![[2503.20711] Demand Estimation with Text and Image Data](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2503.20711] Demand Estimation with Text and Image Data
Summary
This article presents a novel demand estimation method that utilizes unstructured data from text and images to enhance substitution pattern analysis in economics.
Why It Matters
The approach addresses limitations in traditional demand estimation models by incorporating hard-to-quantify attributes, potentially leading to more accurate market predictions and better-informed business strategies. This is particularly relevant in today's data-driven economy where visual and textual information is abundant yet underutilized in economic modeling.
Key Takeaways
- Introduces a mixed logit demand model leveraging text and image data.
- Demonstrates improved counterfactual predictions over traditional models.
- Validates the method across 40 product categories, highlighting its versatility.
- Addresses challenges in estimating demand without detailed product attribute data.
- Emphasizes the importance of unstructured data in understanding consumer behavior.
Economics > General Economics arXiv:2503.20711 (econ) [Submitted on 26 Mar 2025 (v1), last revised 17 Feb 2026 (this version, v4)] Title:Demand Estimation with Text and Image Data Authors:Giovanni Compiani, Ilya Morozov, Stephan Seiler View a PDF of the paper titled Demand Estimation with Text and Image Data, by Giovanni Compiani and 2 other authors View PDF HTML (experimental) Abstract:We propose a demand estimation approach that leverages unstructured data to infer substitution patterns. Using pre-trained deep learning models, we extract embeddings from product images and textual descriptions and incorporate them into a mixed logit demand model. This approach enables demand estimation even when researchers lack data on product attributes or when consumers value hard-to-quantify attributes such as visual design. Using a choice experiment, we show this approach substantially outperforms standard attribute-based models at counterfactual predictions of second choices. We also apply it to 40 product categories offered on this http URL and consistently find that unstructured data are informative about substitution patterns. Subjects: General Economics (econ.GN); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG) Cite as: arXiv:2503.20711 [econ.GN] (or arXiv:2503.20711v4 [econ.GN] for this version) https://doi.org/10.48550/arXiv.2503.20711 Focus to learn more arXiv-issued DOI via DataCite Submission history From: Giovanni Compiani [view email] [v1] We...
