Estifanos-Abera/house-price-predictor

This project started as a beginner ML exercise and has grown into a full Ethiopian real estate prediction system. It compares 4 machine learning models on a custom dataset of 500 houses across Addis Ababa, with prices and features grounded in real market research. Key improvements over v1: Dataset expanded from 6 → 500 houses 3 features → 14 feat # 🏠 Ethiopian House Price Predictor > A machine learning project that predicts residential property prices across **19 Addis Ababa neighborhoods**, built on realistic 2025–2026 Ethiopian market data. --- ## 📌 About This project started as a beginner ML exercise and has grown into a full Ethiopian real estate prediction system. It compares **4 machine learning models** on a custom dataset of **500 houses** across Addis Ababa, with prices and features grounded in real market research. **Key improvements over v1:** - Dataset expanded from 6 → 500 houses - 3 features → 14 features (location proximity, condition, parking, etc.) - 1 model → 4 models compared with cross-validation - Console-only → full interactive web app (Streamlit) - Generic data → realistic Ethiopian/Addis Ababa market prices in ETB --- ## 🏙️ Neighborhoods Covered | Zone | Neighborhoods | Price Range | |------|--------------|-------------| | **Premium** | Bole, Old Airport, Kazanchis, Sarbet | 40M – 150M ETB | | **Mid-High** | Megenagna, CMC, Gerji, Summit | 20M – 60M ETB | | **Mid** | Yeka, Ayat, Lebu, Lideta, Piassa, Kirkos, Bole Bulbula | 8M – 35M ETB | | **Emerging** | Goro, Akaki Kality, Kolfe, Jemo | 4M – 20M ETB | > Prices reflect 2025–2026 market data. Bole & Kazanchis command premiums due to proximity to embassies, the airport, and international businesses. --- ## 🧠 Models Compared | Model | Strengths | Weaknesses | |-------|-----------|------------| | **Linear Regression** | Fast, interpretable | Assumes linearity | | **Ridge Regression** | Handles multicollinearity | Still linear | | **Random Forest** | Captures non-linear patterns, robust | Less interpretable | | **Gradient Boosting** | Highest accuracy on tabular data | Slower to train | --- ## 🗂️ Project Structure ``` house-price-predictor/ ├── ethiopian_house_prices.csv ← 500-row dataset (14 features) ├── house_price_predictor.py ← CLI model training & comparison ├── app.py ← Streamlit web app …