ClickToPrice: Incorporating Visual Features of Product Images in Price Prediction

Le résumé fourni par la source

E-commerce websites provide product images in addition to textual description and other metadata about a product to shoppers. While the use of images in online product positioning is ubiquitous, its effect on shoppers and product prices has not been studied meaningfully in literature. In this paper, we make two primary contributions: (i) we find evidence of a causal link between the features of a product's image and its price, (ii) we conduct a comprehensive evaluation of regression algorithms to predict product price using metadata and image features of the product and find that the Random Forest algorithm provides the best prediction performance on this task. We discuss implications of our findings for e-commerce portals and mobile app creators, and present directions for future research to further investigate the predictive power and statistical significance of product images on their prices.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.