amanbr86/Ethiopian-Food-Price-Prediction-Model

# 🌾 Ethiopian Food Price Prediction — ML Pipeline > **WFP Market Price Data · 60,343 records · 96 Commodities · 128 Markets · 2000–2026** > Masters Program — Machine Learning Final Project · 2026 --- ## 📋 Project Overview Food price volatility in Ethiopia directly impacts food security for millions of people. This project builds a full **machine learning pipeline** that predicts retail and wholesale food prices (ETB) across 128 Ethiopian markets and 96 commodity types, using 26 years of WFP market price data. The project is deployed as a **Streamlit web application** where users can select a commodity, market, and forecast date to instantly predict the expected price. ### Results | Metric | Value | |---|---| | Best Model | RandomForest (tuned) | | Test R² | **0.872** — 87.2% of variance explained | | Test RMSE | **1,245 ETB** average error | | Test MAE | **668 ETB** median absolute error | | Train R² | 0.966 (healthy gap — not severe overfitting) | --- ## 📁 Project Structure ``` food_price_prediction/ ├── data/ │ └── wfp_food_prices_eth.csv ← WFP price data (place here) ├── models/ ← auto-created by train.py │ ├── best_model.pkl ← fitted sklearn Pipeline │ ├── encoders.pkl ← LabelEncoders for Streamlit dropdowns │ ├── feature_names.json ← metadata, metrics, commodity lists │ ├── training_report.txt ← human-readable performance summary │ ├── eda_report.png ← 4-panel EDA chart │ └── diagnostics.png ← residuals + feature importance chart ├── train.py ← full 8-step ML pipeline ├── app.py ← Streamlit deployment app └── requirements.txt ← Python dependencies ``` --- ## ⚙️ Quick Start ### 1. Install dependencies ```bash pip install -r requirements.txt ``` ### 2. Train the model (runs all 8 pipeline steps) ```bash python train.py ``` Expected runtime: **5–15 minu …