MarialRK/flood-prediction-sudd-wetland

Interpretable Spatiotemporal Flood-Risk Prediction System for the Sudd Wetland Region of South Sudan - ALU Capstone Project # 🌊 Interpretable Spatiotemporal Flood-Risk Prediction System ## Sudd Wetland Region, South Sudan ### ALU Capstone Project | BSc. Software Engineering **Student:** Daniel Marial Reng Kudum **Supervisor:** Hubert Apana **Date:** June 13, 2026 --- ## 📌 Project Description This project develops an **interpretable machine learning-based flood-risk prediction system** for the Sudd Wetland Region of South Sudan – one of the most flood-vulnerable areas in Africa. The system uses environmental data (rainfall, water levels, vegetation health) to predict flood occurrence with **94.44% accuracy** and **100% recall** (catches every flood event). ### Key Features - ✅ **Data Engineering** – 15 years of monthly environmental data (2010-2024) - ✅ **Data Visualization** – 6 plots showing distributions, correlations, and time-series analysis - ✅ **Machine Learning Model** – Random Forest Classifier with 100 trees - ✅ **Performance Metrics** – Accuracy (94.44%), Precision (80%), Recall (100%), F1-Score (88.89%) - ✅ **Feature Importance** – Identifies Water Level and Rainfall as primary flood drivers --- ## 📂 Repository Structure ``` flood-prediction-sudd-wetland/ ├── flood_prediction_demo.ipynb # Main Jupyter notebook with model ├── screenshots/ # Visual outputs from the notebook │ ├── 01_rainfall_distribution.png │ ├── 02_correlation_heatmap.png │ ├── 03_performance_metrics.png │ └── 04_feature_importance.png ├── requirements.txt # Python dependencies └── README.md # This file ``` --- ## 🚀 How to Set Up the Environment ### Prerequisites - Python 3.11 or higher - Git - VS Code (recommended) or any code editor ### Step 1: Clone the Repository ```bash git clone github.com cd flood-prediction-sudd-wetland ``` ### Step 2: Create a Virtual Environment **Windows:** ```bash python -m venv venv venv\Scripts\activate ``` **Mac/Linux:** ```bash python3 -m ve …