birhanumoges/Ethiopia-food-security-analytics

NLP and Machine Learning analysis of Ethiopian food markets, crop yields, and agricultural practices to monitor food security, identify trends, and support data-driven policy decisions. # Ethiopian Food Security & Price Analytics System A comprehensive data analytics and machine learning system for forecasting food prices, detecting market crises, and monitoring food security in Ethiopia. This project combines text analytics, time-series forecasting, and early warning systems to support food security monitoring and policy-making. ## 📋 Project Overview This project analyzes World Food Programme (WFP) food price data from Ethiopian markets to: - **Forecast** commodity prices across regions using advanced ML models - **Detect** price anomalies and early warning signals for food crises - **Classify** commodities by market behavior (stable, seasonal, volatile) - **Visualize** regional food security risks on interactive maps - **Monitor** inflation shocks and market volatility ## 🎯 Key Components ### 1. **Market Text Analytics** (`market_text_analytics.ipynb`) Natural Language Processing and descriptive analytics on food commodity data. **Features:** - Data cleaning and NLP preprocessing - Commodity standardization and text normalization - TF-IDF vectorization and topic modeling (LDA) - Commodity clustering analysis (K-means) - Demand and consumption pattern analysis - Price volatility detection (food security risk indicator) - Geographic commodity distribution heatmaps - Interactive Folium maps with risk-weighted markers **Key Insights Generated:** - Top demanded commodities by region - Price volatility rankings (food risk indicator) - Market distribution analysis - Seasonal price trends by commodity category - Regional price inequality patterns --- ### 2. **Price Forecasting** (`Price_Forecasting.ipynb`) Multi-model ensemble forecasting system with recursive 6-month predictions. **Models Included:** - Linear Regression (baseline) - Random Forest (400 estimators) - Gradient Boosting - XGBoost (with hyperparameter tuning) - LightGBM (for large datasets) - **Weighted Ensemble** (automatically weights models by inverse error) **Advanced Features …