Disentangling Spatial and Structural Drivers of Housing Prices through Bayesian Networks: A Case Study of Madrid, Barcelona, and Valencia

2025-06-11

AI Summary

This study investigates the heterogeneous mechanisms through which spatial factors (e.g., transport accessibility) and structural factors (e.g., building attributes, housing typologies) differentially influence housing prices across Madrid, Barcelona, and Valencia. Leveraging over 180,000 geocoded property listings, we develop an interpretable discrete Bayesian network model that integrates multi-source drivers to enable joint inference, scenario simulation, and sensitivity analysis. Our analysis reveals city-specific pricing logics: Madrid exhibits infrastructure-driven spatial stratification; Barcelona is predominantly shaped by unit typology classification; and Valencia relies on synergistic interactions between spatial and structural fundamentals. This framework enhances policy transparency and operationalizability by quantifying context-dependent factor contributions, offering a robust, reproducible analytical tool for equitable urban governance and evidence-based housing policy design.

Technology Category

• Reasoning under Uncertainty: Relational Probabilistic Models
• Machine Learning: Calibration & Uncertainty Quantification
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding

Abstract

Understanding how housing prices respond to spatial accessibility, structural attributes, and typological distinctions is central to contemporary urban research and policy. In cities marked by affordability stress and market segmentation, models that offer both predictive capability and interpretive clarity are increasingly needed. This study applies discrete Bayesian networks to model residential price formation across Madrid, Barcelona, and Valencia using over 180,000 geo-referenced housing listings. The resulting probabilistic structures reveal distinct city-specific logics. Madrid exhibits amenity-driven stratification, Barcelona emphasizes typology and classification, while Valencia is shaped by spatial and structural fundamentals. By enabling joint inference, scenario simulation, and sensitivity analysis within a transparent framework, the approach advances housing analytics toward models that are not only accurate but actionable, interpretable, and aligned with the demands of equitable urban governance.

Problem

• Analyzing spatial and structural factors affecting housing prices
• Developing interpretable models for urban housing market segmentation
• Comparing city-specific price formation in Madrid, Barcelona, Valencia

Innovation

• Uses Bayesian networks for housing price analysis
• Analyzes 180,000 geo-referenced housing listings
• Enables scenario simulation and sensitivity analysis