About this project
An end-to-end ML project: scraping real estate listings, engineering regional features, training regression models, and evaluating price predictions across different Tunisian governorates.
Focus was on realistic feature engineering for a market where listing quality varies wildly, and on making the model actually useful rather than just accurate on a test split.
Key features
Data scraping & cleaning
Listings collected and normalized from multiple sources.
Regional feature engineering
Location, amenities, and infrastructure signals.
Regression benchmarks
Compared linear, tree-based, and gradient boosting models.
Tags
machine learningreal estatetunisiaregressiondata science