Key Takeway

Marketing models analyze structured product attributes to determine consumer preferences. Deriving such attributes from images, text and other unstructured data typically requires human judgment and can be hard to scale. Using available marketing data, machine learning techniques for disentangled representation learning can derive meaningful attributes without human intervention, often revealing consumer decision factors missed by other approaches.