders to extract high-level features from complex soil and environmental datasets. High Performance: Implementation of the CatBoost algorithm for robust classification and regression tasks. Sensitivity Analysis: Includes detailed sensitivity analysis on raw soil datasets to identify primary erosion drivers. Technology Stack Programming Language: Python Libraries:CatBoost for predictive modeling PyTorch/TensorFlow for Autoencoder implementation Scikit-learn for clustering and preprocessing Pandas & NumPy for data manipulation Project Structure/models: Contains the trained catboost_erosion_model.cbm and .joblib files. /src: Source code for feature extraction (Autoencoders) and model training. /ux: UI/UX design assets for the project interface. /data: Sample datasets and RUSLE-based cluster information.