Demand Estimation with Unstructured Product Data

2503.20711This paper is primarily a research paper exploring a novel method for demand estimation by incorporating unstructured data like product images and text (titles, descriptions, reviews). The authors propose using deep learning models to extract relevant features from this data and integrate them into a random coefficients logit model, allowing for the inference of consumer substitution patterns. The paper validates this approach using a choice experiment where they demonstrate its superior ability to predict second choices compared to traditional attribute-based models and a simple logit model. Furthermore, the methodology is applied to a wide range of product categories on Amazon.com, consistently showing that text and image data offer valuable information for identifying product substitutes.keepSave to notecopy_alldocsAdd noteaudio_magic_eraserAudio OverviewmapMind Map