customer patterns, you can start to predict a lot about what a person may or may not do in a shop. If, say, there’s a size-10 woman wearing a gold necklace walking quickly towards the sock aisle, you can use that data to predict she wants to, well, buy socks. That could allow a retailer to automatically put targeted ads on screens aimed specifically at that person. If she looks like the type of person who wants to buy socks, they will show her adverts for socks. If it sounds familiar, it’s because the online world has been using techniques like these for years. If you search for something on Amazon, you’ll be hounded by targeted banners for similar products on other sites. Express a vague interest in canoeing and you’ll get ads for canoes wherever you go. Yet bringing these systems into the physical world isn’t a simple case of copy and paste. It turns out that people do not react to cameras in the same way as they do to browser cookies. Hoxton Analytics, a London-based team of data scientists, has developed a technology that makes use of machine learning and artificial intelligence to categorize people based on the shoes they are wearing. By analysing the style and size of people’s footwear as they walk past the sensor, the system can identify a customer’s gender with between 75 and 80% accuracy. Owen McCormack, Hoxton Analytics CEO, says that the focus of