Graph neural networks for residential location choice: connection to classical logit models

2025-07-28

AI Summary

Classical discrete choice models (DCMs) explicitly capture structural relationships among alternatives—such as nesting or spatial adjacency—but existing deep learning approaches lack mechanisms to encode such dependencies explicitly within a theoretically grounded framework. Method: We propose Graph Neural Network-based Discrete Choice Models (GNN-DCMs), which represent spatial adjacency among alternatives as a graph and employ message-passing to model spatial interdependence, while rigorously embedding the Random Utility Maximization (RUM) principle. Contribution/Results: GNN-DCMs unify Multinomial Logit (MNL), Nested Logit, and Spatially Correlated Logit within a single, interpretable architecture that preserves theoretical consistency while enhancing expressive power. Empirical evaluation on residential choice across 77 Chicago neighborhoods demonstrates that GNN-DCMs significantly outperform MNL, Spatially Correlated Logit (SCL), and feedforward neural networks in predictive accuracy. Moreover, the model effectively uncovers individual heterogeneity and spatially aware choice patterns, establishing a novel paradigm that bridges deep learning with classical choice theory.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Reasoning under Uncertainty: Graphical Models
• Game Theory and Economic Paradigms: Social Choice / Voting

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Economics, Online Markets and Human Computation: Social networks and social learning
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Researchers have adopted deep learning for classical discrete choice analysis as it can capture complex feature relationships and achieve higher predictive performance. However, the existing deep learning approaches cannot explicitly capture the relationship among choice alternatives, which has been a long-lasting focus in classical discrete choice models. To address the gap, this paper introduces Graph Neural Network (GNN) as a novel framework to analyze residential location choice. The GNN-based discrete choice models (GNN-DCMs) offer a structured approach for neural networks to capture dependence among spatial alternatives, while maintaining clear connections to classical random utility theory. Theoretically, we demonstrate that the GNN-DCMs incorporate the nested logit (NL) model and the spatially correlated logit (SCL) model as two specific cases, yielding novel algorithmic interpretation through message passing among alternatives' utilities. Empirically, the GNN-DCMs outperform benchmark MNL, SCL, and feedforward neural networks in predicting residential location choices among Chicago's 77 community areas. Regarding model interpretation, the GNN-DCMs can capture individual heterogeneity and exhibit spatially-aware substitution patterns. Overall, these results highlight the potential of GNN-DCMs as a unified and expressive framework for synergizing discrete choice modeling and deep learning in the complex spatial choice contexts.

Problem

• Captures complex feature relationships in location choice
• Models dependence among spatial alternatives explicitly
• Connects deep learning with classical utility theory

Innovation

• GNN captures spatial choice dependencies
• Connects GNN-DCMs to classical logit models
• Outperforms MNL, SCL, and neural networks