Comparative Analysis of Machine Learning Models for Real Estate Price Prediction in Urban City

Real estate price estimations play a crucial role in enabling informed decisions by buyers, investors, and policymakers. Traditional statistical methods often fail to capture nonlinear patterns and diverse feature interactions present in housing data. This paper presents a comparative analysis of three advanced machine learning models (XGBoost, CatBoost, and TabNet) for real estate price prediction using structured tabular data. The dataset comprises key property features, such as area, location, number of bedrooms, and amenities, and was sourced from an open-access Kaggle repository. Rigorous pre-processing, including feature encoding, outlier removal, and log transformation, was applied. The models were evaluated using Mean Squared Error (MSE), Mean Absolute Error (MAE), and Coefficient of Determination or R-squared (R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>) metrics. Among the three, CatBoost achieved superior performance across all evaluation criteria. The results affirm that machine learning techniques, particularly CatBoost, offer a robust alternative to traditional approaches in real estate valuation, laying the groundwork for automated and data-driven price-estimation systems.