Fashion sales forecasting is a challenging problem: how to predict sales for items that have no sales history? In this talk we will present a high level overview on how Zalando uses machine learning models to capture both the nuances of the consumer marketplace as well as customer behaviour.
• Challenging problem: predicting sales of items without history
• Zalando Research: new ways to make use of novel data, solve hard engineering problems
• Performance in predicting millions of products
• Modelling user and fashion influencer behaviour