This is an old revision of the document!
Introduction
Product recommendations are usually implemented by using collabrative filtering technique which recommends items by calculating similar items to the given item as well as finding similar users to whom recommendations are being made. This is very useful algorithm and is core of many big companies like amazon,netflix etc. But the major concern with collabrative filtering is that it doesn’t handle cold start problem(eg: If user has no enrollments) well so to handle this case we need content based filtering which suggests items based on user profile.So ideally a hybrid recommendation system which implements both collabrative and content based filtering systems is required.Lightfm is the library which implements the hybrid recommendation systems.
