tactikalkofi/PREDICTING-WEEKLY-AND-MONTHLY-RAINFALL-IN-GHANA-WEST-AFRICA-USING-LONG-SHORT-TERM-MEMORY

# 🌧️ Ghana Rainfall Prediction System > **AI-Powered Weekly and Monthly Rainfall Forecasting across Ghana's Climate Zones** Predicting weekly and monthly rainfall in Ghana using **Long Short-Term Memory (LSTM)** networks, **XGBoost**, **Random Forest**, and **ARIMA** models. This system supports climate-resilient planning for agriculture, water resource management, and disaster preparedness. --- ## 📋 Table of Contents - Overview - Key Features - Project Structure - Installation - Usage - Model Performance - Data Description - Methodology - Results - Deployment - Contributing - Citation - License - Contact --- ## 🌍 Overview This research project develops and evaluates machine learning models for predicting rainfall across **six meteorological stations** representing Ghana's three distinct climate zones: | Climate Zone | Stations | Characteristics | |-------------|----------|----------------| | **Savannah** | Tamale, Navrongo | Unimodal rainfall pattern | | **Forest** | Kumasi, Ho | Bimodal rainfall pattern | | **Coastal** | Accra, Takoradi | Asymmetrical bimodal pattern | ### 🎯 Research Objectives 1. **Develop optimized LSTM models** for weekly and monthly rainfall prediction 2. **Compare four forecasting approaches**: LSTM, XGBoost, Random Forest, and ARIMA 3. **Analyze spatial patterns** across Ghana's distinct climate zones --- ## ✨ Key Features - 🤖 **4 Machine Learning Models**: LSTM, XGBoost, Random Forest, ARIMA - 📊 **Dual Prediction Horizons**: Weekly and monthly forecasts - 🗺️ **6 Meteorological Stations**: Complete coverage of Ghana's climate zones - 📈 **High Accuracy**: R² > 0.85 for monthly predictions using LSTM - 🌐 **Interactive Dashboard**: Real-time predictions with Streamlit - 📱 **Responsive Design**: Works on desktop, tablet, and mobile - 📉 **Comprehensive Visualizations**: Historical trends, predictions, and model comparisons --- ## 📁 Project Structure ``` ghana-rainfall-prediction/ │ ├── 📄 app.py # Main St …