With all the products available to a national supermarket, which ones should be stocked in your local grocery store? In this talk, I’ll show how we can recommend new products to small convenience stores, by leveraging recommendation algorithms similar to those used by Netflix. By treating product ranging as a recommendation problem, we’re able to suggest products in a customer-centric way; ensuring that local stores stock products that are suited to the specific tastes of their customers. I’ll explain how collaborative topic modelling helps us to build up a profile of each store and then make relevant recommendations about which products to stock. I’ll also discuss how retailers could use insights from these models to generate ideas for new products that haven’t yet been developed.
Supported by
