Tabular Foundation Models for Discrete Choice Estimation

2026-07-14

AI Summary

This study addresses the limitations of conventional tabular foundation models in discrete choice modeling, which typically neglect choice-set dependencies and heterogeneity in consumer preferences. The work proposes a novel adaptation of tabular foundation models to discrete choice problems through a data restructuring approach that explicitly encodes choice-set structure and individual-specific preferences within a row-wise learning framework. By integrating in-context learning with population-level fine-tuning, the method significantly enhances predictive performance for users with sparse behavioral data. Empirical evaluation on a yogurt purchase panel dataset demonstrates that the proposed model achieves an 8% improvement in log-likelihood and a 3.6% increase in hit rate compared to a hierarchical Bayesian benchmark, while accelerating inference by 16×. These gains are particularly pronounced at moderate data scales, underscoring the model’s practical efficacy in realistic market research settings.

Abstract

Tabular foundation models (TFMs) generate predictions on structured data via in-context learning, without task-specific estimation. We ask whether TFMs can be effectively applied to discrete choice, a central demand estimation framework in marketing and operations, and find that directly applying TFMs yields limited performance. The gap is structural: TFMs assume row-independent observations, whereas discrete choice is inherently set-valued and subject to persistent consumer preference heterogeneity. We propose a reformulation that encodes both choice-set dependence and individual heterogeneity within a row-based learning framework. Evaluated on a yogurt scanner panel, individual-level heterogeneity encoding is the dominant driver of predictive accuracy. The best reformulation outperforms hierarchical Bayesian estimation by 8\% in holdout log-likelihood and 3.6\% in hit rate, running 16 times faster, a practical advantage for large-scale demand estimation. The advantage is largest in the medium-data regime (10--40 purchase occasions per consumer), where parametric Bayesian shrinkage most distorts estimates for atypical consumers. Fine-tuning on population choice data provides additional gains for consumers with shallow purchase histories, where in-context learning has limited individual-specific signal to condition on. These results establish a principled approach for applying foundation models to consumer choice problems more broadly.

Problem

• Tabular Foundation Models
• Discrete Choice
• Preference Heterogeneity
• Choice-Set Dependence
• Demand Estimation

Innovation

• Tabular Foundation Models
• Discrete Choice
• Preference Heterogeneity
• In-context Learning
• Demand Estimation