House price prediction in Andalusia: a comparative study of hedonic and machine-learning approaches using online data

Abstract

This study explores housing price prediction in six major cities of Andalusia using a combination of traditional and machine learning models. Online property listings were compiled to build a detailed dataset incorporating structural, socioeconomic, and locational variables. A traditional hedonic model was estimated using Ordinary Least Squares (OLS), and its performance was compared with that of Random Forest, K-Nearest Neighbours, and XGBoost. Results show that linear models tend to underperform, highlighting the relevance of non-linear approaches in capturing the complex dynamics of the housing market. To enhance interpretability, SHapley Additive exPlanations (SHAP) values were applied to the XGBoost model, enabling the identification of the most influential predictors in each city. While structural variables remained important across all models, socioeconomic and locational features exhibited varying effects depending on the city. These findings underscore the potential of combining rich data sources with interpretable machine learning techniques for urban housing market analysis.Este estudio analiza la prediccion de precios de la ´ vivienda en seis ciudades principales de Andaluc´ıa mediante una combinacion de modelos tradicionales ´ y de aprendizaje automatico. Se recopilaron anuncios ´ inmobiliarios en l´ınea para construir un conjunto de datos detallado que incorpora variables estructurales, socioeconomicas y de localizaci ´ on. Se estim ´ o un mod- ´ elo hedonico tradicional mediante M ´ ´ınimos Cuadrados Ordinarios (MCO) y se comparo su rendimiento ´ con el de Random Forest, K-Nearest Neighbours y XGBoost. Los resultados muestran que los modelos lineales tienden a tener un desempeno inferior, lo ˜ que destaca la relevancia de los enfoques no lineales para capturar la compleja dinamica del mercado in- ´ mobiliario. Para mejorar la interpretabilidad, se aplicaron valores SHapley Additive exPlanations (SHAP) al modelo XGBoost, lo que permitio identificar los ´ predictores mas influyentes en cada ciudad. Si bien las ´ variables estructurales siguieron siendo importantes en todos los modelos, las caracter´ısticas socioeconomicas ´ y de localizacion mostraron efectos variables seg ´ un´ la ciudad. Estos hallazgos subrayan el potencial de combinar fuentes de datos con tecnicas de aprendizaje ´ automatico interpretables para el an ´ alisis del mercado ´ de la vivienda urbana.