arXiv 14 Jul 2026 · Machine Learning
arXiv:2607.13314 · PDF · DOI · OpenAlex · Extracted main text
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 on both holdout log-likelihood and 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.
appendix boundary found by appendix_command · 95% of the source is main text. Read the extracted text to check this.
The works this paper leans on most, across its whole bibliography — not restricted to papers in our corpus. Ranked by composite intensity, which combines how often a work is mentioned, how many sections mention it, and how much of that falls in the main text rather than the appendix.
| Reference | Intensity | Mentions | Sections | Main text | |
|---|---|---|---|---|---|
| 1 | Hollmann, Noah and Müller, Samuel and Purucker, Lennart and Krishnak… (2025) Accurate predictions on small data with a tabular foundation model | 1.000 | 6 | 4 | 100% |
| 2 | Rossi, Peter E and Allenby, Greg M and McCulloch, Robert (2005) Bayesian Statistics and Marketing | 0.950 | 7 | 6 | 86% |
| 3 | Allenby, Greg M and Rossi, Peter E (1998) Marketing Models of Consumer Heterogeneity | 0.928 | 5 | 4 | 80% |
| 4 | Train, Kenneth E (2009) Discrete choice methods with simulation | 0.843 | 4 | 4 | 75% |
| 5 | Hollmann, Noah and Müller, Samuel and Eggensperger, Katharina and Hu… (2023) TabPFN: A Transformer That Solves Small Tabular Classification Problems in a Second | 0.843 | 3 | 3 | 100% |
| 6 | Evgeniou, Theodoros and Pontil, Massimiliano and Toubia, Olivier (2007) A Convex Optimization Approach to Modeling Consumer Heterogeneity in Conjoint Estimation | 0.737 | 3 | 2 | 100% |
| 7 | Müller, Samuel and Hollmann, Noah and Arango, Sebastian Pineda and G… (2022) Transformers Can Do Bayesian Inference | 0.737 | 3 | 2 | 100% |
| 8 | Nagler, Thomas (2023) Statistical Foundations of Prior-Data Fitted Networks | 0.737 | 3 | 2 | 100% |
| 9 | Brown, Tom B and Mann, Benjamin and Ryder, Nick and Subbiah, Melanie… (2020) Language Models are Few-Shot Learners | 0.644 | 2 | 2 | 100% |
| 10 | Eremeev, Dmitry and Bazhenov, Gleb and Platonov, Oleg and Babenko, A… (2025) Turning Tabular Foundation Models into Graph Foundation Models | 0.644 | 2 | 2 | 100% |
Showing the top 10 of 60 scored citations.