abdotamer55/Egypt-Houses-Price

An end-to-end Machine Learning project for predicting Egyptian real estate prices using data analysis, feature engineering, model development, and interactive deployment with Streamlit and Power BI. An end-to-end Machine Learning web application designed to predict residential property prices in Egypt. The project processes raw real estate data through a robust data pipeline—encompassing rigorous cleaning, advanced feature engineering, target encoding to prevent data leakage, and feature importance selection—before serving predictions via an intuitive Streamlit multi-page interface. --- ## 🚀 Features & Architecture The project is structured into a modular, production-ready pipeline split into distinct operational layers: 1. **Data Cleaning (`data_cleaning.py`)**: Outlier rejection, duplicate filtering, and structural numeric coercion. 2. **Feature Engineering (`featuer_engineering.py`)**: Construction of non-linear interaction features and domain-specific ratios. 3. **Feature Selection & Encoding (`featuer_selection.py`)**: Safe Target Encoding split mechanics to prevent data leakage and Random Forest feature importance filtering. 4. **Model Training & Evaluation (`Model.py`)**: Benchmarking multiple regression algorithms (Ridge, Decision Trees, Random Forest, LightGBM, XGBoost). 5. **Prediction Pipeline (`Predection.py`)**: A backend routing script that passes raw web user input safely through the saved preprocessing assets. 6. **Multi-page UI Application (`App.py`)**: An elegant interactive client interface built using Streamlit. --- ## 📊 Detailed Pipeline Breakdown ### 1. Data Cleaning & Outlier Rejection * **Handling Missing Values**: Drops missing target instances (`Price`) and standardizes continuous attributes via median statistical imputation. * **Percentile-Based Trimming**: Extreme luxury properties or pricing anomalies are cut dynamically using defensive percentile filters (Lower: 0.5%, Upper: 99%) across Price, Area, Bedrooms, and Bathrooms to stabilize gradients. ### 2. Advanced Feature Engineering To extract maximum predictive power from the real estate attributes, several engineering steps were implemented: * **Ratios & Proportions**: Com …