Gohar-Hany/Egypt-Real-Estate-Predictive-Analytics

An end-to-end Machine Learning Pipeline, Flask API Serving Engine & Glassmorphic Analytics Dashboard for Egypt's Real Estate Market. # NileValue AI - Egypt Real Estate Predictive Analytics & Valuation Engine ### *An End-to-End scikit-learn ML Pipeline, Flask API Serving Engine & Premium Glassmorphic Analytics Dashboard* ?logo=chartdotjs&logoColor=white&style=for-the-badge) NileValue AI is a state-of-the-art predictive analytics and automated valuation model (AVM) designed for the Egyptian real estate market. The project features a fully automated, self-contained scikit-learn machine learning pipeline that cleans raw listings, engineers high-value features using Regex and Natural Language Processing (NLP), and returns highly accurate EGP price estimates. Predictions are served in real-time by a Flask API and displayed on a stunning, responsive **Glassmorphic Dark Theme Dashboard** featuring floating neon elements, dynamic text highlights, and interactive analytical charts. --- ## 📸 Dashboard Preview ### Scenario A: Luxury Golf-View Villa (Giza) > Instant AI Valuation showing active NLP keyword tagging and comparative regional intelligence in Giza. ### Scenario B: Premium Installments Apartment (Cairo) > Valuation output illustrating customized pricing maps, payment plans, and down payment details. --- ## 🚀 Key Features * **Custom Feature Preprocessing Pipeline**: A modular, custom-engineered scikit-learn transformer parses mixed formatting, synchronizes square feet to square meters, extracts studio configurations, and splits address hierarchies into Governorates and Districts. * **NLP Text-Based Feature Extraction**: Analyzes property descriptions in English and Arabic in real-time, engineering boolean indicator columns for high-value amenities (e.g. finishes, private pools, sea views). * **Log-Scale Target Regressions**: Resolves the extreme right-skewed nature of real estate values (ranging from 100K EGP to 500M EGP) using log-transformed target variables ($y' = \log(y + 1)$) inside a gradient boosting regression framework. * **Production Serving Engine**: A solid, multit …