Item Popularity Prediction in E-commerce Using Image Quality Feature Vectors

Online retail is a visual experience- Shoppers often use images as first order information to decide if an item matches their personal style. Image characteristics such as color, simplicity, scene composition, texture, style, aesthetics and overall quality play a crucial role in making a purchase decision, clicking on or liking a product listing. In this paper we use a set of image features that indicate quality to predict product listing popularity on a major e-commerce website, Etsy. We first define listing popularity through search clicks, favoriting and purchase activity. Next, we infer listing quality from the pixel-level information of listed images as quality features. We then compare our findings to text-only models for popularity prediction. Our initial results indicate that a combined image and text modeling of product listings outperforms text-only models in popularity prediction.