# Egypt Real Estate Appraiser đźŹ
**AI-powered real estate price prediction system for the Egyptian market**
A production-ready machine learning application that predicts property prices using advanced regression models trained on 16,000+ Egyptian real estate listings.
---
## 🎯 Project Overview
This graduation project delivers a complete ML pipeline for real estate price estimation in Egypt, featuring:
- **Data-driven preprocessing**: Intelligent handling of semi-structured real estate data
- **Feature engineering**: Location grouping, price normalization, size categorization
- **Multiple ML models**: Linear, Ridge, Random Forest, Gradient Boosting, XGBoost
- **Hyperparameter tuning**: GridSearchCV for optimal model performance
- **Interactive web app**: Streamlit interface for instant price predictions
- **Production quality**: Full pipeline with validation, error handling, and logging
---
## 📊 Dataset Characteristics
### Data Source
- **File**: `data/egypt_real_estate_listings.csv`
- **Records**: ~16,000 properties
- **Features**: 7 raw columns
### Key Columns
- `price`: Target variable (EGP) - property price in Egyptian Pounds
- `size`: Area extracted as `size_sqm` (square meters)
- `bedrooms`: Number of bedrooms (formatted as "3", "7+", "3+ Maid")
- `bathrooms`: Number of bathrooms
- `location`: Geographic location of property
- `type`: Property category (Apartment, Villa, Duplex, etc.)
- `payment_method`: Payment type (Cash or Installment)
### Data Challenges & Solutions
| Challenge | Solution |
|-----------|----------|
| Mixed character formats in bedrooms ("7+", "3+ Maid") | Regex extraction of leading number |
| Size formatted as text with units | Extraction of numeric value after "/" |
| High location cardinality (500+ unique values) | Intelligent grouping into 6 regions |
| Missing values in bathrooms/bedrooms | Median/mode imputation |
| 5% of price/size outliers | IQR-based removal |
| Duplicates in dataset | Exact match deduplication |
---
## …